JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY
11, 2017
Listen to this manuscript’s
audio summary by
JACC Editor-in-Chief
Dr. Valentin Fuster.
VOL. 69, NO.
ª 2017 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION ISSN 0735-1097/$36.00
PUBLISHED BY ELSEVIER http://dx.doi.org/10.1016/j.jacc.2016.12.035
Short-Term Effects of Tolvaptan in
Patients With Acute Heart Failure
and Volume Overload
Marvin A. Konstam, MD,a Michael Kiernan, MD, MS,a Arthur Chandler, MD,b Ravi Dhingra, MD, MPH,c
Freny Vaghaiwalla Mody, MD,d Howard Eisen, MD,e W. Herbert Haught, MD,f Lynne Wagoner, MD,g
Divya Gupta, MD,h Richard Patten, MD,i Paul Gordon, MD,j Kenneth Korr, MD,j Russell Fileccia, MD,k
Susan J. Pressler, PHD, RN,l Douglas Gregory, PHD,m Patricia Wedge, RN,m Douglas Dowling, BS,m
Matthew Romeling, MHA,m Jeremy M. Konstam, MS,m Joseph M. Massaro, PHD,n James E. Udelson, MD,a
for the SECRET of CHF Investigators, Coordinators, and Committee Members
ABSTRACT
BACKGROUND In patients with acute heart failure (AHF), dyspnea relief is the most immediate goal. Renal dysfunction,
diuretic resistance, and hyponatremia represent treatment impediments.
OBJECTIVES It was hypothesized that the addition of tolvaptan to a background diuretic improved dyspnea early in
patients selected for an enhanced vasopressin antagonism response.
METHODS In a double-blind trial, patients were randomized to tolvaptan 30 mg/day or placebo. Study entry required
hospitalization within the previous 36 h, active dyspnea, and any of the following: 1) estimated glomerular filtration
rate <60 ml/min/1.73 m2; 2) hyponatremia; or 3) diuretic resistance (urine output #125 ml/h following intravenous
furosemide $40 mg). The primary endpoint was a 7-point change in self-assessed dyspnea at 8 and 16 h, using a novel
standardized approach.
RESULTS We randomized 250 patients. There was no difference in the primary endpoint of day 1 dyspnea reduction,
despite significantly greater weight reduction with tolvaptan (
2.4  2.1 kg vs.
0.9  1.8 kg; p < 0.001). At day 3,
dyspnea reduction was greater with tolvaptan (p ¼ 0.01). There were 2 significant treatment-by-subgroup interactions:
patients without elevated jugular venous pressure and those without ascites showed directional favorability of tolvaptan
over placebo for the primary endpoint compared with patients with these findings.
CONCLUSIONS Despite rapid and persistent weight loss with tolvaptan compared with placebo, in patients with AHF
who were selected for greater potential benefit from vasopressin receptor inhibition, tolvaptan was not associated
with greater early improvement in dyspnea. Apparent subsequent differences in dyspnea warrant further exploration
of the temporal relationship between diuresis and dyspnea relief and a possible clinical role for tolvaptan.
(Randomized, Double-Blind, Placebo Controlled Study of the Short Term Clinical Effects of Tolvaptan in Patients
Hospitalized for Worsening Heart Failure With Challenging Volume Management [SECRET of CHF]; NCT01584557)
(J Am Coll Cardiol 2017;69:1409–19) © 2017 by the American College of Cardiology Foundation.
From the aCardiovascular Center, Tufts Medical Center, Boston, Massachusetts; bUniversity Cardiology Associates, Augusta,
Georgia; cCardiovascular Division, University of Wisconsin, Madison, Wisconsin; dDivision of Cardiology, VA Greater Los Angeles,
Los Angeles, California; eDivision of Cardiology, Drexel University, Philadelphia, Pennsylvania; fHeart Center Research, Hunts-
ville, Alabama; gThe Heart Institute, Mercy Hospital, Fairfield, Ohio; hDivision of Cardiology, Emory University, Atlanta, Georgia;
i
Cardiovascular Medicine, Lahey Medical Center, Burlington, Massachusetts; jCardiovascular Institute, Miriam Hospital, Provi-
dence, Rhode Island; kAdvanced Cardiovascular Specialists, Shreveport, Louisiana; lCenter for Enhancing Quality of Life in
Chronic Diseases, Indiana University School of Nursing, Indianapolis, Indiana; m
Cardiovascular Clinical Science Foundation,
Boston, Massachusetts; and the nBoston University School of Public Heath, Boston, Massachusetts. This investigation was an
investigator-initiated study funded by Otsuka Pharmaceuticals Co., Rockville, Maryland. Each of the authors and/or their orga-
nizations received support originating from Otsuka Pharmaceuticals to support this research. Dr. Gupta has received additional
research support from Otsuka. Drs. M.A. Konstam and Wagoner have received honoraria from Otsuka. Randall Starling, MD,
served as Guest Editor for this paper.
Manuscript received November 30, 2016; revised manuscript received December 23, 2016, accepted December 29, 2016.
1410 Konstam et al.
The SECRET of CHF Trial
T
ABBREVIATIONS he public health burden and costs of
AND ACRONYMS treating patients with acute heart
failure (AHF) are substantial (1). Fac-
AHF = acute heart failure
tors such as renal insufficiency and hypona-
eGFR = estimated glomerular
tremia challenge effective management and
filtration rate
independently predict morbidity and mortal-
GEE = generalized estimating
equation ity (2,3). Volume overload in such settings is
HF = heart failure
often managed by increasing loop diuretic
dosing at the expense of worsening renal
JVP = jugular venous pressure
function and hyponatremia (4). Many pa-
tients have suboptimal responses to initial diuretic
doses and have difficult-to-control volume, which
subsequently leads to needing a high diuretic dose,
an additional marker of unfavorable outcomes (5).
An alternative approach to volume homeostasis in
patients with AHF who also have renal dysfunction,
hyponatremia, or inadequate response to initial
diuretic dosing would be clinically advantageous.
Several lines of evidence suggest that the addition of
a vasopressin antagonist to a loop diuretic may be
useful in such a setting. Single-dose studies showed
favorable changes in hemodynamics (6,7). In the
EVEREST (Efficacy of Vasopressin Antagonism in
Heart Failure: Outcome Study with Tolvaptan) trial,
the addition of tolvaptan compared with placebo was
associated with reduced body weight as early as day 1,
in addition to early improvement in dyspnea and
long-term safety (8,9). In the EVEREST trial subpop-
ulation with hyponatremia at baseline, dyspnea
improvement was even more marked (10). Several
studies showed enhanced urine output and improved
dyspnea with tolvaptan in the presence of renal
dysfunction, although these studies were small,
uncontrolled, and/or open label trials (11–13).
SEE PAGE 1420
We hypothesized that addition of the vasopressin
antagonist tolvaptan in the setting of AHF accompa-
nied by hyponatremia, renal dysfunction, or inade-
quate initial diuretic response would be associated
with more rapid improvement in dyspnea versus
placebo. To test this hypothesis, we conducted the
SECRET of CHF (Study to Evaluate Challenging Re-
sponses to Therapy in Congestive Heart Failure)
multicenter, randomized, double-blind, placebo-
controlled trial. The study evaluated the short-term
clinical effects of tolvaptan in patients hospitalized
for AHF who had challenging volume management.
METHODS
Details of the design have been previously published
(14). The trial assessed the effects of tolvaptan 30 mg/day
versus placebo in subjects hospitalized for AHF who
JACC VOL. 69, NO. 11, 2017
MARCH 21, 2017:1409–19
also had a clinical feature that suggested challenging
decongestion: hyponatremia, renal dysfunction, or a
suboptimal diuretic response to initial treatment.
Eligible patients were screened and randomized
within 36 h of hospitalization to active therapy or
placebo during hospitalization. Trial assessments
were performed at 8, 16, and 24 h, and daily thereafter
while patients were hospitalized up to day 7, with an
outpatient visit at day 7 if discharged before day 7. All
patients received a telephone call at 30 days to collect
safety information and any events post-discharge. All
subjects had to continue conventional therapy, which
might have included diuretics, digoxin, angiotensin-
converting enzyme inhibitors, angiotensin II recep-
tor blockers, hydralazine, nitrates, beta-blockers, or
aldosterone inhibitors, as well as any standard medi-
cation for treating AHF.
A data and safety committee monitored data to
ensure patient safety. All sites received investiga-
tional review board approval, and all subjects pro-
vided written informed consent.
Eligible subjects were those hospitalized for AHF
who could be randomized within 36 h of initial pre-
sentation to the hospital and/or emergency depart-
ment, were age 18 years or older, had New York Heart
Association functional class III or IV symptoms on
admission, and at least a 1-month history of heart
failure (HF) treatment. Subjects were required to have
signs of extracellular volume expansion, defined as $2
of these findings: jugular venous distention, pitting
edema ($1þ), ascites, pulmonary congestion on chest
x-ray, and/or pulmonary rales. Dyspnea was required
to be present at baseline, measured by the 5-point
current dyspnea scale (moderately short of breath or
worse), within 2 h of randomization and dosing.
Within 12 h before randomization, patients were
required to have at least 1 of the following: estimated
glomerular filtration rate (eGFR) <60 ml/min/1.73 m 2;
serum sodium #134 mEq/l; or urine output #125 ml/h
during any $2-h period during the initial 8 h after
administration of intravenous furosemide of at least
40 mg (or equivalent dose of an alternative loop
diuretic). These criteria were chosen to identify pa-
tients likely to have an enhanced response to vaso-
pressin antagonism based on previous trials or
subgroup analyses (10–13). Patients with either a
reduced or preserved ejection fraction were eligible.
Patients were excluded if they had the following:
supine systolic blood pressure <90 mm Hg; serum
creatinine >3.5 mg/dl or were undergoing dialysis;
serum sodium >144 mEq/l; refractory, end-stage HF;
an acute coronary syndrome or percutaneous coro-
nary intervention within 30 days before the study;
planned revascularization procedures or other
JACC VOL. 69, NO. 11, 2017
MARCH 21, 2017:1409–19
cardiac surgery within 30 days following study
randomization; hemodynamically significant uncor-
rected primary cardiac valvular disease; hypertrophic
or amyloid cardiomyopathy or history of primary
significant liver disease; or comorbid conditions with
an expected survival of <6 months.
Randomization was conducted by a web-based
central randomization system. Subjects were ran-
domized 1:1 to tolvaptan 30 mg or placebo daily dur-
ing hospitalization, for a maximum of 7 days of
treatment. Patients concomitantly received all clini-
cally indicated therapies, including guideline-
directed medical therapies for patients with reduced
ejection fractions. Dosing of standard diuretic thera-
pies was at the discretion of the care providers based
on clinical parameters and was not protocol-directed.
TRIAL ENDPOINTS. The primary endpoint was change
in self-assessed dyspnea score, assessed at 8 and 16 h,
by a 7-point Likert scale. To minimize variability in
dyspnea measurement, we: 1) trained coordinators
using a script that encouraged symptom aggregation
over the preceding “few hours”; and 2) facilitated
patients’ recall through comparison of sequential
self-assessed 5-point absolute dyspnea scores solicited
at baseline and each treatment time point. The
importance of uniform, consistent assessment, with
scripting, was emphasized during site-study training.
Secondary endpoints, in order of clinical priority,
included: change from baseline in body weight,
measured daily while hospitalized up to 7 days;
diuretic dose collected daily while hospitalized up to
7 days; change from baseline in eGFR at discharge or
day 7, whichever came first; days alive and out of the
hospital over 30 days; change from baseline in
cognitive function at 48 h or discharge (using the
Mini-Mental Status Exam); and death or rehospitali-
zation for HF through 30 days.
STATISTICAL ANALYSIS. Demographic analyses
were performed on the intent-to-treat population,
including all randomized subjects. The safety popu-
lation included all randomized subjects who received
the study drug. The primary efficacy population
consisted of all randomized subjects who received the
study drug and had available Likert dyspnea data at
8 and/or 16 h.
The primary efficacy analysis was the comparison of
the 7-point Likert scale self-assessed dyspnea score
measured across 8 and 16 h, between the treatment and
control groups, using a repeated-measures general-
ized estimating equations (GEE) model applied to pa-
tients with available data, with a type I error (alpha) of
0.05. Each Likert category was assigned a numerical
value from 1 to 7, with 1 being markedly better and 7
Konstam et al. 1411
The SECRET of CHF Trial
being markedly worse. Each patient contributed 2 ob-
servations to the analysis; the linear model had Likert-
scale dyspnea scores as the dependent variable, and
treatment, investigational center (nominal variable),
and time (8 or 16 h) as the main effects. Presence or
absence of treatment group–center interaction was
confirmed using a repeated-measures mixed model
with the center as a random effect. The compound
symmetry within-patient correlation structure was
assumed. For sensitivity analysis in the event of
nonrandom missing data, we conducted a multiple
imputation with a conservative pattern mixture model
approach, which did not appreciably change results or
conclusions.
A supportive analysis was performed, comparing
the 2 treatment groups in terms of percent of patients
reporting moderate or marked dyspnea improvement
during the primary endpoint period, using a GEE
model with covariate adjustment for right (vs. left)
HF as the principle clinical syndrome, and an ejection
fraction #45% versus >45%. In addition, treatment
groups were compared across all time points using a
GEE model with a binomial distribution.
Secondary efficacy endpoints were ordered in
terms of clinical importance (as previously dis-
cussed), and comparisons across treatments were
performed in a sequential manner, in that comparison
of treatment and placebo groups was conducted for
each secondary endpoint, conditional on demon-
strating statistical significance for the previous
endpoint. Each analysis was a 2-tailed statistical test
with type I error set at 0.05.
The study sample size was derived based on as-
sumptions that the mean dyspnea score difference
between the treatment and placebo groups would
be $0.35 at each of the 8 and 16 h, that the within-
treatment SD of dyspnea scores was 1.0, and that
the autocorrelation between the 8- and 16-h time
points was #0.4. With these assumptions, a sample
size of 121 evaluable subjects per group would yield at
least 90% power to detect a significant difference
over post-baseline time points between tolvaptan
versus placebo using a 2-sided 0.05 level of signifi-
cance. We planned to enroll 250 subjects, which
allowed for 8 withdrawals or patients with unevalu-
able data. The assumed difference and SD in dyspnea
scores between the treatment and placebo groups was
based on dyspnea score data from the EVEREST,
ASCEND-HF (Acute Study of Clinical Effectiveness of
Nesiritide in Decompensated Heart Failure), and
REVIVE-HF (Randomized Evaluation of Intravenous
Levosimendan Efficacy) trials (8,15,16).
Statistical analysis tests were performed at a 2-sided
alpha level of 0.05, unless otherwise specified.
1412 Konstam et al. JACC VOL. 69, NO. 11, 2017

The SECRET of CHF Trial MARCH 21, 2017:1409–19

placebo. The groups were well balanced in baseline

T A B L E 1 Baseline Characteristics
characteristics (Table 1). Mean age was 69 years,
Placebo Tolvaptan 33% were women, 29% were African American, 32%
(n ¼ 128) (n ¼ 122) p Value
had an ejection fraction $45%, 83% had an eGFR
Demographics and medical history
<60 ml/min/1.73 m 2, and 19% had serum sodium
Age, yrs 67  13 70  11 0.08
Female 35 30 0.42 #134 mEq/l. Of the study participants, 12% had
African American 29 29 0.31 neither renal impairment nor hyponatremia, and
Ischemic etiology 54 49 0.37 were enrolled based on diuretic resistance alone. In
BMI, kg/m2 33.8  9.1 34.5  10.3 0.57 all, 119 tolvaptan patients and 125 placebo patients
NYHA functional class received the study drug, and all of these patients had
III 61 63 0.82
data available at 8 and/or 16 h for the primary endpoint
IV 39 37
evaluation, representing both the efficacy and safety
Diabetes 60 60 0.84
History of atrial fibrillation 37 44 0.22
populations.
Clinical and laboratory measurements DYSPNEA. By 8 h, both groups reported improvement
EF, % 33  17 35  16 0.27 in 7-point Likert dyspnea scores, with no difference
EF $45% 30 34 0.51
between groups in the dyspnea score at 8 and 16 h
Systolic blood pressure, mm Hg 122  20 122  22 0.97
(primary endpoint). In the tolvaptan and placebo
Serum Na, #134 mEq/l 21 18 0.3
groups, respectively, the Likert category numerical
eGFR, ml/min/1.73 m2 47  19 48  19 0.88
eGFR, <60 ml/min/1.73 m 2
81 85 0.48 equivalent was 2.9  1.1 and 3.0  1.1 at 8 h and
Serum creatinine, mg/dl 1.7  0.6 1.7  0.5 0.76 2.7  1.3 and 2.7  1.2 at 16 h, which represented
BUN, mg/dl 36.9  18.5 38.5  17.5 0.51 slightly better than minimal improvement for each
BNP, pg/ml* 728 (378–1,425) 577 (211–1,269) 0.115* population. There was no treatment group–center
Clinical signs interaction. Among tolvaptan and placebo patients,
Rales 59 70 0.05
respectively, either marked or moderate dyspnea
Congestion on CXR 42 49 0.27
improvement was reported in 38% and 33% at 8 h and
Elevated JVP 73 68 0.42
Ascites 19 22 0.51
45% and 43% at 16 h.
Edema 90 81 0.05 The distribution of Likert scores between groups
Treatments was compared at each assessment time point through
ICD 48 39 0.15 3 days (Central Illustration). There was a progressive
CRT 20 27 0.16 separation between groups over time that reached
ACEI or ARB 52 60 0.23
nominal statistical significance at day 3 for both nu-
Beta-blocker 92 92 0.91
merical equivalents (Table 2) and percent improved
MRA 41 37 0.54
(tolvaptan vs. placebo: 81.2% vs. 66.3%; p ¼ 0.02).
Values are mean  SD, %, or median (interquartile range). *p value calculated using the Wilcoxon rank sum test. Table 2 lists mean Likert category numerical equiva-
ACEI ¼ angiotensin-converting enzyme inhibitor; ARB ¼ angiotensin receptor blocker; BMI ¼ body mass index; lents, along with p values, for each of those same time
BNP ¼ B-type natriuretic peptide; BUN ¼ blood urea nitrogen; CRT ¼ cardiac resynchronization therapy;
CXR ¼ chest X-ray; EF ¼ ejection fraction; eGFR ¼ estimated glomerular filtration rate; ICD ¼ implantable points and out to day 7. As patients were discharged,
cardioverter-defibrillator; JVP ¼ jugular venous pressure; MRA ¼ mineralocorticoid receptor antagonist;
NYHA ¼ New York Heart Association.
the number at each time point decreased.

OTHER EFFECTS. Over the course of treatment, the

Between-group descriptive analyses used Student
t tests for continuous data, and Cochran-Mantel
Haenzel methods or chi-square tests for stratified cat-
egorical data. If departures in the assumptions of
normality and homogeneity of variance occurred, log
transformation or nonparametric methods were used
for continuous data. Treatment-by-subgroup in-
teractions were assessed for pre-specified subgroups,
including those defined by clinical signs of right versus
left HF.
RESULTS
From July 2012 through June 2016, 37 U.S. sites ran-
domized 250 patients, 122 to tolvaptan and 128 to
tolvaptan group had greater weight loss than the
placebo group (p < 0.01). At post-treatment day 1,
mean body weight change was
2.36  2.08
and
0.94  1.80 kg in the tolvaptan and placebo
groups, respectively (p < 0.001) (Central Illustration).
Over the subsequent 2 days, the 2 groups lost similar
additional amounts of weight. Over the same 3 days,
loop diuretic doses trended higher in the placebo arm
than in the tolvaptan arm (Figure 1). In terms of kid-
ney function over time (Table 3), there were no sig-
nificant differences in eGFR or serum creatinine.
There was no between-group difference in the inci-
dence of worsening renal function (defined as any
increase in serum creatinine $0.3 mg/dl) over the first
3 days of hospitalization (tolvaptan 31% vs. placebo
JACC VOL. 69, NO. 11, 2017 Konstam et al. 1413
MARCH 21, 2017:1409–19 The SECRET of CHF Trial

C ENTR AL I LL U STRA T I ON Dyspnea Responses and Body Weight Changes With Tolvaptan

A Likert dyspnea scale responses over 3 days

100% p=0.02
p=0.14
81.2%
80% 73.5%
p=0.60 66.3%
64.3%
p=0.78
60% p=0.46 52.1% 42.4%
44.9% 48.8% 37.2%
43.1% 28.6% 36.7%
37.9% 19.7%
40% 33.3% 17.0% 21.1%
10.3% 17.9%
Patients (%)

11.4%
20% 32.5% 35.7% 36.3% 38.8%
22.0% 27.6% 25.2% 28.0% 27.6% 29.6%
0%
14.2% 18.4% 16.5%
26.7% 27.6% 22.3%
31.7% 31.7% 31.4% 30.8% 8.0%
20% 10.2%
10.7%
40% 17.8% 18.7% 12.8%
31.0% 21.1%
30.9%
60%
n=123 n=116 n=123 n=118 n=123 n=117 n=112 n=113 n=98 n=85
80%
Placebo Tolvaptan Placebo Tolvaptan Placebo Tolvaptan Placebo Tolvaptan Placebo Tolvaptan
8 HOURS 16 HOURS 24 HOURS 48 HOURS Day 3
% Markedly or Moderately Better
DYSPNEA LIKERT SCALE
% Markedly better % Minimally better % Minimally worse % Markedly worse
% Moderately better % No Change % Moderately worse

B
Body Weight Change
DAY 1 DAY 2 DAY 3
0

-0.5 -0.94

-1 -1.89
-2.36 -2.41
-1.5 -3.09
-3.54
-2
kg

-2.5

-3

-3.5

-4

p<0.001 p<0.001 p=0.006

-4.5

Placebo Tolvaptan

Konstam, M.A. et al. J Am Coll Cardiol. 2017;69(11):1409–19.

Patients with acute heart failure and active dyspnea received tolvaptan or placebo to determine whether the vasopressin receptor 2 antagonist would improve dyspnea
early. There was no difference in the primary endpoint of the Likert dyspnea scale response at day 1 (A) despite a significant drop from baseline in body weight (B) at
the same time point. At day 3, dyspnea reduction was greater with the study drug, and the difference in weight reduction remained significant. In A, percentages in bold
are the percent of patients in each group with either moderately or markedly better scores at that time point; in B, values are mean  95% confidence interval; in
A and B, p ¼ between-group differences.
1414 Konstam et al. JACC VOL. 69, NO. 11, 2017

The SECRET of CHF Trial MARCH 21, 2017:1409–19

T A B L E 2 Changes in Likert Numerical Dyspnea Score*

8H 16 H 24 H Day 2 Day 3 Day 4 Day 5 Day 6 Day 7

Placebo 3.0  1.1 (123) 2.7  1.2 (123) 2.6  1.2 (123) 2.3  1.1 (112) 2.2  1.2 (98) 2.0  1.1 (81) 2.2  1.2 (56) 2.1  1.0 (49) 1.9  1.0 (41)
Tolvaptan 2.9  1.1 (116) 2.7  1.3 (118) 2.5  1.2 (117) 2.1  1.3 (113) 1.8  0.8 (85) 2.0  0.9 (64) 2.0  1.1 (55) 2.0  1.0 (44) 2.2  1.1 (36)
p Value 0.84 0.89 0.68 0.44 0.01 0.68 0.24 0.70 0.18

Values are mean  SD (n). *Based on assigned numerical value to each Likert scale category: 1 ¼ markedly better; 2 ¼ moderately better; 3 ¼ minimally better; 4 ¼ no change; 5 ¼ minimally
worse; 6 ¼ moderately worse; and 7 ¼ markedly worse.

25%; p ¼ 0.298), nor were there any between-group
difference in the change in B-type natriuretic peptide
levels from baseline to discharge or day 7 (tertiary
endpoint): median tolvaptan
88 pg/ml (IQR:
413 to
3 pg/ml) versus median placebo
201 pg/ml (IQR:
598
to
39 pg/ml; p ¼ 0.06).
For clinical outcomes during the 30-day follow-up
(Table 4), there were no differences in changes in
the Mini-Mental Status Score, days alive out of hos-
pital, HF rehospitalization, or death.
There were no significant differences between
study groups in the frequency of any adverse event.
Table 5 lists adverse events of special interest and
those that occurred in at least 5% of either popula-
tion. Compared with the placebo group, patients
randomized to tolvaptan had numerically fewer car-
diac failure and hypotension events, numerically
F I G U R E 1 Diuretic Doses
greater hypernatremia events, and a similar fre-
quency of renal events.
SUBGROUP ANALYSES. Figure 2 shows primary end-
point findings within patient subgroups. Of all of the
subgroup pairs investigated, 2 manifested significant,
qualitative treatment-by-subgroup interactions—
those defined by the presence or absence of elevated
jugular venous pressure (JVP) (p < 0.001) and of ascites
(p ¼ 0.009). In each case, patients with the finding
absent showed directional favorability of tolvaptan
over placebo for early dyspnea improvement, whereas
those with the finding present showed favorability for
placebo. In particular, patients without JVP elevation
had nominally significantly greater early dyspnea
improvement with tolvaptan. In an exploratory anal-
ysis, neither subgroup finding persisted at 3 days. At
that time point, all 4 subgroups—with and without
elevated JVP and with and without ascites—had
directionally favorable dyspnea improvement with
tolvaptan compared with placebo.

DISCUSSION
p = 0.06
250
p = 0.08
p = 0.06 In this trial, despite significant early and sustained
augmentation in weight reduction with tolvaptan, we
200 saw no between-group differences in the primary
endpoint of dyspnea improvement at 8 and 16 h
following initial tolvaptan dosing.
150
Volume overload is the principal trigger for HF
mg

hospitalization, and dyspnea, which is presumed to

be directly linked with congestion, is the most
100
181 186 frequent symptom. The goal of documenting a drug’s
175
159 154
141 138 141 beneficial effect on dyspnea has been elusive. In a
50 population selected for greater likelihood of response
to a vasopressin receptor antagonist, and using a
more standardized approach to dyspnea assessment,
0 we attempted to replicate the secondary endpoint
DAY 0 DAY 1 DAY 2 DAY 3 finding in EVEREST, of greater early dyspnea
Placebo Tolvaptan improvement with tolvaptan than with placebo, on
top of loop diuretic treatment (8). That finding was

Over the first 3 days post-treatment, with day 0 the day of randomization, loop diuretic
not replicated, although a signal of improved dyspnea
doses trended higher in patients who took placebo rather than tolvaptan. Values are appeared at day 3.
mean  95% confidence interval; p ¼ between-group difference in change from day 0. AHF trials are hampered by: 1) inhomogeneity of
patient populations and triggers for hospitalization;
JACC VOL. 69, NO. 11, 2017 Konstam et al. 1415
MARCH 21, 2017:1409–19 The SECRET of CHF Trial

T A B L E 3 Renal Function

Day 0 Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7

eGFR, ml/min/1.73 m2
Placebo 47.4  19.2 (107) 47.0  19.4 (106) 45.4  19.2 (100) 47.5  20.5 (81) 47.2  24.3 (57) 39.2  15.7 (48) 40.4  15.7 (42) 44.9  20.3 (87)
Tolvaptan 47.8  19.1 (106) 46.9  20.9 (107) 47.5  21.0 (93) 43.5  19.4 (62) 45.9  20.2 (47) 43.1  18.3 (40) 46.8  21.8 (35) 44.2  17.2 (83)
p Value* 0.399 0.842 0.348 0.389 0.477 0.160 0.660
Serum creatinine, mg/dl
Placebo 1.68  0.57 (107) 1.70  0.61 (106) 1.78  0.64 (100) 1.73  0.63 (81) 1.82  0.73 (57) 2.02  0.70 (48) 1.94  0.67 (42) 1.82  0.69 (88)
Tolvaptan 1.66  0.48 (106) 1.72  0.55 (107) 1.72  0.59 (93) 1.88  0.69 (62) 1.82  0.66 (47) 1.88  0.64 (40) 1.79  0.59 (35) 1.80  0.60 (84)
p Value* 0.095 0.903 0.207 0.69 0.458 0.299 0.793

Values are mean  SD (n). *To adjust for baseline values, p values were calculated using Student t tests of the difference in change scores from baseline between treatment groups. Day 0 ¼ day of
randomization.
Abbreviation as in Table 1.

2) difficulty affecting long-term outcomes with a
short-term intervention; 3) variability and lack of
standardization of metrics, such as those of dyspnea;
and 4) questionable direct linkage between those
endpoints and improvement in hemodynamics and
congestion (8).
We attempted to augment the probability of
observing improved dyspnea with tolvaptan by, first,
selecting patients with dyspnea within 36 h of pre-
sentation who also had either hyponatremia, diuretic
resistance, or kidney impairment. We reasoned that
hyponatremia marked higher vasopressin levels, pa-
tients resistant to loop diuretics would show greater
benefit to adjunctive tolvaptan treatment, and kidney
impairment might limit diuretic dosing, rendering
tolvaptan a more relevant adjunct. Secondly, we
standardized Likert score assessment by giving study
coordinators uniform scripts (Online Appendix) to ask
about dyspnea improvement over the preceding few
hours, as well as helping the patient focus by asking
for a current (5-point) absolute dyspnea assessment
as an aid to the (7-point) assessment of change from
baseline.
Despite these efforts, we observed incongruity
between the marked day 1 weight loss with tolvaptan
and our neutral first-day dyspnea primary endpoint
finding. In a population hospitalized for HF, Kociol
et al. (17) observed no correlation between dyspnea
relief and either weight or fluid loss. The observation
in this trial of apparently greater dyspnea relief with
tolvaptan compared with placebo on day 3 (Table 2)
was consistent with a similar finding in the TACTICS-
HF (Targeting Acute Congestion with Tolvaptan in
Congestive Heart Failure) trial (18). TACTICS-HF also
randomized patients with decompensated HF to tol-
vaptan versus placebo, and similarly reported greater
weight loss in tolvaptan-treated patients as early as
day 1 post-randomization, but they also reported no
between-group difference in early measures of
dyspnea. Thus, in both trials, greater early weight
loss (presumably reflecting volume loss) was not
associated with earlier dyspnea relief. However, by
day 3, the responder analysis in TACTICS-HF sug-
gested a favorable effect of the vasopressin receptor
antagonist, with Likert scores showing a trend in the
same direction. Thus, in 2 trials of patients with
decompensated HF—one focused on patients with a
likely enhanced response to vasopressin receptor
antagonism (SECRET of CHF) and the other with a
broader enrollment (TACTICS-HF)—tolvaptan was
associated with signs of greater early decongestion
(weight loss), with no early effect on dyspnea mea-
sures, but it did show a consistent signal of a favor-
able effect on day 3. Taken together, these consistent
results reinforced the notion of a temporal discon-
nect between early decongestion and amelioration of
the factors driving dyspnea. The data also suggested
that future trials in such patients should refocus on
the timing of dyspnea assessment to recognize this
phenomenon.
SUBGROUP FINDINGS. The assumption of direct
linkage between lung water and dyspnea may not
always be correct. Our subgroup findings, which
suggested greater early dyspnea benefit from tol-
vaptan in patients without elevated JVP or without
ascites, raised the hypothesis that mechanisms for
dyspnea differ in patients without right HF versus
patients with right HF. In the latter case, it might take
time for relief of systemic congestion to translate into
dyspnea reduction. Central venous pressure eleva-
tion drives systemic congestion and both abdominal
and peripheral fluid accumulation. A large proportion
of this fluid must be mobilized before central
plasma volume can diminish. Besides increased lung
water, dyspnea might be driven by right ventricular
distension, ventricular interdependence, tricuspid
regurgitation, and reduced cardiac output, limiting
1416 Konstam et al. JACC VOL. 69, NO. 11, 2017

The SECRET of CHF Trial MARCH 21, 2017:1409–19

symptomatic hyponatremia, including those with HF.

T A B L E 4 Clinical Outcomes
In EVEREST, in patients hospitalized with HF,
Placebo Tolvaptan p Value reduced left ventricular ejection fractions, and evi-
Mini-Mental Status Score dence of volume overload, those treated with tol-
Baseline 27  3 26  3 0.987
vaptan, compared with placebo, showed
48 h 27  3 27  3
approximately 0.9 kg greater body weight reduction
Length of stay, days 7.4  6.9 7.4  10.5 0.991
Days alive and out of hospital through day 30 21.6  7.7 22.6  7.0 0.139
after 1 day that was associated with an improvement
HF rehospitalization or death through day 30 16 (12.8) 14 (11.8) 0.740 in the percent of patients reporting moderate or
Death through day 30 6 (4.8) 6 (5.0) 0.972 marked improvement in dyspnea (8). The even
greater augmentation in weight loss and fluid
Values are mean  SD or n (%).
removal that we presently observed might be a
HF ¼ heart failure.
consequence of the more targeted population that we
studied. Nevertheless, this greater effect on fluid
volume did not translate into overall greater dyspnea
systemic oxygen delivery. Although not definitive,
improvement during the initial day.
these subgroup findings suggested the need for
Along with reduced body weight, we observed a
greater fluid removal to achieve dyspnea relief in the
relative decrease in the use of loop diuretics in pa-
presence of right HF. More effective clinical trials will
tients randomized to tolvaptan, without an excess of
require an improved understanding of these specu-
renal functional impairment or renal adverse events.
lative physiological interdependencies within various
Several studies showed adverse outcomes associated
patient subgroups.
with worsening kidney function in patients hospital-
VASOPRESSIN ANTAGONISM IN HF. Regulation of ized with HF (2,20), although this association might
vasopressin secretion is reset in severe HF, with not hold when congestion is concomitantly relieved
diminished suppression by hypo-osmolarity. The (21). Although we saw no overall dyspnea benefit
result is excess urinary water reabsorption and within the initial 24 h, the benefit we observed over
reduced serum sodium concentration and osmolarity the subsequent 2 days might signal a role for vaso-
(19). Hyponatremia is associated with a worse prog- pressin receptor antagonism as an adjunct to loop
nosis. Several studies in HF demonstrated increased diuretic treatment in AHF, with the time course of
free water excretion, improved hemodynamics, and benefit linked to the underlying physiology.
correction of hyponatremia after administration of V2
STUDY LIMITATIONS. Results might have differed if
receptor antagonists (6,7,10). In the United States,
tolvaptan is approved for treating patients with we had enrolled patients earlier than the initial 36 h
following presentation, although dyspnea was
required to be present at enrollment to minimize the
T A B L E 5 Patients With Adverse Events* possibility that major improvement had already
occurred. We might have observed even greater
Placebo Tolvaptan
(n ¼ 125) (n ¼ 119) p Value between-group differences in body weight and dys-
Cardiac failure 20 (16.0) 13 (10.9) 0.27 pnea if we had made attempts to maintain more
Hypotension 21 (16.8) 13 (10.9) 0.20 uniform diuretic doses. However, an imposed limi-
Increased 7 (5.6) 7 (5.9) 1.0 tation on diuretic dosing would not have replicated a
creatinine
Renal disorder 8 (6.4) 10 (8.4) 0.63
real-life clinical scenario. Rather, reduced diuretic
Acute renal failure 12 (9.6) 10 (8.4) 0.83 use in patients who received tolvaptan strengthened
Hypernatremia 1 (0.8) 5 (4.2) 0.11 the importance of body weight changes seen in these
Hyponatremia 4 (3.2) 3 (2.5) 1.0 patients.
Hyperkalemia 7 (5.6) 5 (4.2) 0.77
Hypokalemia 14 (11.2) 10 (8.4) 0.52 CLINICAL IMPLICATIONS. In multiple randomized,
Hypomagnesemia 7 (5.6) 4 (3.4) 0.54 controlled trials of vasopressin antagonism with
GU infection 7 (5.6) 8 (6.7) 0.79 tolvaptan, there was significant incremental weight
Constipation 8 (6.4) 4 (3.4) 0.38 loss beyond that achieved with standard-of-care
Nausea 12 (9.6) 9 (7.6) 0.65 therapy (8,13,18,22). In this trial, among patients
Headache 7 (5.6) 9 (7.6) 0.61
selected for a potentially enhanced response to such
Values are n (%). *Number of patients (%) with adverse events occurring in $5%
a therapy, the magnitude of weight loss was
of patients within either treatment group, of special interest, or with statistical substantially higher than that seen in the more un-
difference between groups.
GU ¼ genitourinary.
selected patients enrolled in the EVEREST or
TACTICS-HF trials (8,18). Overall, the data suggested
JACC VOL. 69, NO. 11, 2017 Konstam et al. 1417
MARCH 21, 2017:1409–19 The SECRET of CHF Trial

F I G U R E 2 Primary Endpoint: Subgroups

Treatment Difference (95% CI)

Subgroup Patients (n) Pint
Mean Difference in Likert Score Over 16 h

Overall 241 0.898

Female 80
0.400
Male 164
Age≤69 yrs 126
0.186
Age>69 yrs 118
Race not White 75
0.548
Race White 169
Non-ischemic Etiology 120
0.615
Ischemic Etiology 124
No Diabetes 98
0.116
Diabetes 146
No Atrial Fibrillation 145
0.183
Atrial Fibrillation 99
LVEF≤45% 183
0.578
LVEF>45% 61
SBP≤120 mm Hg 125
0.624
SBP>120 mm Hg 118
NYHA Class III 149
0.673
NYHA Class IV 94
Sodium≤134 mEq/L 41
0.594
Sodium>134 mEq/L 171
Cr≤1.57 mg/dl 106
0.693
Cr>1.57 mg/dl 106
eGFR≥60 ml/min/1.732 36
0.716
eGFR<60 ml/min/1.732 176
No rales 86
0.109
Rales 158
No congestion on x-ray 133
0.363
Congestion on x-ray 111
No elevated JVP 71
0.001
Elevated JVP 173
No ascites 193
0.009
Ascites 51
No pitting edema 34
0.698
≥1+ pitting edema 210

-1.25 -0.75 -0.25 0.25 0.75 1.25

Tolvaptan better Placebo better

*eGFR subgroups not pre-specified (cut off values reflect study inclusion criteria)

The values plotted are mean  95% confidence interval (CI) for the Likert score numerical equivalent, with 1 ¼ markedly better, ranging to
7 ¼ markedly worse. p ¼ treatment-by-subgroup interaction. BNP ¼ B-type natriuretic peptide; Cr ¼ creatinine; eGFR ¼ estimated
glomerular filtration rate; JVP ¼ jugular venous pressure; LVEF ¼ left ventricular ejection fraction; NYHA¼ New York Heart Association;
SBP ¼ systolic blood pressure.

that vasopressin antagonism is pharmacologically assessment at day 7 or at discharge in EVEREST (a

active, and results in volume and weight loss. The component of a co-primary endpoint), and early
trial efficacy results were neutral based on the pro- dyspnea relief within 24 h in both the SECRET of
spective choice of primary endpoints: patient global CHF and TACTICS-HF trials. In these trials,
1418 Konstam et al. JACC VOL. 69, NO. 11, 2017

The SECRET of CHF Trial MARCH 21, 2017:1409–19

secondary or exploratory endpoints
signals of efficacy on the dyspnea rating on day 1,
on multiple subsequent days in EVEREST, and on
day 3 in both the SECRET of CHF and TACTICS-HF
trials. Clinically, such a therapy could be consid-
ered in a patient who is challenging to diurese,
especially in the setting of hyponatremia, because
tolvaptan is available for clinical use. From a trial
perspective, the insights gained from both the
SECRET of CHF and the TACTICS-HF trials might
inform future trial design regarding the timing of
dyspnea assessments.
CONCLUSIONS
showed without right HF in clinical practice and during sub-
sequent AHF investigations.
ADDRESS FOR CORRESPONDENCE: Dr. Marvin A.
Konstam, The CardioVascular Center, Tufts Medical
Center, Box 108, 800 Washington Street, Boston,
Massachusetts 02111. E-mail: mkonstam@
tuftsmedicalcenter.org.
PERSPECTIVES
COMPETENCY IN PATIENT CARE: During the first
day of therapy, the vasopressin antagonist tolvaptan
did not improve dyspnea better than placebo in
patients with acute heart failure complicated by

Despite rapidly reduced body weight compared with hyponatremia, renal dysfunction, or diuretic resis-

placebo, in patients hospitalized with HF, dyspnea, tance, despite substantially greater weight loss.

and either hyponatremia, kidney impairment, or Tolvaptan produced greater improvement in dyspnea

diuretic resistance, tolvaptan 30 mg/day did not later in the hospital course, and patients with and

affect greater improvement in dyspnea within the without signs of right HF responded differently.

initial 24 h. Delayed dyspnea improvement, together

TRANSLATIONAL OUTLOOK: In future trials,
with absence of worsened kidney function, might
markers of decongestion may not correlate with early
signal a role for tolvaptan as an adjunct to diuretic
relief of dyspnea, and outcomes should be assessed seri-
treatment in select patients. The subgroup findings
ally over time, especially in patients with signs of right HF.
suggested the importance of additional studies for
distinguishing patients with right HF versus patients

REFERENCES

1. Heidenreich PA, Albert NM, Allen LA, et al.
Forecasting the impact of heart failure in the
United States: a policy statement from the
American Heart Association. Circ Heart Fail 2013;
6:606–19.
2. Damman K, Navis G, Voors AA, et al. Worsening
renal function and prognosis in heart failure: sys-
tematic review and meta-analysis. J Card Fail
2007;13:599–608.
3. Klein L, O’Connor CM, Leimberger JD, et al.,
OPTIME-CHF Investigators. Lower serum sodium is
associated with increased short-term mortality in
hospitalized patients with worsening heart failure:
results from the Outcomes of a Prospective Trial
of Intravenous Milrinone for Exacerbations of
Chronic Heart Failure (OPTIME-CHF) study. Circu-
lation 2005;111:2454–60.
4. Testani JM, Chen J, McCauley BD, Kimmel SE,
Shannon RP. Potential effects of aggressive
decongestion during the treatment of decom-
pensated heart failure on renal function and sur-
vival. Circulation 2010;122:265–72.
5. Hasselblad V, Gattis Stough W, Shah MR, et al.
Relation between dose of loop diuretics and out-
comes in a heart failure population: results of the
ESCAPE trial. Eur J Heart Fail 2007;9:1064–9.
6. Udelson JE, Smith WB, Hendrix GH, et al. Acute
hemodynamic effects of conivaptan, a dual V1A
and V2 vasopressin receptor antagonist, in
patients with advanced heart failure. Circulation
2001;104:2417–23.
7. Udelson JE, Orlandi C, Ouyang J, et al. Acute
hemodynamic effects of tolvaptan, a vasopressin
V2 receptor blocker, in patients with symptomatic
heart failure and systolic dysfunction: an interna-
tional, multicenter, randomized, placebo-
controlled trial. J Am Coll Cardiol 2008;52:
1540–5.
8. Gheorghiade M, Konstam MA, Burnett JC Jr.,
et al. Short-term clinical effects of tolvaptan, an
oral vasopressin antagonist, in patients hospital-
ized for heart failure: the EVEREST Clinical Status
Trials. JAMA 2007;297:1332–43.
9. Konstam MA, Gheorghiade M, Burnett JC, et al.
Effects of Oral Tolvaptan in Patients Hospitalized
for Worsening Heart Failure: The EVEREST
Outcome Trial. JAMA 2007;297:1319–31.
10. Hauptman PJ, Burnett J, Gheorghiade M,
et al., EVEREST Investigators. Clinical course of
patients with hyponatremia and decompensated
systolic heart failure and the effect of vasopressin
receptor antagonism with tolvaptan. J Card Fail
2013;19:390–7.
11. Otsuka T, Sakai Y, Ohno D, Murasawa T, Sato N,
Tsuruoka S. The effects of tolvaptan on patients
with severe chronic kidney disease complicated by
congestive heart failure. Clin Exp Nephrol 2013;17:
834–8.
12. Vaduganathan M, Gheorghiade M,
Pang PS, et al., on behalf of the EVEREST
Investigators. Efficacy of oral tolvaptan in
acute heart failure patients with hypotension
and renal impairment. J Cardiovasc Med 2012;
13:415–22.
13. Matsue Y, Suzuki M, Torii S, et al. Clinical
effectiveness of tolvaptan in patients with acute
heart failure and renal dysfunction. J Card Fail
2016;22:423–32.
14. Felker GM, Mentz RJ, Adams KF, et al. Tol-
vaptan in patients hospitalized with acute heart
failure: rationale and design of the TACTICS and
the SECRET of CHF Trials. Circ Heart Fail 2015;8:
997–1005.
15. O’Connor CM, Starling RC, Hernandez AF, et al.
Effect of nesiritide in patients with acute decom-
pensated heart failure. N Engl J Med 2011;365:
32–43.
16. Packer M, Colucci W, Fisher L, et al. Effect of
levosimendan on the short-term clinical course of
patients with acutely decompensated heart fail-
ure. J Heart Fail 2013;1:103–11.
17. Kociol RD, McNulty SE, Hernandez AF, et al.,
NHLBI Heart Failure Network Steering Committee
and Investigators. Markers of decongestion, dys-
pnea relief, and clinical outcomes among patients
hospitalized with acute heart failure. Circ Heart
Fail 2013;6:240–5.
JACC VOL. 69, NO. 11, 2017
MARCH 21, 2017:1409–19
18. Felker GM, Mentz RJ, Cole R, et al. Efficacy and
safety of tolvaptan in patients hospitalized with
acute heart failure. J Am Coll Cardiol 2016 Sep 13
[E-pub ahead of print].
19. Pruszczynski W, Vahanian A, Ardaillou R,
et al. Role of antidiuretic hormone in impaired
water excretion of patients with congestive
heart failure. J Clin Endocrinol Metab 1984;58:
599–605.
20. Blair JE, Pang PS, Schrier RW, et al.
Changes in renal function during hospitalization
and soon after discharge in patients admitted
for worsening heart failure in the placebo group
of the EVEREST trial. Eur Heart J 2011;32:
2563–72.
21. Metra M, Davison B, Bettari L, et al. Is wors-
ening renal function an ominous prognostic sign in
patients with acute heart failure? The role of
congestion and its interaction with renal function.
Circ Heart Fail 2012;5:54–62.
22. Gheorghiade M, Gattis WA, O’Connor CM, et al.,
Acute and Chronic Therapeutic Impact of a Vaso-
pressin Antagonist in Congestive Heart Failure
(ACTIV in CHF) Investigators. Effects of tolvaptan, a
Konstam et al. 1419
The SECRET of CHF Trial
vasopressin antagonist, in patients hospitalized
with worsening heart failure: a randomized
controlled trial. JAMA 2004;291:1963–71.
KEY WORDS clinical trials, diuresis,
dyspnea, jugular venous pressure,
vasopressin antagonism
A PP END IX For the Dyspnea Questionnaire
instructions and the Dyspnea Scales, please see
the online version of this article.