The CIPAZ study protocol: an open label randomised controlled trial of azithromycin versus ciprofloxacin for the treatment of children hospitalised with dysentery in Ho Chi Minh City, Vietnam [version 1; peer review: awaiting peer review]

Background: Diarrhoeal disease remains a common cause of illness and death in children <5 years of age. Faecal-oral infection by Shigella spp. causing bacillary dysentery is a leading cause of moderate-to-severe diarrhoea, particularly in low and middle-income countries. In Southeast Asia, S. sonnei predominates and infections are frequently resistant to first-line treatment with the fluoroquinolone, ciprofloxacin. While resistance to all antimicrobials is increasing, there may be randomised (1:1) to treatment with either oral ciprofloxacin (15mg/kg/twice daily for 3 days, standard-of-care) or oral azithromycin (10mg/kg/daily for 3 days). The primary endpoint will be the proportion of treatment failure (defined by clinical and microbiological parameters) by day 28 (+3 days) and will be compared between study arms by logistic regression modelling using treatment allocation as the main variable. dissemination: The study has been and the of (1341/NĐ2-CĐT). The study has also been by the (5044/QĐ-BYT).


Introduction
Diarrhoeal disease is a major childhood health issue and a leading cause of death in children <5 years of age 1 . Each year an estimated 1.73 billion episodes of diarrhoea occur worldwide 2 , and, while there have been reductions in the global burden of diarrhoea and diarrhoea-related deaths, the effect has been modest in Southeast Asia. In this region, data indicate that each child still experiences 2.4 episodes of diarrhoea every year, resulting in 427.4 million cases and 227,700 deaths annually 2 . Reasons for disparities in rates of reduction are thought to include differences in vaccination implementation, variation in breast-feeding practices and communitylevel case management practices 1 . Of note, there have been no major changes to international diarrhoea treatment algorithms over this time period 3 .
Faecal-oral infection by the Gram-negative bacilli Shigella spp. is a leading cause of moderate-to-severe diarrhoea in children, particularly in low and middle-income countries 4,5 . Shigellosis is particularly problematic in Southeast Asia where S. sonnei predominates and resistance to antimicrobials has reached a critical level [6][7][8] . Diarrhoea and dysentery (diarrhoea containing blood and/or mucus) due to infection by Shigella spp. are known to cause a signiicant burden of disease in Vietnam 6,9,10 , accounting for 49% of hospitalized cases of dysentery 10 . A key concern for the control and management of dysentery caused by Shigella spp. and other bacteria in Vietnam and elsewhere is the emergence of antimicrobial resistance, speciically against luoroquinolones [11][12][13] .
Bacterial resistance to luoroquinolones and 3 rd generation cephalosporins has increased markedly over the last decade in Vietnam, in tandem with S. sonnei becoming the predominant Shigella species 10 . Most national and international guidelines advocate the antimicrobial treatment of all cases of dysentery in children, due to the high frequency of isolating Shigella sp. as the aetiological agent and the high rate of complications and death if not treated promptly 14,15 . In addition to preventing death and morbidity, effective treatment reduces shedding and transmission 14,16 . World Health Organization recommendations are to use the luoroquinolone, ciproloxacin (CIP), as 1 st -line treatment and pivmecillinam, ceftriaxone or azithromycin, as alternative 2nd-line regimes according to local susceptibility data [15][16][17] . These recommendations were supported by a 2010 review of historical treatment failures, which estimated a 99% cure rate with these antimicrobials 18 . This high rate of success is contradicted by a previous study performed in our setting (Ho Chi Minh City, Vietnam) which suggested an 11% clinical failure rate when treating culture-conirmed Shigella dysentery cases with CIP 19 . Since these data were acquired, resistance rates have further increased over the last 10 years 20 . Currently, local resistance data suggests Shigella spp. nonsusceptibility to CIP is ~60% with ~15% of patients failing to show a clinical response to treatment and trending upwards each year 21 .

Rationale
With such high rates of resistance in Southeast Asia, there may be several direct and indirect beneits to using the 2 nd -line recommendation, azithromycin (AZI), in preference for the treatment of children with dysentery. In addition to its widespread availability and low cost, AZI has a large volume of distribution (30L/kg) and a long half-life (70hrs), which results in high tissue and intracellular concentrations at concentrations up to 100-fold those in serum 22,23 . Furthermore, incomplete absorption of AZI (low bioavailability; 17-37%) results in high concentrations of drug within colonic epithelial cells, which is likely to be particularly beneicial in colonic/diarrhoeal infection. Further theoretical indirect beneits to using AZI for the treatment of dysentery may include anti-inlammatory effects through IL-1 inhibition or CD 4 T-cell suppression 24,25 . These effects may result in a clinical treatment response even in the absence of direct in vitro microbicidal activity.
Only two previous randomised controlled trials have evaluated AZI for the treatment of dysentery. The irst, performed in Dhaka in 1997, demonstrated that AZI was effective for treating Shigella dysentery in adult men 26 . Most cases were caused by epidemic S. dysenteriae ST1 strain and rates of multi-drug resistance (MDR) and nalidixic acid resistance were high. The second study, performed in Paraguay in 2003, demonstrated AZI was more effective than ceixime for bacterial eradication in children aged 6m -5yrs, and suggested a trend for better clinical outcomes 27 . Bacterial isolates in this study were predominantly S. flexneri (~80%).
Antimicrobial resistance (AMR) is a well-established international emergency and children with diarrhoeal infection represent a signiicant proportion of the total global infectious disease burden. With the increasing rates of AMR observed in children presenting with dysentery in Vietnam and evidence of international transmission events [28][29][30] , new data supporting alternative treatment options, such as AZI, in particular for the new highly-antimicrobial resistant S. sonnei serotype, is urgently needed. While antimicrobial resistance to AZI is also now not infrequent 31 , we hypothesize that, due to the difference in pharmacokinetic/pharmacodynamic characteristics, bioavailability and theoretical indirect beneits, AZI may be superior treatment for children hospitalised with dysentery in Ho Chi Minh City compared to the current standard-of-care, CIP.

Objectives and endpoints
The primary objective of our two-group superiority study is to measure the eficacy of azithromycin (AZI) compared with ciproloxacin (CIP, standard of care) oral antimicrobials for the treatment of children hospitalised with dysentery in Ho Chi Minh City, Vietnam. The primary endpoint will be the proportion of treatment failures detected in each antimicrobial arm of the study, as deined by: • Clinical treatment failure (fever ≥38.0°C or the persistence of signs or symptoms of the infection: vomiting, abdominal pain, +/-≥ 3 loose stools +/-blood, mucus or both) after 120 hours of start of either treatment; or Secondary objectives include: 1. Quantify differences in symptom duration between treatment groups, stratiied by detection of Shigella spp. DNA in stool by PCR.
2. Assess the time to resolution of objective markers of infection and inlammation, including cessation of culture-and PCR-conirmed Shigella shedding, normalization of blood total white cell count, C-reactive protein and stool lipocalin concentrations.
3. Assess the rates of adverse events associated with exposure to the antimicrobial agents used.
4. Assess the effects of antimicrobial exposure on the host microbiome and resistome, including diversity and abundance of bacterial species in stool.

Methods and analysis
Study design This study is designed as a single centre, open label, randomized controlled trial of two therapeutic options for the antimicrobial treatment of children hospitalised with dysentery. Eligible patients will be randomly assigned to treatment with either oral ciproloxacin (15mg/kg/twice daily for 3 days, standard-of-care) or oral azithromycin (10mg/kg/daily for 3 days). Randomisation will be performed by using a set of computer-generated random numbers generated by the local CTU. Patients will be randomised 1:1 using a variable block design with blinding effected by use of identical sealed packs containing the allocated treatment generated by the central unblinded study pharmacist. Eligible participants will be assigned to the next available sequential treatment as recorded in the ward enrolment log.

Patient recruitment
Study participants will be recruited from those children aged 6 to 60 months presenting to Children's Hospital 2, Ho Chi Minh City, Vietnam, with symptoms and signs of dysentery. Children's Hospital 2 is one of three major children's hospitals in the city, it has a 1,400-bed capacity, serving the local community and acts as tertiary referral center for children with severe infectious diseases and noncommunicable diseases in southern Vietnam. After identiication, the admitting paediatric physician will approach the child's parents/guardians to discuss the study, to ask them to consider enrolment, and to explain the informed consent process. If eligible the attending study physician will obtain written informed consent.

Eligibility criteria
Criteria for study inclusion are: • Age 6 months to 60 months at time of presentation; • Have signs/symptoms of dysentery, speciically passing stools containing mucus and/or blood with/without abdominal pain, tenesmus or fever (≥37.8°C); • Be within 72 hours of the onset of signs/symptoms; • Have a parent/guardian present at admission willing to provide written informed consent.
Study exclusion criteria include: • Those known to have speciic medical/surgical conditions that may affect disease severity/presentation or response to treatment (e.g. affecting antimicrobial absorption), including: gastrointestinal abnormalities, including short bowel syndrome, chronic (inlammatory or irritable) bowel disease; inherited or acquired immune system deiciency rendering the patient immunocompromised, including chronic/ long-term steroid treatment or other immunosuppressive treatment; patients with known congenital or acquired osteoarthropathy, prolongation of the QT interval, congenital long QT syndrome; • Presentation with severe infection requiring parenteral antimicrobial treatment, including shock jaundice, extensive gastrointestinal bleeding, convulsion, drowsiness or coma, reduced movements on stimulation, tachypnoea >60 times per minute, grunting, chest retraction, poor sucking relex; • Known hypersensitivity to any of the trial drugs (ciproloxacin or azithromycin); • Coexisting infection requiring other or additional antimicrobials to be prescribed/administered.

Interventions measured
The interventions measured will be ciproloxacin 15mg/kg BW/ twice daily, with doses 12 hours apart against azithromycin 10mg/kg BW/once daily and administered by a study physician to ensure adherence.
Ciprofloxacin. CIP has a dual-ringed quinolone structure with the addition of luorine at position 6 and a piperazinyl group at position 7. The bactericidal activity of CIP, which extends to Gram negative bacteria, is a result of inhibiting both the type II topoisomerase (DNA-gyrase) and type IV topoisomerase enzymes, which are required for bacterial DNA transcription, replication, repair and recombination. Inhibition of this mechanism results in rapid bacterial cell death.
Quinolone antimicrobials are well absorbed from the gastrointestinal system; CIP has an oral bioavailability of 70%. Peak concentrations of CIP are reached 1-3 hours after dose ingestion with a half-life of 4 hours. While food does not inhibit the absorption of quinolone agents, it may result in a delay in reaching peak concentrations. Enteral feeding may reduce the absorption of quinolone antimicrobials, whether given via the oral, nasogastric or jejunal route. CIP has a high volume of distribution (V D 231L) penetrating well into most tissues and the intracellular macrophage compartment, where concentrations are 2-100x plasma. CIP is eliminated through a combination of renal and non-renal routes, including intestinal secretion, which accounts for 10-15% of drug excretion. Other routes of elimination include hepatic metabolism to less active forms, which similarly accounts for 10-20% of CIP clearance.
CIP has a broad spectrum of activity, including the Enterobacteriaceae, Gram-negative cocci, intracellular atypical pneumonia pathogens and some Gram-positive organisms including Mycobacteria and Staphylococci species. Bacterial resistance mechanisms include alterations of the target enzymes or permeability mechanisms. CIP resistance by Shigella sonnei, was irst reported in 1993 and has been recently been characterised as occurring in a single MDR lineage (III) which is now dominant globally 32 . Genome sequencing has identiied that the emergence of this globally dominant CIP-resistant strain has occurred as a single selection event likely in South East Asia, with resistance being acquired sequentially though accumulation of gyrA and parC chromosomal mutations within the DNA gyrase and topoisomerase IV genes, respectively.
Azithromycin. AZI is an azalide antimicrobial related to the macrolide erythromycin although with more favourable pharmacokinetic and side-effects proiles and a broader spectrum of activity. Structural differences with erythromycin include substitution of nitrogen for a methyl group within the lactone ring. This substitution enhances the stability of AZI in acidic conditions and thus improves the gastrointestinal absorption and bioavailability of the drug. The mechanism of action for AZI is similar to other macrolide antimicrobials; inhibition of the bacterial 50S ribosomal subunit inhibiting RNA-dependent protein synthesis.
The oral bioavailability of AZI is 36% and is further inhibited by food (50%), thus administration should be 1 hour before or 2 hours after meals. Absorption is also slowed by co-administration with magnesium or aluminium-containing antacids. AZI has a high volume of distribution (23-31L/kg) and is widely dispersed in body tissues including the lung and sputum with concentrations reaching 10-100x serum. Very high concentrations are found in macrophages and neutrophils, which is thought to be due to cellular uptake of the basic compound into acidic lysozymes by ionic trapping. AZI has a half-life of 2-4 days and antibacterial activity may persist for several days after completion of a course. Most elimination is of the nonmetabolised drug via the gallbladder or trans-intestinal route into faeces.
AZI has more activity against Gram-negative bacteria in comparison to the other macrolides, erythromycin and clarithromycin, but is less active against various Gram-positive organisms including Streptococci and Staphylococci. This enhanced activity is thought to be due to enhanced ability to penetrate the Gram-negative cell wall. Resistance to AZI is by alteration of the target site (methylation of mRNA nucleotides or mutation of ribosomal components, e.g. ermA/B/C), by macrolide modiication (by esterases or phosphotransferases e.g. ereA/B or mphA/B/D) or by eflux pump mechanisms (mefA or msrA) 33 .

Sample size calculation
In this study, children aged 6 months to 60 months, presenting with symptoms of dysentery will be randomised to receive antimicrobial treatment with either AZI or CIP. The primary endpoint will be clinical or microbiological treatment failure. To account for potential inadequacies in our assumptions and some loss to follow-up, sample size was increased by 30%. Thus, a total sample size of 364 participants, 182 in each arm, will be recruited. Parents of children meeting the study criteria will be approached daily until the participant number has been met. The ward will contain posters and lyers regarding the study and may be expanded to other sites if recruitment targets are not met.

Standard-of-care procedures
Aside from the study speciic procedures described, all patients admitted to participating wards and enrolled to the study will receive standard-of-care treatment according to national guidance 39 . Alterations to the management of each participant will be at the discretion of the treating physician; whilst in hospital, study participants may require additional procedures to be performed to optimise clinical management. In cases where the patient is failing to respond to the treatment allocated, rescue treatment with an alternative (AZI or thirdgeneration cephalosporin) will be used. If antimicrobial susceptibility data becomes available, treatment will be altered as required.

Data collection
After enrolment, data will be collected from participants during daily review, or if discharged before the inal study visit on day 28, on days 3, 7 (+3 days allowable variation), and 28 (+3 days). Data sources will include the study participant and samples collected from them, the study participant's parent(s)/guardian(s) and the clinical care records, including but not limited to hospital case notes, imaging, laboratory results and nursing/observation record charts.
Comprehensive demographic data will be collected at the baseline/day 1 visit including complete previous medical and surgical history with participant birthing, breast-feeding and immunization record, and details of current and previous medication use including antimicrobials and visits to pharmacies/medicine sellers. Data collected from physical examinations performed will include assessment of hydration and nutritional status (weight, mean upper arm circumference, skin turgor, and mucus membrane moistness) and standard physical observations (heart rate, blood pressure, respiratory rate, oxygen saturations, and oral temperature measurement). Study visits will be performed twice-daily until day 5 with only physical observation and adverse event reporting data collected during the afternoon visit (Table 1; afternoon visits not shown).
Laboratory methods. Stool specimens will be collected in sterile containers as soon as possible on the day of hospital admission after informed consent has been obtained. After delivery to the laboratory, specimens will be investigated to determine a microbiological or parasitological cause for the dysentery episode. In addition, two aliquots of stool sample will be stored at -80°C as 10% suspensions in distilled PBS for batched viral identiication and secondary analysis.
To examine stool for the presence of parasites, a fresh smear of a stool specimen will be prepared in phosphate buffered saline. 10µL of this solution will be examined at 400x magniication and examined for E. histolytica and Giardia lamblia. Further examination for Cryptosporidium cysts will be performed using Ziehl-Neelsen stain. This procedure will be performed on day of admission to the ward only.

Culture methods.
All stool specimens will be examined for the presence of blood or mucus and cultured for the presence of bacterial pathogens. Fresh stool smears of each specimen will be examined by x400 microscopy for presence of red/ white blood and pus cells, indicating an inlammatory colonic process. Stool cultures will be performed according to standard protocols; in brief, specimens will be cultured on MacConkey Agar (MC), Xylose-Lysine-Deoxycholate Agar (XLD), Selenite broth, and selective Campylobacter media. After overnight incubation at 37°C or ≥48 hours incubation at 42°C in a micro-aerophilic environment (for Campylobacter culture only), identiication of potential pathogens will be performed using standard local protocols incorporating clinical mass-spectrometry identiication approaches (Bruker MALDI-TOF Biotyper).
Shigella spp. causing dysentery will be identiied through this method; primarily as non-lactose fermenting colonies growing on MC or XLD agar. Speciation will be performed by slide agglutination using polyvalent somatic (O) and monovalent serotype-speciic antibodies (Denka Seiken, Japan), and conirmed by sequencing methods. Susceptibility testing of Shigella spp isolates will be performed using a disc diffusion method and by minimum inhibitory concentration antimicrobial gradient diffusion, where necessary, on Mueller-Hinton agar (Oxoid). Assessment of extended-spectrum B-lactamase phenotype organisms will be performed using interpretative reading of the disc diffusion results and with the double-disc diffusion synergy test 40 . Further characterisation of resistance mechanisms will be performed using established in-house genotyping methods as appropriate, as previously described 11,31 .
All isolated pathogens will be stored for future testing and the isolated organism will be recorded in the study data. Pathogen identiication and relevant susceptibility data will be reported to the clinical team as soon as it is available.

Molecular detection methods.
To enhance our ability to identify putative pathogens causing dysenteric symptoms, additional non-culture-based diagnostics will be performed using nucleic acids (DNA or RNA) extracted from stool samples collected on the day of hospital admission. Total nucleic acids will be extracted from faecal specimens (homogenised and diluted 10% in PBS) either using the QIAamp viral RNA Mini kit (QIAGEN, Hilden, Germany) or a Roche MagNA pure 96 automated nucleic acid extraction machine (Roche). After extraction, an aliquot of RNA will be converted to complementary DNA (cDNA) by reverse transcription prior to storage at -80ºC until further analysis. Bacterial (and viral) pathogen sequences will be detected in stool-extracted cDNA/ DNA by multiplex real-time PCR using established in-house protocols 41 .

Data analysis plan
Source documents are where data are irst recorded, and from which participants' CRF data are obtained. These include, but are not limited to, hospital records (from which medical history and previous and concurrent medication may be summarised into the CRF), clinical and ofice charts, laboratory and pharmacy records, and correspondence. CRF entries will be considered source data if the CRF is the site of the original recording (e.g. there is no other written or electronic record of data). All documents will be stored safely in conidential conditions. On all trial-speciic documents, other than the signed consent, the participant will be referred to by the trial participant number/code, not by name.
Analysis of the clinical trial data will be performed after dataset cleaning and not before the inal enrolled participant has attended for their Day 3 visit. The primary analysis population consists of all participants enrolled and randomised to a treatment arm. Outcomes will be analysed according to randomised arm (intention-to-treat population). In addition, the primary endpoint will be assessed using the per-protocol population, which will include all participants completing the allocated full 3-day antimicrobial course. After data cleaning the only people with access to the dataset will be the head of the CTU at OUCRU and the trial statistician.
Description of statistical methods. The primary endpoint, the proportion of treatment failure (deined by clinical and microbiological parameters) by day 28 (+3 days) after enrolment will be compared between the study arm (either CIP or AZI) based on a logistic regression model with treatment arm is the main variable. The primary analysis will not adjust for any covariates, but in a second step we will also explore the effect of the following covariates on the treatment failure (in addition to the treatment arm): duration of diarrhoea prior to enrolment, age, and pathogen (norovirus, rotavirus, Salmonella spp., Shigella spp., Campylobacter spp., unknown). The secondary endpoints will be compared between the treatment arms based on a lognormal accelerated failure time regression model for time-to-event endpoints, logistic regression for rate of relapse within 7-days (+ 3 days) endpoint.
The primary endpoint, the proportion of treatment failure (deined by clinical and microbiological parameters) by day 28 (+3 days) after enrolment will be compared between study arms (either CIP or AZI) by logistic regression modelling using treatment allocation as the main variable. The primary analysis will not adjust for any covariates, but in a second step we will also explore the effect of the following additional covariates on treatment failure: duration of diarrhoea prior to enrolment, age, and pathogen identiied (norovirus, rotavirus, Salmonella spp., Shigella spp., Campylobacter spp., unknown).
The secondary endpoints will be compared between the treatment arms based on a lognormal accelerated failure time regression model for time-to-event endpoints and logistic regression for rate of relapse within the day 7 (+3 days) endpoint. . The study has also been approved by the Vietnamese Ministry of Health (5044/QĐ-BYT). The investigators will submit and, where necessary, obtain approval from the above parties for all substantial amendments to the original approved documents. All trial potential study participant's parents/guardians will provide written informed consent in accordance with the Declaration of Helsinki 43 . SAEs and SARs will be notiied to the OUCRU CTU immediately and within no more than 24 hours of the investigator becoming aware of the event. The trial will be audited annulally by the OUCRU CTU audit team.

Ethics and dissemination.
No information concerning the study or the data will be released to any unauthorized third party without prior written approval of the sponsor. The study monitor, other authorized representatives of the sponsor, representatives of the IRB may inspect all documents and records required to be maintained by the investigator, including but not limited to, medical records (ofice, clinic, or hospital) and pharmacy records for the participants in this study. The clinical study site will permit access to such records. The study participant's contact information will be securely stored at each clinical site for internal use during the study. At the end of the study, all records will continue to be kept in a secure location for as long a period as dictated by local IRB and Institutional regulations. Study participant research data, which is for purposes of statistical analysis and scientiic reporting, will be transmitted to and stored at the Oxford University Clinical Research Unit, Vietnam (Ho Chi Minh City). This will not include the participant's contact or identifying information. Rather, individual participants and their research data will be identiied by a unique study identiication number. The study data entry and study management systems used by clinical sites and by OUCRU-Vietnam research staff will be secured and password protected. At the end of the study, all study databases will be de-identiied and archived at OUCRU, Vietnam (Ho Chi Minh City).

Study status
The trial has now received the necessary approvals and recruitment started in November 2019. We anticipate recruitment will continue for 18 months and we aim to complete data analysis by December 2021.

Dissemination
We will disseminate the results from this study to local and regional policy makers to inform updates to current treatment guidelines in order to optimise the antimicrobial treatment for Shigella dysentery in the new era of widespread antimicrobial resistance. We would also aim to present these data at national (Vietnam) and international conferences and to report the results in open access journals, in keeping with the recommendations from the Funder. All people contributing to the protocol and downstream analysis will be eligible for inclusion as an author; authorship will be determined by the Principal investigator; no professional writers will be sought. Public access to the full protocol, participant-level dataset, and statistical code can be requested from the Principal investigator.

Data availability
Underlying data No data is associated with this article.