Short-course aminoglycosides as adjunctive treatment in adults with sepsis

In the Netherlands, there is large variation in the empirical antibiotic
treatment of sepsis, especially regarding the adjunctive short-term treatment
with aminoglycosides added to the widely used beta-lactam antibiotics. In a
recent multicenter observational study of684 patients with Gram-negative
bacteremia we found large heterogeneity in the use of aminoglycosides
(gentamicin or tobramycin)across seven hospitals.(Deelen et al, submitted)
Proportions of patients receiving aminoglycosides added to beta-lactam
antibiotics ranged from 11% to 49% per hospital. Over 80% of these infections
concerned community-onset infections. In a multivariable model, *hospital* was
by far the most important determinant of aminoglycoside prescription, and
explained variation five times better than sepsis severity (the second most
important determinant of aminoglycoside prescription).

The hypothesized added value of aminoglycosides results from a presumed more
rapid clearance of bacteremia, synergistic activity with beta-lactams due to
acting on two different targets, and increased probability of appropriate
empirical antibiotic therapy (defined as treatment that covers the pathogen
based on its in vitro susceptibility to antibiotics). Appropriateness of
empirical antibiotic therapy is becoming less certain due to growing antibiotic
resistance. In the Netherlands, aminoglycosides are often combined with second
generation cephalosporins (usually cefuroxime), while the usual alternative
strategy is monotherapy with a 3rd generation cephalosporins (usually
ceftriaxone). The combination of a 2nd generation cephalosporin with an
aminoglycoside provides a better coverage of S. aureus and better coverage of
ESBL+ gram-negative bacteria. Indeed, in the aforementioned study of patients
with Gram-negative bacteremia, the proportion of inappropriate empirical
treatment was 14.8% among patients not receiving an aminoglycoside and 1.9%
among those receiving both beta-lactam and aminoglycoside antibiotics. However,
this was not associated with a better outcome. In fact, the adjusted odds ratio
for day-30 mortality was 1.39 (95% CI 0.78-2.48) among patients receiving
aminoglycosides, with the point estimate pointing towards higher mortality for
patients that had received aminoglycosides.(Deelen et al) The apparent lack of
benefit is in line with results from a previous observational study in the ICU
setting in the Netherlands. Several studies on this topic have been performed
in other countries. Naturally, observational studies suffer from potential
residual confounding by indication, for which some included intermediates as
confounders in their analysis, and others used a limited set of confounders,
impeding direct causal interpretation. The few trials performed in this
population are of limited size and mostly studied prolonged use of
aminoglycosides. Yet, the results of these studies, together with the observed
large heterogeneity between hospitals in the Netherlands, justify a randomized
trial to determine the true effect of adding aminoglycosides to a beta-lactam
antibiotic on patient-relevant clinical outcomes and healthcare efficiency in
sepsis patients. The need of trials to determine the benefit of combination
therapy has also been expressed by an international consensus committee on the
research priorities in the management, epidemiology, outcome and underlying
causes of sepsis and septic shock.

The most important risks associated with aminoglycosides are nephrotoxicity
(sometimes requiring dialysis) and ototoxicity. Incidences of acute kidney
injury (AKI) in sepsis patients treated with aminoglycosides range from 14% to
46%, depending on the population studied and the definition of AKI. There are
two RCTs and six comparative observational studies assessing AKI incidence in
sepsis patients treated or not treated with aminoglycosides which yielded
heterogeneous results. E.g. a propensity score-matched cohort study of patients
with bacteremia found no association between gentamicin and nephrotoxicity with
an OR 0.90(95% CI 0.68-1.20; point estimate in favor of gentamicin). Another
study, among patients with severe sepsis or septic shock in the ICU, found an
adjusted OR for renal failure days of 1.39 (95% CI 1.00-1.94).(1) The risk of
aminoglycoside-induced nephrotoxicity might be reduced by therapeutic drug
monitoring (TDM) which is not always well described in these studies. Among the
RCTs, one was too small to be conclusive, and the other adult patients with
serious hospital-acquired infections found a small increased risk of 3% in
patients treated with aminoglycosides. The literature on ototoxicity with
short-term once-daily dosed aminoglycosides is muchless abundant. In a
meta-analysis comparing beta-lactam monotherapy to once-daily aminoglycosides
added to beta-lactams,ototoxicity outcomes were found to be not well reported
and could, therefore, not be analyzed. To the best of our knowledge,
noobservational studies have compared ototoxicity rates between regimens with
or without aminoglycosides. There is some indirectevidence to suggest that the
ototoxicity risk induced by short-course aminoglycosides is limited.
Aminoglycoside-induced nephrotoxicity is usually reversible, ototoxicity is
often permanent, and both increase patients* morbidity and health care costs,
for instance because of a temporary need of renal replacement therapy in some
patients.

The Dutch sepsis guideline does not recommend in favor or against the use of
aminoglycosides due to the lack of evidence, but instead states that *the
decision should be guided by local etiology and resistance data.* Obviously,
aminoglycosides should be withheld if there is no added value. Colloquially it
should be recommended if there is a clear positive risk-benefit balance. There
areno trials on the effectiveness of short-course aminoglycosides adjunctive to
beta-lactams in sepsis and it is not possible to answer this research question
with existing observational data, as there is strong indication bias which
cannot be fully corrected for. The heterogeneity in antibiotic treatment in
sepsis patients in the Netherlands calls for a well-designed study to determine
which strategy has the best risk-benefit balance.

The objective of the proposed study is to determine the effectiveness, safety
and cost-effectiveness of a strategy of cefuroxime combined with short course
treatment with aminoglycosides compared to a strategy of ceftriaxone
monotherapy in patients with sepsis presenting to the emergency room (ER) and
admitted to the hospital. We hypothesize that the ceftriaxone monotherapy
strategy is non-inferior for mortality. Yet, we will design the analysis such
that we can also test whether cefuroxime + aminoglycosideis superior (i.e. we
will use a two-sided alpha). For nephrotoxicity we will test the hypothesis
that a strategy of ceftriaxone monotherapy is superior to a strategy of
cefuroxime combined with short course aminoglycosides.

A secondary aim is to determine whether there is a difference in nephrotoxicity
between gentamicin and tobramycin. Here the null-hypothesis is that there is no
difference in nephrotoxicity and the alternative hypothesis is that tobramycin
is less nephrotoxic compared to gentamicin.

We will preform a cluster-randomized cross-over trial. During two consecutive
periods of 12 months, hospitals will be randomized to alternating antibiotic
policies for patients admitted with sepsis. The antibiotic policy serves as the
preferred empiric treatment of sepsis during the 12 month-period. This implies
that treating physicians may deviate from the policy if considered clinically
indicated and that patients with treatment deviation will be included in the
intention-to-treat analysis. In total 10 hospitals will participate in the
study. We will ensure that a representative number of academic and non-academic
hospitals are included to obtain a representative sepsis population.

By including 3,140 patients, the trial is large enough to statistically
demonstrate or rule out a 5% absolute reduction in mortality risk,while taking
into account possible deviations from the assigned treatment and clustering per
hospital.

In a 1/3 of the patients (active follow-up subset) we will determine the
quality of life and costs after discharge during one year. We will also
actively look at nephrotoxicity by means of 1 vena puncture 10 days after
admission. In a small group, an extra veni puncture will be necessary. This
will be in the patients with acute kidney injury without recovery 10 days after
admission or on discharge. Ototoxicity is also examined by a questionnaire.
For the PK / PD part we want to include 900 patients. These can be patients
from the active follow-up subset, but can also be participants from the main
study only. Here, residual material is used to determine the PK / PD (blood
taken in the first 48 hours of admission). The positive blood cultures will
also be saved to determine the MIC.

The planned duration of the project is 4.5 years. The proposed project is
feasible and will provide important information to improvethe health outcomes
of patients with sepsis or to avoid unnecessary health loss and costs.

-

Since both treatments are currently standard care in this patient population
(depending on which hospital you are admitted to), there is a negligible risk
in terms of treatment. Additional blood draws in the active follow-up subset
may present a risk of pain during blood collection and possible hematoma.

For the active follow-up subset, there are 3 questionnaires that must be
completed. This takes between 10-30 minutes and consists of low-impact
questions such as: how does the patient feel, has the patient been admitted or
visited a doctor and is there hearing loss.