TAILOR-AHF: randomized controlled trial investigating a tailored diuretic algorithm in acute heart failure patients

ACUTE HEART FAILURE: SCOPE OF THE PROBLEM
Hospitalizations for acutely decompensated heart failure (AHF) are not only
prevalent - 30*985 in the Netherlands in 2018 - but also put a high burden on
resource use because of their duration (mean length of stay 6.3 days in 2006 in
a US Medicare registry; 9.7 days in Maastricht UMC in 2019). The total costs
for heart failure (HF) in the Netherlands were estimated at 817 million euros
in 2017; mainly driven by hospitalizations which are responsible for almost two
thirds of the total HF costs [1]. On top of this, AHF hospitalizations impose a
high risk of short-term mortality (in-hospital 4-9%, 30-day 11-21%, 1-year
17-27%) and re-admissions (30-day 16-21%; 1-year 31-44%). The already high and
further increasing prevalence of HF (currently ~240*000 in the Netherlands;
expected increase of 72% between 2018-2040 purely based on the ageing
population) will only put more pressure on already tight resources.

AHF: NEED FOR A TAILORED DIURETIC ALGORITHM
Treatment of AHF relies mainly on diuretics to relieve congestion, by
increasing renal water and sodium excretion. Early installation of diuretics
and an adequate or *complete* decongestion are favourable factors with regard
to re-admissions and mortality. However, in clinical practice this is often not
achieved because it is very difficult to predict the individual response to
diuretic therapy and thus to choose the right dose and type of diuretics for
sufficient decongestion. One randomized trial previously investigated a high-
versus low-dose of loop diuretics and a bolus versus continuous administration
(2:2 factorial design). Both interventions did not improve patient-related nor
clinical endpoints. The trial was limited by cross-over because clinicians were
allowed to increase dosage of diuretics and indeed did so more often in the
low-dose arm. Being too aggressive may be dangerous in terms of causing
worsening renal failure. A more individualized, tailored approach taking into
account a patient*s diuretic response is warranted.

TAILORED DIURETICS: NEED FOR A RANDOMIZED TRIAL
A tailored diuretic approach using repeated measures of urinary sodium (Ur-Na)
was recently suggested by a working group paper [2] and mentioned by the
European clinical practice guidelines. However, the guideline states that *this
algorithm is entirely based on expert opinion.* Observational studies showed
that Ur-Na can predict diuretic response and speed-up decongestion. Still, no
study thus far investigated solid clinical endpoints and such an Ur-Na tailored
approach has never been compared to a control group.
The need for a decent evaluation of this algorithm is underlined by several
ongoing initiatives: i) an ongoing multicentre study investigating the
feasibility of this approach (ENACT-HF), however this study is non-randomized
and the primary endpoint is a non-clinical endpoint of natriuresis after 1 day;
ii) an ongoing single-centre randomized study in UMC Groningen (PUSH-AHF;
NCT04606927), using a more intense algorithm in a single-centre nature,
prohibiting extrapolation to non-academic centres and prohibiting evaluation of
implementation issues; iii) a randomized trial has been announced in the USA
(NCT04481919), however results may be not extrapolated due to a very different
organization of care in the Netherlands. Thus, we are still lacking an
adequately powered, clinical endpoint driven evaluation of this diuretic
approach including a cost-effectiveness analysis. This is needed to obtain a
sufficient level of evidence (B) to support implementation of this approach
because it requires more effort from medical staff. That is, treatment by the
diuretic algorithm becomes individualized and intensified, as medical staff
(nurses, doctors) are required to evaluate and adjust diuretic dosing at 6
hourly intervals, whereas usual care currently entails only 1 evaluation and
therapy adjustment each day. Moreover, evaluation of the intervention in
different types of hospitals (academic vs non-academic, small vs large) is
important to empower implementation in the entire NL. Finally, such a trial is
needed to address safety concerns, specifically with regard to worsening renal
failure.
In summary, diuretics are given on a daily basis in every Dutch hospital to our
32*500 annually admitted AHF patients. The individual response to diuretics
varies greatly yet there is no sufficient evidence to support any specific
dosing algorithm. It is hypothesized that a Ur-Na based, tailored and
intensified algorithm can help tailor diuretics in an individual way, but
sufficient evidence to support its implementation is lacking. This urgently
calls for an adequately powered trial to address this issue: TAILOR-AHF.

Primary Objective: Investigate if a tailored diuretic algorithm based on Ur-Na
has a positive effect on a composite endpoint of mortality, HF events (HFE) and
a change in quality of life (QoL) (assessed with the Kansas city cardiomyopathy
questionnaire total symptom score (KCCQ-TSS) of >=5 points) versus standard
clinical care in patients hospitalized with AHF. Secondary Objective(s): i)
Show the effect of this algorithm on the single endpoints included in the
combined primary endpoints (i.e., mortality and CV mortality, HF events, HF
rehospitalizations, and QoL following KCCQ), effectiveness of decongestion
based on a congestion score, days alive outside the hospital and change in
N-terminal Pro-Brain Natriuretic Peptide from baseline (day of admission) to
discharge (day of discharge +/- 2 days) ii) Show the safety of this protocol in
terms of worsening renal failure. iii) Care consumption: number of outpatient
and inpatient visits during 90 days follow-up iv) Provide an implementation
plan for clinical practice. v) Correlate urinary spot sodium measurement at
~2hours after loop diuretic bolus with 8 hourly urine collection and with
urinary chloride. vi) Correlate urinary potassium levels with treatment
response (interaction analysis with MRA). vii) Exploratory subgroup analyses,
being GFR>=30 versus <30 ml/min, age >=75 vs <75, gender and previous HF versus
*de novo* HF.

Study design: Single-centre, Randomized, Single-blinded, Blinded-Endpoint Trial.
Individual randomization is performed because cluster randomisation will
introduce too much bias because only a limited number of clusters can be
created on our ward.

Blinding: This design prohibits blinding of the treating physicians/nurses
since the treating physician is using the Ur-Na measures to guide and intensify
therapy. The patient will be blinded.
Endpoints however will be blinded and evaluated by an independent endpoint
committee. Also, the researchers/data-analysts performing data analysis will be
blinded. Endpoints are chosen in a way to minimize bias by non-blinding (see
endpoints).

Study arms: Arm I: Tailored, urinary-sodium based, intensified diuretic
strategy;
Arm II: Usual care

Duration: the intervention will take place only during the hospitalization,
follow-up 90 days. A study flow chart is provided in figure 1 in the study
protocol.

Setting: Cardiology ward and First Heart AID / emergency department at a large
non-academic top-clinical hospital (STZ) being Zuyderland Medical Centre (ZMC)
location Heerlen.

The proposed intervention (treatment arm I) is shown in Figure 3. It is adapted
from a recent working-group paper based on findings from several observational
studies and has been designed in consensus of an expert panel of HF
specialists. After initial assessment and diagnostics (preferable <1 hour) the
patient will be treated with a bolus of loop diuretics that can vary from a
minimal dose of 40 mg to a maximum of 250 mg furosemide depending on whether
the patient already uses diuretics or not and based on the eGFR, following the
dosage conversion table in Figure 4. Acetazolamide (500 mg intravenously) once
daily in the morning (7:00 am) is advised during the first 3 days of
hospitalization (as long as the patient is still congested) according to the
ADVOR-trial as part of routine clinical care and thus this is advised in both
study arms. Two hours after the loop diuretic bolus, Ur-Na will be evaluated.
Insertion of a urinary bladder catheter (UBC) is strongly recommended. In
patients where this doesn*t prove possible (patient refuses UBC or has a
medical contra-indication) the patient will be instructed to empty his/her
bladder (which will be verified by means of bladder-scan) before sample
collection. In case of urinary incontinence or insufficient bladder emptying, a
UBC is mandatory. If Ur-Na is below target of 100 mmol/L, the next bolus will
be doubled - up to 250 mg furosemide. Bolus administration of loop diuretics
and Ur-Na evaluation thereafter are repeated 3x a day (approx. 7 am , 3 pm and
11 pm) in the first 48 hours, and diuretics will be further increased if the
response remains sub therapeutic, including addition of other types of
diuretics (See 5.2 use of co-interventions). Once the patient has reached an
adequate diuretic response, the effective loop diuretic dose (+/- other
diuretic types) will be continued until the patient is decongested, or at
discretion of the physician. Thus, the algorithm is designed to install an
effective diuretic response as soon as possible in an individualized way. As
long as patients are still showing signs of congestion (clinical congestion
score >= 3, see section 8.1.2) a rise of serum creatinine up to 50% of baseline
is accepted and is not a reason per-se to not further increase diuretic dosage.
If creatinine increases above 50% of baseline (in combination with clinical
signs such as hypotension etc), further diagnostics are recommended for example
a (repeated) transthoracic echocardiography to confirm filling status of the
patient, a renal ultrasound, a consultation of the nephrologist, etc. This will
occur similarly as is currently done in clinical standard care. Standard care
(treatment arm II) (reflecting usual care in NL) is a loop diuretic iv (either
shots or continuous administration, to the discretion of the treating
physician) and adjustment of therapy based on other parameters such as body
weight changes or volume-balance. In both study arms, physicians are
recommended to discharge the patient only when fully decongested (clinical
congestion score <= 2 (see section 8.1.2) and NT-proBNP reduction of >=30%),
evaluated by clinical evaluation and if needed by echocardiography or
chest-X-ray. In both study arms, physicians are recommended to install optimal
HF therapy and treat underlying triggers of HF according to standard care and
guidelines. In both study arms, standard regimens on the ward are a sodium
restricted diet (3 grams/24u) and a fluid restriction of 1.8L including meals
per 24 hours.

As stated in the background section, intravenous loop-diuretics have been the
cornerstone of treatment of AHF for decades. However, there is very little
evidence about individual dosing in daily practise. Therefore it is often up to
the treating physician with which dose to start and how fast to increase this
dose. This trial will use an intensified Ur-Na based treatment algorithm which
will provide physicians with the tools to be able to reach an adequate dose
faster. Since there is no new medication involved in this trial, there are no
medication related side-effects to be expected in the intervention group that
weren*t already present during usual care. Possibly, a more aggressive approach
at increasing diuretic dose could result in more prevalent regression of renal
function due to dehydration. However, this is closely monitored by frequent
laboratory analysis (in line with usual care but also in terms of a safety
endpoint) so that diuretic dosing can be adjusted accordingly.
Furthermore, swift increase of diuretic dose will lead to a therapeutic dosage
earlier during hospitalisation. Expectations are that therefore it will lead to
earlier decongestion and shorter hospital admissions.