LOS is commonly defined as sepsis occurring 48 to 72 hours after birth, and is a serious disease with a mortality rate of 17% to 27% depending on the studied population 2 3 4 5 . Although LOS is predominantly caused by coagulase negative staphylococci (CoNS) accounting for 36-66% of cases 3 4 6 7 8 , Gram-negative rods Klebsiella pneumoniae, Escherichia coli, Serratia spp and Enterobacter cloacae are responsible for about 26-36% of cases 4 6 9 . In recent years the incidence of Gram negative infections appears to be increasing in many NICUs worldwide 10 11 12 . In an UK based microbiology laboratory survey, after exclusion of CoNS, the majority of LOS isolates were susceptible to the commonly used combinations of empiric antibiotics flucloxacillin, gentamicin, ampicillin and/or cefotaxime, however, relatively high resistance rates of Enterobacteriaceae to cephalosporins were reported 3 13 .

Meropenem is a low protein-bound (2%) broad-spectrum carbapenem with the bactericidal activity resulting from inhibition of bacterial peptidoglycan synthesis in the cell wall. It is active against Gram-positive and Gram-negative bacteria including anaerobes and extended spectrum beta-lactamase and AmpC chromosomal β-lactamase producing Enterobacteriaceae 14 . The pharmacodynamics (PD) of meropenem is determined by the fraction of time free drug exceeds MIC (f%T>MIC) with the target value of about 70% suggested for immunosuppressed adults and potentially for neonates 15 .

The PK of meropenem in neonates has been described in three studies 16 17 18 but firm dosing recommendations have not yet been made. By using Monte Carlo simulations (MCS) two studies 16 18 concluded that for most microorganisms a dose of 10 mg/kg or 20 mg/kg given twice or three times a day, depending on the gestational (GA) and postnatal age (PNA), is sufficient since >90% of treated neonates will achieve PK/PD target attainment. One possible short-coming of the simulation approach taken by Bradley et al 18 was that MIC values were randomly allocated to simulated patients, thereby assuming MICs were known a priori, and many of the allocated MIC values were likely well below the accepted breakpoints for sensitivity. When designing dosing regimens, MIC is not known in advance and it is therefore preferable to recommend doses that will cover all organisms up to the sensitivity breakpoint which is usually 2 mg/L for meropenem (http://www.eucast.org). This approach was taken by van den Anker et al. 16 and in fact both authors emphasized that higher doses and 4-hour infusion may be needed for microorganisms with increased MIC values, more specifically for Pseudomonas aeruginosa 16 18 . However, this may not be feasible due to the instability of meropenem (Summary of Product Characteristics). Whilst it is recommended that meropenem should be given within 1 hour of reconstitution, a recent study has shown that at a concentration of 4% at room temperature (≤ 25°C) the degradation will be less than 10% over 12 hours 19 .

The advantage of meropenem over standard of care might be its wider antibacterial coverage and thus the use of mono- instead of combination therapy.

Despite the fact that a number of clinical studies in neonatal sepsis have been conducted previously, several methodological issues and the changing patient population make their results out of date and not applicable to the present population affected by LOS. It is remarkable that all 13 randomised controlled trials (RCT) are relatively old, dating from 1973 to 1992 20 . With the exception of two studies, 21 22 data on EOS and LOS were combined and neonatal and paediatric patients were reported together 23 24 . The variable proportion of missing outcomes as well as the small sample sizes in the two trials which included only LOS patients makes drawing meaningful conclusions from these studies also difficult. Our proposed NeoMero1 study is therefore designed to address many of these gaps in the literature.

The study will include patients with confirmed as well as clinical sepsis (Figure 1). Confirmed sepsis will be defined as at least one positive bacterial culture (except CoNS) from a normally sterile site at baseline together with an abnormal clinical or laboratory parameter within the last 24 hours prior to randomisation (as listed below). Clinical sepsis in patients equal or older than 44 weeks of postmenstrual age is based on the criteria defined by the International Pediatric Sepsis Consensus Conference 25 . For patients below 44 weeks of postmenstrual age, the criteria of clinical sepsis were agreed at the Expert Meeting on Neonatal and Paediatric Sepsis on 8 June 2010, EMA London (http://www.ema.europa.eu/docs/en_GB/document_library/Report/2010/12/WC500100199.pdf) and consist of a combination of at least two abnormal clinical and two abnormal laboratory parameters within the 24 hours prior to randomisation. The relevant clinical criteria are (1) hyper- or hypothermia or temperature instability; (2) reduced urinary output or hypotension or mottled skin or impaired peripheral perfusion; (3) apnea or increased oxygen or increased ventilatory support requirement; (4) bradycardia spells or tachycardia or rhythm instability; (5) feeding intolerance or abdominal distension; (6) lethargy or hypotonia or irritability; (7) skin and subcutaneous lesions such as petechial rash or sclerema. The relevant laboratory criteria are: (1) white blood cells count (WBC) < 4 or > 20 × 10⁹ cells/L; (2) immature to total neutrophil (I/T) ratio > 0.2; (3) platelet count < 100 × 10⁹/L; (4) C-reactive protein (CRP) > 15 mg/L or procalcitonin ≥ 2 ng/mL; (5) glucose intolerance when receiving normal glucose amounts (8-15 g/kg/day) as expressed by blood glucose values > 180 mg/dL or hypoglycemia (< 40 mg/dL) confirmed on at least two occasions; (6) acidosis with base excess (BE) < -10 mmol/L or lactate above 2 mmol/L.

Patients who have received systemic antibiotics for more than 24 hours (except when judged as treatment failure), have organisms suspected or known to be resistant to study therapies or are not expected to survive for more than 3 months due to congenital abnormalities, will be excluded as well as patients with renal failure or known intolerance or contraindications to the study medications.

Meropenem dose for this study was set at 20 mg/kg q8h (q12h for neonates with GA < 32 weeks and PNA <2 weeks) based on the results of a previous PK study 18 . The main question to be addressed was infusion length, as giving the same dose but varying the infusion length will change f%T>MIC without affecting the area under the curve (AUC) and therefore overall exposure 16 . Using PK parameters and their uncertainty from an ongoing study in the USA [unpublished interim model kindly provided by Dr EV Capparelli], we performed simulations to investigate the effect of infusion length on f%T>MIC using the current EUCAST susceptibility cut-off for Enterobacteriaceae of 2 mg/L (http://www.eucast.org). These simulations showed that infusion times of 0.5, 1 and 2 hours gave 90% of patients with f%T>MIC of at least 44%, 48%, and 50%, respectively with a skewed distribution meaning that at least 60% in all age groups had f%T>MIC of 100% (Figure 2). Clearly increasing infusion length increased f%T>MIC, but given concerns over meropenem's stability and that at least 44% f%T>MIC was achieved for 90% of patients, a 30 minute infusion was finally recommended.

The two most commonly used antibiotic regimens for LOS (ampicillin + gentamicin or cefotaxime + gentamicin) were selected as SOC knowing that both regimens cover the vast majority of LOS causing agents and have been in use for several decades 13 . To minimise the potential bias of an open label design, each participating unit will select just one of these two regimens prior to start of the study and will then use it throughout the study. The dosages for the comparator agents were selected as recommended in the BNFC (edition of 2010-2011; http://www.bnfc.org).

The use of other systemic antibacterials will not be allowed with the exception of vancomycin and other suitable antibiotics for the treatment of infections proven or suspected to be caused by methicillin-resistant CoNS or Staphylococcus aureus. The use of topical anti-infectives, systemic antifungals, antivirals, immunglobulins and probiotics is allowed and will be recorded in the case report form.

Since the treatment of LOS is an emergency situation requiring immediate commencement of antibacterial treatment it will not always be possible to obtain the informed consent (IC) prior to the first dose of any antibiotic. Therefore the study will allow the enrolment of patients who have been given antibiotics for LOS according to the local guidelines for less than 24 hours. This avoids restricting the study to the less severe cases where treatment can be delayed until IC has been signed. For the study analysis patients will be stratified according to whether or not they received prior antibiotics.

The study aims to minimise interventions made purely for study purposes. First, all study antibiotics (including meropenem) are widely used for the treatment of LOS in the participating units despite the fact that none of them is labelled for neonates 26 . Second, few procedures will be conducted only for study purposes; PK samples, genetic samples and specific microbiologic tests will preferably be collected when routine clinical sampling is done; the collection of stool samples does not cause any extra distress to the baby/infant. Furthermore, the amount of blood collected for the PK-study (0.3 ml on a maximum of 3 times) has proven to be safe for ELBW babies 27 .

The study has been approved by the Ethics Review Committee University of Tartu (205T-18).

In addition to routine clinical monitoring, patients will be assessed for specific clinical signs and symptoms and for laboratory abnormalities at baseline, on Day 3, at the end of the antibacterial- and study treatment and in patients with a successful outcome, two days after stopping study treatments (test of cure visit - TOC). At all time points, clinical symptoms and laboratory parameters will be graded and compared with those at baseline. They will be categorised as improved, stable or worsened according to prespecified criteria. For monitoring relapses, reinfections, new infections and late safety issues a Day 28 follow up visit will be conducted for all patients with a successful outcome at TOC visit. This follow-up (FU) visit is optional and may be replaced by a telephone interview if necessary in patients with treatment failure. An auditory assessment will also be undertaken at any time up until the FU assessments. Finally, a separate study will invite patients at the corrected age of 20 to 24 months to neurodevelopmental and auditory testing if the latter was abnormal in the acute period.

Microbiological samples will be collected at baseline and on Day 3 and also whenever clinically indicated; they will be processed in local microbiology laboratories. Clinically significant isolates from blood or CSF will be stored at -80°C and sent for identification and antibacterial susceptibility to a central microbiology laboratory in order to ensure uniformity.

To compare the effect of study antibiotics on colonisation by antibiotic resistant microorganisms and fungi, stool samples or perianal swabs will be collected on admission, at the end of study therapy and at hospital discharge or the follow up visit (whichever is earliest). For a combination of techniques such as the detection of 16S ribosomal DNA 28 , organism specific PCRs and microarray techniques 29 30 , 0.5 mL of whole blood will be taken on admission and again on Day 3. Molecular techniques will be used to improve the detection of pathogens where conventional bacterial cultures are negative.

Using the previously mentioned interim PK model parameters and their uncertainty, we undertook an ED-Optimal design exercise to ascertain when to take samples given the following scenarios: a small cohort of full profile subjects with 3 post-dose samples, and the remaining subjects providing a single PK sample. Optimal sampling times were found to be 0.5, 7-8 and 12 hours post dose for those dosed 12 hourly and 0.5, 5-6 and 8 hours post-dose for those dosed 8 hourly. For the single sample subjects, the optimal time was a trough sample.

We opted for an efficacy trial. The superiority of meropenem over SOC is mainly expected from its wider coverage of responsible pathogens and from its potential use as monotherapy. We estimate that, in the control arm, 15% of patients will die before the TOC visit and among those who survive the proportion of failing subjects will be 25% 4 31 . Thus, the proportion of neonates who will die or fail therapy in the comparator arm is expected to be 36.25%. It is hypothesized that in the meropenem arm the survival will be improved to 90% and the failure rate will drop to 15% in surviving neonates. Thus, the expected proportion of neonates who will die or fail therapy should be reduced to 23.5%. Based on these assumptions, the required sample size to provide 80% power to show the superiority of meropenem over SOC, using a continuity-corrected chi-square test with a two-sided 5% alpha level, is 220 subjects per arm (NQuery software). We also anticipate that 15 to 20% of randomised neonates will ultimately have conditions not amenable to study antibiotics, although fulfilling the initial criteria of clinical sepsis, and thus decrease the apparent treatment size effect. Consequently, the sample was conservatively increased by 25%, thus the total population of the study will be 275 neonates per study arm.

The primary endpoint is a combination of clinical and microbiological criteria measured at the TOC visit and defined to a success (the patients is alive, all baseline symptoms and laboratory parameters are resolved or improved so that there is no need to continue antibacterial treatment, the baseline microorganisms are eradicated, there are no new microorganisms identified and patient has received allocated treatment for 11 ± 3 days with no modification) and a failure (all remaining cases).

Primary efficacy analysis will compare success rates in the two arms in all randomised subjects in an intent to treat approach. Subsidiary analyses of the primary endpoint by the stratification factors (SOC regimen, timing of antibiotic therapy initiation) will also be performed in the full analysis set and the confirmed sepsis subset. Various secondary analyses including comparison of survival and relapse rates or new infections by Day 28, clinical response at Day 3 and end of therapy visits, clinical and bacteriological response and duration of hospital stay will also be evaluated.