Natriuretic Peptide Response and Outcomes in Chronic Heart Failure with Reduced Ejection Fraction

Abstract

Background:

The GUIDing Evidence Based Therapy Using Biomarker Intensified Treatment in Heart Failure (GUIDE-IT) trial demonstrated a strategy to “guide” application of guideline directed medical therapy (GDMT) by reducing NT-proBNP was not superior to GDMT alone.

Objective:

To examine prognostic meaning of NT-proBNP changes following HF therapy intensification relative to the goal NT-proBNP value of 1000 pg/mL explored in the GUIDE-IT trial.

Methods:

We examined 638 study participants alive and with available NT-proBNP results 90 days after randomization. Rates of subsequent cardiovascular (CV) death/HF hospitalization or all-cause mortality during follow-up and Kansas City Cardiomyopathy Questionnaire (KCCQ) overall scores were analyzed.

Results:

198 (31.0%) subjects had an NT-proBNP ≤1000 pg/mL at 90 days with no difference in achievement of NT-proBNP goal between biomarker-guided or usual care arms. NT-proBNP ≤1000 pg/mL by 90 days was associated with longer freedom from CV/HF hospitalization or all-cause mortality (P <0.001 for both) and lower adjusted hazard of subsequent HF hospitalization/CV death (Hazard Ratio [HR] = 0.26; 95 % confidence interval [CI] = 0.15 – 0.46; P <0.001) and all-cause mortality (HR = 0.34; 95 % CI = 0.15 – 0.77; P=0.009). Regardless of elevated baseline concentration, an NT-proBNP ≤1000 pg/mL at 90 days was associated with better outcomes and significantly better KCCQ overall scores (P =0.02).

Conclusions:

In patients with HFrEF, those with decrease in NT-proBNP ≤1000 pg/mL during GDMT had better outcomes. These findings may help to understand results of the GUIDE-IT trial ().

Clinical Trial:

GUIDE IT-HF

Condensed Abstract

We used data from the GUIDing Evidence Based Therapy Using Biomarker Intensified Treatment in Heart Failure trial to examine long-term outcomes relative to amino-terminal pro-B type natriuretic peptide (NT-proBNP) concentrations at 90 days following randomization. Those who achieved NT-proBNP ≤1000 pg/mL had lower adjusted risk for subsequent HF hospitalization/CV death (Hazard Ratio [HR] = 0.26; 95 % confidence interval [CI] = 0.15 – 0.46; P <0.001) and all-cause mortality (HR = 0.34; 95 % CI = 0.15 – 0.77; P=0.009). Low or improving NT-proBNP was also associated with better outcome. Those attaining lower NT-proBNP also had better quality of life.

Concentrations of amino-terminal pro-B type natriuretic peptide (NT-proBNP) predict adverse outcome in chronic heart failure with reduced ejection fraction (HFrEF), particularly when measured after initiation or titration of guideline directed medical therapy (GDMT) (1–3). Work to refine understanding of this relationship has suggested that those patients with persistently elevated NT-proBNP concentrations after GDMT titration have a higher long-term risk of adverse outcomes, worse quality of life, and higher propensity towards deleterious LV remodeling (4,5) compared to those with low NT-proBNP concentrations. Based on these observations, studies have been performed to define whether elevated NT-proBNP is a modifiable risk factor, by testing medical interventions aimed at suppressing NT-proBNP concentrations. The GUIDing Evidence Based Therapy Using Biomarker Intensified Treatment in Heart Failure (GUIDE-IT) trial (6,7) recently tested a strategy of GDMT for HFrEF “guided” by a goal to suppress NT-proBNP ≤1000 pg/mL versus a usual care strategy of GDMT only. The trial was stopped early due to futility, with outcomes similar in both arms; in retrospect, the NT-proBNP reduction achieved in both study arms was substantial (nearly 50% in both treatment groups reached the goal value of 1000 pg/mL or less by one year following randomization), which might be one reason for similar outcomes between arms of the trial.

The aim of this analysis was to better understand the phenomenon of NT-proBNP response to GDMT within the GUIDE-IT trial population. As a substantial percentage of GDMT titration occurred within the first 90 days of the trial, we wished to focus on change in NT-proBNP during this period relative to subsequent events. We hypothesized reduction in NT-proBNP below 1000 pg/mL (the target value sought in the study) during the first 90 days of the trial would strongly predict subsequent outcomes and would be linked to improved quality of life.

Methods

The study was approved by the institutional review board at each enrolling site and at the coordinating center and all patients provided written informed consent.

The rationale and design and primary results of the GUIDE-IT Trial () have been recently published (6,7). The study was a randomized multicenter clinical trial conducted between January 16, 2013, and September 20, 2016, at 45 clinical sites in the United States and Canada. Patients with HFrEF (EF ≤40%), NT-proBNP concentration >2000 pg/mL or BNP concentration >400 pg/mL within the prior 30 days, and a history of a prior HF event (HF hospitalization or equivalent) were randomized to either an NT-proBNP–guided strategy or usual care. Those randomized to the guided strategy had GDMT titrated to achieve guideline-directed drug targets when possible (8), but with a parallel goal of achieving a target NT-proBNP of ≤1000 pg/mL. Patients randomized to the usual care arm were managed as recommended in HF clinical practice guidelines. A substantial amount of GDMT adjustment and NT-proBNP lowering occurred during the first 90 days of the trial (7). The primary endpoint of GUIDE-IT was the composite of time-to-first HF hospitalization or cardiovascular mortality. Prespecified secondary endpoints included all-cause mortality and adverse events. As noted, the study was stopped due to futility when 894 of the planned 1100 patients had been enrolled. The median (25th-75th percentile) follow-up for the present cohort was 17 (10–24) months.

A flow diagram for the present analysis is shown in Figure 1. For the purposes of this analysis, given neutral result of the study, the two treatment strategy arms were considered as a single group. We sought to evaluate association of GDMT on NT-proBNP during the titration phase early during the study, relative to achievement of the target NT-proBNP value of 1000 pg/mL sought in the trial and subsequent outcomes. Thus, for most analyses, we examined the 638 patients who were alive and had an available NT-pro-BNP result at 90 days ± 2 weeks (henceforth referred to as “90 days”) following randomization. The time window was selected to closely align achieved NT-proBNP results to the follow-up of interest. For outcomes analyses including HF hospitalization, we excluded patients with such an event prior to 90 days. A comparison of baseline variables between those included for this analysis versus those excluded are detailed in Supplemental Table 1.

Figure 1: Study flow diagram.

Patients free of clinical events prior to 90 days following randomization were included.

NT-proBNP “response”

Given design of the GUIDE-IT HF trial, our primary goal was to examine outcomes relative to achievement of an NT-proBNP, ≤1000 pg/mL by 90 days. However, given importance of even lower values to prognosticate, we also examined categories below 1000 pg/mL, best achieved NT-proBNP by 90 days and in categories based on baseline/90 day measurement.

Statistical analysis

Baseline clinical characteristics between those study participants with NT-proBNP response by 90 days were compared using Student’s T-test or the Wilcoxon rank sum test for continuous variables and the chi-square test or Fisher’s exact test for categorical variables. Baseline characteristics are expressed as median (25^th-75^th percentile) for continuous variables and frequencies (%) for categorical variables.

To better understand a possible association between intensity of GDMT and NT-proBNP, HF therapies associated with mortality reduction in HFrEF administered at baseline and 90 days were converted to dose equivalents (9,10) and summarized with a scaled “score” of 0–5 for beta blockers and angiotensin converting enzyme inhibitors (ACEi), or 0–4 for angiotensin II receptor blockers (ARB) and mineralocorticoid receptor antagonists (MRA). The scales used for this GDMT score are detailed in Supplemental Table 2. Due to very small numbers, patients received a score of 2 for taking ivabradine, hydralazine, nitrates or sacubitril/valsartan. Mean GDMT scores between responders and non-responders were compared using Student’s T-test. Unadjusted logistic regression models were fit to test the association between medication dose change from baseline to 90 days and NT-pro-BNP ≤1000 pg/mL at 90 days.

Predictors of NT-proBNP ≤1000 pg/mL at 90 days were evaluated using multivariable logistic regression. Candidate predictors are listed in Supplemental Table 3. Multiple imputation was used to impute missing data among the candidate predictors. Forward selection, with a liberal entry criterion of 0.1, was used to identify a set of key predictors of response within each of 25 imputed datasets. During model selection, all continuous variables were modeled using natural cubic splines to account for the possibility that relationships with outcomes were not linear. The list of key predictors includes only those variables that were chosen in at least 20 of the 25 imputed datasets. Heart rate was non-linearly related to response. In the final model, heart rate was modeled using a two-part piece-wise linear spline (inflection point at 80 beats per minutes). Odds ratios (OR) with 95% confidence intervals (CI) were reported for each predictor in the final model.

Clinical Outcomes

Cox proportional hazards regression models were used to assess landmarked outcomes (11) after 90 days post randomization (HF hospitalization/cardiovascular [CV] death, all-cause mortality) as a function of NT-proBNP response at this time point. To focus on first events, patients were excluded from the analysis if they were hospitalized prior to 90 days (N=576; characteristics of these subjects at study enrollment are detailed in Supplemental Table 4). NT-proBNP response was considered continuously with a log₂ transformation and categorically.

Covariates for each outcome analysis are listed in Supplemental Table 5. Notably, all outcomes analyses were adjusted for baseline NT-proBNP concentration. Hazard ratios (HR) with 95% CI were reported.

Unadjusted Cox proportional hazards models were used to assess the association between change in NT-proBNP categories between baseline and 90 days and time to HF hospitalization/CV death and all-cause mortality. Change in response between baseline and 90 days was defined as “low-low” (≤1000 pg/mL at both baseline and 90 days), “low-high” (≤1000 pg/mL at baseline/>1000 pg/mL at 90 days), “high-low” (>1000 pg/mL at baseline/≤1000 pg/mL at 90 days) or “high-high” (>1000 pg/mL at both time points); the “high-high” group was the reference group in all models.

Kaplan-Meier cumulative event curves for HF hospitalization/CV death and all-cause mortality were generated by change groups and by 90-day response groups. Differences between groups were compared using the log-rank test.

Finally, incidence rates of subsequent outcomes were computed for groups classified by achievement of NT-proBNP≤1000 pg/mL at 90 days and GDMT score as a continuous variable. Incidence rates were calculated as the number of events per 100 patient-years of follow-up.

Patient Reported Outcomes

Generalized linear models with a normal distribution and identity link were used to examine the association between the quality of life metric, the Kansas City Cardiomyopathy Questionnaire (KCCQ), at 90 days and NT-proBNP response at this time. NT-proBNP response was again considered both continuously and dichotomously. Models were adjusted for baseline KCCQ score and other baseline variables identified through a stepwise selection process with an entry criterion of 0.1. The estimated mean changes (95% CI) for each KCCQ score were generated.

All p-values are two-sided with values<0.05 considered statistically significant. All analyses were conducted by members of the Duke Clinical Research Institute (Duke University, Durham NC) using SAS version 9.4 (SAS Institute, Inc, Cary NC).

Results

During the first 90 days, the average number of study visits was very similar between both arms of the trial; the majority (more than 80%) in both treatment groups were seen 4 times. By 90 days, 198 study participants (31.0%) had an NT-proBNP concentration ≤1000 pg/mL. Supplemental Figure 1 details distribution of baseline and 90 day NT-proBNP concentrations, along with the change in concentrations by 90 days. Table 1 details baseline characteristics of those patients with an NT-proBNP ≤1000 pg/mL at 90 days compared to those who did not achieve this target value. Characteristics of patients who had an NT-proBNP ≤1000 pg/mL at 90 days differed considerably at baseline from “non-responders” with lower baseline NT-proBNP, younger age, more female sex, and greater prevalence of new-onset HF of non-ischemic etiology. NT-proBNP responders had fewer medical co-morbidities and had a higher body-mass index and better kidney function; they also had fewer baseline signs of congestion on physical examination. 90-day responders also had higher (more favorable) overall KCCQ scores (63 vs 58; P=0.003) and longer 6-minute walk distance (336 vs 273 feet; P<0.001) at baseline. The median (25^th-75^th percentile) baseline LVEF of responders was no different than non-responders (22 [20–30] vs 25 [20–30]%; P =0.63).

Table 1:

Baseline characteristics as a function of NT-proBNP concentration ≤1000 pg/mL at 90 days. Patients achieving NT-proBNP response differed significantly from non-responders in several ways.

	NT-proBNP Concentration at 90 days
Baseline characteristic	≤1000 pg/mL (N=198)	>1000 pg/mL (N=440)	P
Age, years	57, 48–64	65, 56–74	<0.001
Male sex	121 (61.1%)	316 (71.8%)	0.007
Black race	74 (38.1%)	148 (34.5%)	0.38
Duration of heart failure (months)	1, 1–43	22, 2–72	<.001
Days from hospitalizatoin to randomization	27, 17–74	43, 20–133	0.02
Any hospitalization for heart failure	131 (66.2%)	339 (77.0%)	0.004
ICD or pacemaker	48 (24.2%)	228 (51.8%)	<0.001
History of ischemic heart disease	62 (31.3%)	249 (56.6%)	<0.001
Prior myocardial infarction	28 (14.1%)	145 (33.0%)	<0.001
Ventricular tachycardia/fibrillation	23 (11.6%)	96 (21.9%)	0.002
Atrial fibrillation at baseline	17 (8.6%)	75 (17.2%)	0.005
Peripheral arterial disease	15 (7.6%)	48 (10.9%)	0.19
Stroke	9 (4.5%)	51 (11.6%)	0.005
Hypertension	145 (73.2%)	354 (80.5%)	0.04
Diabetes mellitus	68 (34.3%)	227 (51.6%)	<0.001
COPD	24 (12.1%)	104 (23.6%)	<0.001
History of Smoking	73 (36.9%)	130 (29.5%)	0.07
Alcohol abuse	28 (14.1%)	38 (8.6%)	0.04
Depression treated with medication	31 (15.7%)	63 (14.3%)	0.66
Hyperlipidemia	87 (43.9%)	289 (65.7%)	<0.001
Sleep apnea	51 (25.8%)	92 (20.9%)	0.17
Medications at Baseline
ACE Inhibitor	136 (68.7%)	245 (55.7%)	0.002
Mineralocorticoid receptor antagonist	114 (57.6%)	208 (47.3%)	0.02
Angiotensin receptor blocker	36 (18.2%)	92 (20.9%)	0.43
Beta blocker	189 (95.5%)	410 (93.2%)	0.27
Loop diuretic	185 (93.4%)	413 (93.9%)	0.84
Standardized furosemide equivalent, mg/day	40, 40–80	40, 40–80	0.84
Statin	98 (49.5%)	280 (63.6%)	<0.001
Orthopnea			<0.001
None	81 (42.4%)	127 (29.8%)
Mild	58 (30.4%)	112 (26.3%)
Moderate	31 (16.2%)	136 (31.9%)
Severe	21 (11.0%)	51 (12.0%)
NYHA Class at Enrollment			<0.001
1	23 (11.6%)	25 (5.7%)
2	117 (59.1%)	210 (48.2%)
3	57 (28.8%)	192 (44.0%)
4	1 (0.5%)	9 (2.1%)
Body-mass index, kilograms/meter2	31, 26–37	28, 24–33	<0.001
Heart rate, beats per minute	76, 65–88	76, 68–85	0.75
Systolic blood pressure, millimeters mercury	117, 104–131	111, 100–126	0.01
Diastolic blood pressure, millimeters mercury	71, 63–80	68, 60–78	0.001
Jugular venous distention	26 (14.1%)	109 (25.9%)	<0.001
Rales	9 (4.6%)	55 (12.6%)	0.002
S3 gallop	16 (8.3%)	42 (9.8%)	0.55
Ascites	4 (2.1%)	22 (5.1%)	0.08
Peripheral edema	40 (20.4%)	130 (29.7%)	<0.001
Sodium, millimoles/liter	139, 136–141	138, 136–141	0.65
Blood urea nitrogen, milligrams/deciliter	20, 15–26	24, 16–35	<0.001
Creatinine, milligrams/deciliter	1.1, 1.0–1.3	1.3, 1.1–1.8	<0.001
KCCQ Summary Score	63, 46–80	58, 40–74	0.003
6-minute walk distance, meters	336, 251–406	273, 182–354	<0.001
Left ventricular ejection fraction, %	22, 20–30	25, 20–30	0.63
NT-proBNP, pg/mL	1313, 598–2071	3466, 1945–6452	<0.001

Continuous variables are displayed as median, 25th-75th percentile. ICD denotes: implantable cardioverter/defibrillator; COPD denotes: chronic obstructive pulmonary disease; ACE denotes: angiotensin converting enzyme; NYHA denotes: New York Heart Association; KCCQ denotes: Kansas City Cardiomyopathy Questionnaire; NT-proBNP denotes: amino-terminal pro-B type natriuretic peptide;

Baseline frequency of ACEi and MRA use was higher among those who achieved an NT-proBNP ≤1000 pg/mL at 90 days. When considered in terms of GDMT intensity, however, there was no significant difference between mean (±SD) baseline GDMT intensity score in responders and non-responders (7.0 ± 3.6 vs 6.8 ± 3.5; P =0.37). There was no difference between usual care or guided therapy arms in frequency of 90-day NT-proBNP response (p=0.78).

Baseline Predictors of 90-day NT-proBNP response

Multivariable analyses identified several independent baseline predictors of subsequent NT-proBNP reduction ≤1000 pg/mL (Supplemental Table 6): lower baseline NT-proBNP, non-ischemic etiology, younger age, higher systolic blood pressure and lower heart rate, non-Black race, and absence of chronic obstructive pulmonary disease.

Online Figure 2 demonstrates that patients achieving NT-proBNP response were generally titrated to higher doses of standard GDMT. Thus, by 90 days after enrollment, overall mean GDMT intensity scores were higher in responders compared to non-responders (8.5 ± 3.5 vs 7.7 ± 3.5; P =0.01). The results of the logistic regression models further indicate significant associations were present between higher achieved beta blocker and ACEi doses and “response” (Table 2).

Table 2:

Multivariable logistic regression analysis to assess association of guideline-directed medical therapies on concentrations of NT-proBNP at 90 days. Each analysis was a separate logistic regression, modeling the change in dose while adjusting for baseline dose. The OR expresses likelihood for achieving NT-proBNP response per scaled dose of each therapy.

Medication class	OR (95% CI)	P
Beta Blockers (per 5 mg increase in dose)	1.38 (1.10 – 1.72)	0.005
ACEi (per 5 mg increase in dose)	1.11 (1.01 – 1.21)	0.03
ARB (per 25 mg increase in dose)	1.12 (0.93 – 1.35)	0.24
MRA (per 25 mg increase in dose)	0.80 (0.58 – 1.09)	0.16

ACEi denotes: angiotensin converting enzyme inhibitor; ARB denotes: angiotensin II receptor blocker; MRA denotes: mineralocorticoid receptor antagonist.

Outcomes

Relative to achievement of the target value of ≤1000 pg/mL, those at or below target NT-proBNP concentrations by 90 days had substantially longer subsequent freedom from first HF hospitalization/CV death or all-cause mortality (Central Illustration); in the adjusted analyses, NT-proBNP response at 90 days continued to be associated with substantially lower risk for HF hospitalization/CV death (HR [95% CI] = 0.26 [0.15 – 0.46]; P <0.001) and all-cause mortality (HR [95% CI] = 0.34 [0.15 – 0.77]; P=0.009).

Central Illustration: Natriuretic Peptide Response in Heart Failure Treatment.

Cumulative event curves showing the probability of experiencing the outcomes by time in subjects with NT-proBNP ≤1000 pg/mL and those with NT-proBNP >1000 pg/mL at 90 days. A) CV death or HF hospitalization and B) all-cause mortality as a function of NT-proBNP concentrations. Those with concentrations of NT-proBNP ≤1000 pg/mL at 90 days had lower rates of subsequent events.

Given values below 1000 pg/mL are likely associated with even better outcome (and the fact many patients achieving this value were considerably lower) we then considered outcomes relative to categories of achieved NT-proBNP by 90 days (Figure 2) showing stepwise increase in rates per 100 patient years of HF hospitalization/CV death or all-cause mortality with higher NT-proBNP concentrations. Lastly, to avoid loss of discrimination due to categorizing, we also found in adjusted analyses that continuously modeled log₂-NT-proBNP achieved at 90 days was significantly predictive of HF hospitalization/CV death (HR for halving of log₂-NT-proBNP [95% CI] = 0.65 [0.57 – 0.73]; P <0.001) as well as all-cause mortality (HR for halving of log_2-NT-proBNP [95% CI] = 0.58 [0.50 – 0.68]; P <0.001).

Figure 2: Rates of CV death or HF hospitalization and all-cause mortality as a function of NT-proBNP categories at 90 days.

Higher concentrations of NT-proBNP by 90 days after randomization were associated with worse outcomes.

We then considered change in NT-proBNP from baseline to 90 days; in doing so, we found change in NT-proBNP also significantly associated with HF hospitalization/CV death (HR for 50% reduction in log₂-NT-proBNP [95% CI] = 0.55 [0.47–0.64]; P <0.001) and all-cause mortality (HR for 50% reduction in log₂-NT-proBNP [95% CI] = 0.54 [0.44–0.66]; P <0.001). Relative to change from baseline in a categorical fashion, Figure 3 details Cox proportional hazards analyses and cumulative event curves, which show those study participants with a low NT-proBNP at baseline or at 90 days had a lower risk for events, compared to those with rising NT-proBNP or those elevated at both time points; notably, rising NT-proBNP during the first 90 days was associated with high event rates despite a low concentration at baseline.

Figure 3: Outcomes in GUIDE-IT as a function of change in NT-proBNP response categories from baseline to 90 days.

Patients were considered as “low-low” if ≤1000 pg/mL at both time points, “low-high” if ≤1000 pg/mL at baseline/>1000 pg/mL at 90 days, “high-low” if >1000 pg/mL at baseline/≤1000 pg/mL at 90 days, and “high-high” if >1000 pg/mL at both time points. A) Cox proportional hazards analyses and cumulative event curves for B) HF hospitalization/CV death or C) all-cause mortality show superior subsequent outcomes among those with NT-proBNP ≤1000 pg/mL at baseline and/or 90 days. The hazard ratio is adjusted for baseline NT-proBNP.

Considered as a continuous variable, higher GDMT score showed no clear association with HF hospitalization (HR=1.0; P = 0.88), but higher scores were significantly associated with lower rates of all-cause death (HR =0.91; P =0.003). No interaction between NT-proBNP response and achieved GDMT relative to outcomes was detected. However, calculated incidence rates were dramatically lower among responders regardless of GDMT score (Online Figure 3); for example those with an NT-proBNP ≤1000 pg/mL by 90 days had low rates of subsequent HF hospitalization/CV death (5% below the median GDMT score of 8 and 9% with GDMT≥8) or all-cause mortality (both <5%) regardless of achieved GDMT. In contrast, among those receiving greater than median GDMT intensity, an NT-proBNP >1000 pg/mL at 90 days was associated with a 42% rate of subsequent HF hospitalization/CV death, and a 13% subsequent rate of all-cause mortality.

Patient reported outcomes

Table 3 details 90-day cross-sectional association between lower log₂-NT-proBNP concentrations and improved quality of life scores. Across each KCCQ domain, lower NT-proBNP concentrations were independently associated with improved quality of life. Figure 4 details achieved NT-proBNP concentration categories by 90 days relative to KCCQ scores at the same time point. NT-proBNP ≤ 1000 pg/mL at 90 days was associated with favorable change in all domains (average increase ~5 points), but none of these increases were statistically significant with exception of improvement in Self Efficacy (P=0.04).

Table 3:

Associations between achieved log₂-NT-proBNP by 90 days and change in quality of life scores; these models used 90-day KCCQ score as the outcome and 90-day log₂-NT-proBNP and baseline KCCQ as predictors. The estimates presented correspond to a 1 unit decrease in log₂-NT-proBNP. Lower NT-proBNP was independently associated with better quality of life. KCCQ denotes: Kansas City Cardiomyopathy Questionnaire.

	Unadjusted		Adjusted
Outcome	KCCQ difference per 50% lower concentration of log2-NT-proBNP (95% CI)	P	KCCQ difference per 50% lower concentration of log2-NT-proBNP (95% CI)	P
KCCQ OverallA	1.59 (0.63–2.55)	0.001	1.21 (0.22–2.19)	0.02
Physical LimitationsB	2.25 (1.11–3.40)	<0.001	1.50 (0.34–2.67)	0.01
Social LimitationsC	2.50 (1.12–3.87)	<0.001	1.59 (0.17–3.01)	0.03
Symptom BurdenD	1.67 (0.53–2.82)	0.004	1.12 (−0.04–2.28)	0.06
Self-EfficacyE	0.87 (0.17–1.57)	0.01	0.88 (0.19–1.57)	0.01

Unadjusted analyses included log₂-NT-proBNP, as well as baseline KCCQ score. The model estimate represents the difference in means with respect to a halving of NT-pro-BNP.

^AAdjusted for ICD/pacemaker, BUN, heart rate, SpO2, walk distance and NYHA class.

^BAdjusted for ethnicity, prior HF hospitalization, high BP, ICD/pacemaker, black race, NYHA class, sex, sodium and walk distance.

^CAdjusted for ICD/pacemaker, NYHA class, BUN, SpO2, walk distance.

^DAdjusted for ICD/pacemaker, NYHA class, qualifying event, sex, smoking history, and heart rate.

^EAdjusted for depression treated with medication, ethnicity, high BP, third heart sounds, BUN, and potassium.

Figure 4: Box-and-whisker plot of the Kansas City Cardiomyopathy Questionnaire (KCCQ) at 90 days as a function of NT-proBNP at this same time point.

Lower NT-proBNP was associated with higher KCCQ scores, consistent with superior quality of life. Median values are the black line, boxes represent the 25^th and 75^th percentiles, and whiskers detail the 5^th and 95^th percentiles.

Discussion

Among high risk patients with HFrEF managed with a goal to achieve optimal GDMT with or without a goal reduction in NT-proBNP, we have made several important observations (Figure 2; Figure 5). By 90 days following randomization, reduction in NT-proBNP ≤1000 pg/mL (the target value sought in the trial) occurred in nearly a third of study participants during a period of intensive GDMT titration. Notably, achieved NT-proBNP concentrations by 90 days were monotonically associated with risk for HF hospitalization/CV death as well as all-cause mortality; accordingly, patients achieving NT-proBNP ≤1000 pg/mL by 90 days following randomization had significantly lower risk for these outcomes compared to those who did not, strikingly regardless of achieved GDMT. Lower achieved concentrations were associated with better outcome, regardless of the percent change or baseline NT-proBNP concentration. Additionally, lower absolute NT-proBNP concentrations by 90 days were also associated with improved QOL at that time, which resonates with prior findings (12,13) and presages improved subsequent outcome. Together with recent findings from the GUIDE-IT study suggesting that lower achieved NT-proBNP concentrations were also associated with greater degrees of favorable reverse ventricular remodeling (14), our results affirm importance of lower concentrations of this biomarker during HF therapy, and extend understanding of NT-proBNP interpretation to a contemporary cohort of patients with HFrEF.

Figure 5:

No caption

Unlike other chronic cardiovascular conditions, HFrEF lacks an accepted physiologic target for therapy. Accordingly, HF therapies are titrated per clinical practice guidelines to maximally tolerated doses (8). This guideline-recommended approach to care is expected to have more favorable outcomes compared to less aggressive GDMT application, yet it leaves little room for individualization of care, and offers no guidance for assessing risk among those achieving optimal GDMT targets. Though the prognostic value of NT-proBNP changes to guide therapy in the trial were questioned as a possible explanation for the study results (15), no data regarding prognostic meaning of the biomarker in the trial were available at the time when primary results of the study were published.

Our results help, in part, to understand the outcome of the GUIDE-IT trial. We previously suggested lower NT-proBNP targets and achievement of this goal value to be among the variables predictive of successful “guided” therapy (16); to the extent there was no difference in the achievement of NT-proBNP response between the guided therapy and usual care arms, together with the dominant effect such NT-proBNP reduction would have on outcomes, it would be hard to expect a difference between study arms. Indeed, the NT-proBNP reduction achieved in the usual care arm of GUIDE-IT was among the most substantial of any of the prior guided therapy trials performed to date. However, the prognostic meaning of NT-proBNP in GUIDE-IT had not been explored. In this analysis, we further characterize the biomarker responses in the trial, demonstrating the prognostic meaning of NT-proBNP reduction; the results emphasize the predictive value of NT-proBNP changes in the course of HF therapy and extend understanding of how GUIDE-IT may be interpreted.

Our data suggest that when patients treated for chronic HFrEF have a robust reduction in NT-proBNP during their treatment, it is expected to be accompanied by lower clinical event rates, better QOL and more substantial reverse LV remodeling (14) compared to those with higher (or rising) NT-proBNP. These results are consistent with several studies before ours (1,2,17). Reduction in NT-proBNP, or “response”, is not entirely determined by GDMT, as our data showed numerous non-therapeutic predictors of lower NT-proBNP concentrations 90 days after study entry. Such variables also potently predict inability to titrate GDMT (e.g. ischemic HF, advanced age, lower blood pressure). Thus, non-responders often have more advanced HF characteristics, identifying a patient at a more advanced stage of their disease journey. Such patients without NT-proBNP response despite GDMT had a clearly higher risk, one that may not be rescued by more intensive medical therapy; such patients might therefore be considered for more prompt progression to advanced HF therapy, if appropriate.

Several caveats should be considered in the interpretation of our findings. To align with the design of the GUIDE-IT study and to be clinically applicable, we used a dichotomous cut-off of 1000 pg/mL as the definition of “response”. It is noteworthy that analyses have recently found reverse ventricular remodeling is more obvious when NT-proBNP is below 1000 pg/mL (14). Nonetheless, dichotomization does not generally improve performance of continuous prognostic variables, so we also examined prognostic performance of NT-proBNP continuously as well as affirming higher event rates at more elevated results for the biomarker and conversely, lower event rates with greater NT-proBNP reduction. The GUIDE-IT study aimed for an NT-proBNP below 1000 pg/mL, which was the central focus of this analysis and we show lower concentrations than this cut-off are associated with even better outcome. Though aiming for the “lowest possible” NT-proBNP value (even far below 1000 pg/mL) is likely to result in better outcome, this is a strategy not explored by this trial; future efforts to explore this question are worthwhile. The therapy score we developed attempts a “holistic” view of GDMT, and does not consider individual classes of drugs, which might result in similar scores between patients despite different doses of the various GDMT administered. When analyzed as a continuous variable, the GDMT score was associated with mortality (with lower event rates seen in those receiving higher doses of GDMT); given lack of accepted cut-offs for low, intermediate, and higher GDMT scores, for display purposes, we used the median of the GDMT score in Online Figure 2. The central message remains, which is more intensive therapy was associated with lower mortality rates. Lastly, very few subjects in the trial were treated with sacubitril/valsartan, a therapy that more often leads to substantial NT-proBNP reduction (1); ongoing trials will examine implication of NT-proBNP reduction from this relatively newer therapy for HFrEF (18).

Our results emphasize the importance of a low post-treatment NT-proBNP during HFrEF therapy. Current clinical practice guidelines and consensus statements for management of HF articulate support for NT-proBNP measurement as a tool for prognostication (8,19,20) but do not specifically recommend lower concentrations of this biomarker as a measure of clinical stability. Part of the concern regarding this stance is a reasonable worry regarding accepting under-treatment of patients with lower NT-proBNP concentrations; it is accepted that higher doses of GDMT are associated with more favorable outcomes and as such, therapies should always be maximized, regardless of NT-proBNP values. Indeed, we show that higher intensity of GDMT in GUIDE-IT was associated with lower rates of mortality; this supports current clinical practice guideline recommendations for titration to highest possible doses of these therapies. However, given the substantially lower event rates and superior quality of life among those with lower NT-proBNP concentrations during course of GDMT, it is reasonable to hypothesize “optimal therapy” for HFrEF might be defined as target doses of GDMT together with robust NT-proBNP reduction. Further research and discussion around this concept are needed.

Clinical Perspectives.

Competency in Medical Knowledge:

Clinical practice guidelines and consensus statements recommend measuring NT-proBNP to assess prognosis in patients with chronic HFrEF. Lower NT-proBNP concentrations are associated with a more favorable prognosis and better quality of life.

Translational Outlook:

Further research is needed to expose the mechanisms linking trends in NT-proBNP with the biology of HF and clinical outcomes.

Abbreviations

NT-proBNP

Amino-terminal pro-B type natriuretic peptide

HFrEF

Heart failure with reduced ejection fraction

GDMT

Guideline directed medical therapy

ACEi

Angiotensin converting enzyme inhibitor

ARB

Angiotensin II receptor blocker

MRA

Mineralocorticoid receptor antagonist

CV

Cardiovascular

OR

Odds ratio

CI

Confidence interval

KCCQ

Kansas City Cardiomyopathy Questionnaire

HR

Hazard ratio