The primary objective is to evaluate the decrease in the number of patients hospitalised after at least 48 hours of experimental treatment or requiring hospitalisation for COVID-19. The 48-hour delay allows a more objective attribution of the clinical course to advanced treatment. The embedded second primary objective is to evaluate the decrease in the number of patients requiring mechanical ventilator support.
The secondary objectives are to evaluate:
- The decrease in the number of patients with a SARS-CoV-2-positive PCR test 7 days after the start of the experimental treatment.
- The decrease in the total length of hospital stay.
- The decrease in the length of hospitalisation in the intensive care unit.
- The decrease in the duration of mechanical ventilator support.
- The decrease in the rate of deaths related to SARS-CoV-2 infection.
- An assessment of the safety of doxycycline.
The main outcome is an efficacy outcome corresponding to the percentage of patients hospitalised after at least 48 hours of experimental treatment. The outcome corresponding to the embedded second primary objective corresponds to the percentage of patients requiring mechanical ventilator support. These three outcomes are monitored in the two arms from 48 hours of treatment to 28 days corresponding to the end-of-study visit.
For the secondary outcomes:
Efficacy: These outcomes correspond to the report of the:
- Number of positive SARS-CoV-2 PCR tests at the inclusion visit and on D7.
- Total length of hospital stay.
- Total length of hospital stay in the intensive care unit.
- Duration of mechanical ventilator support.
- Number of deaths related to SARS-CoV-2 infection.
Safety: Report of the number of adverse events (AE) and serious adverse events (SAE) over 28 days.
Apart from the first and the last secondary outcomes, the other outcomes will be recorded at 28 days. If the patient is hospitalised during the study period, this data will be recorded three months after the beginning of the treatment. For all of these secondary outcomes, percentages will be calculated.
For all of the outcomes, the data will be reported as percentages with confidence intervals and a p value. Furthermore, the treatment effects, reported in the results and which meets all the outcomes except that of safety, will be estimated by risk ratio (plus confidence intervals and p value). We will focus on the reduction of hospitalisations (main objective), mortality and the duration of respiratory care. Moreover, we will look at the correlation between the number of risk factors and these outcomes in the results.
The DYNAMIC study is a superiority study of the efficacy of doxycycline. It is a prospective, phase III, randomised, stratified, double-blind, placebo-controlled, multi-centre national study to evaluate the use of doxycycline in patients with at least one risk factor for worsening COVID-19.
As the study is double-blind, the study treatment and its placebo will be managed by the pharmacist of the coordinating centre, CHU Nantes. The oral anti-inflammatory dose of 200 mg of doxycycline or the matching placebo will be taken once a day in the evening one hour before going to bed, for 14 days. It should be noted that the placebo being used is marketed by COOPER; it is made of lactose and is very similar to the doxycycline tablet. The treatment (doxycycline or placebo) will be dispensed at the inclusion/randomisation visit. No dose adjustment is required. The treatment will end if the patient is hospitalised between day 0 (D0) and day 14 (D14).
Six French university hospitals: CHU Nantes, AP-HP [Assistance Publique – Hôpitaux de Paris; Hôpital Avicenne], CHU Bordeaux, CHU Caen, CHU Dijon and CHU Grenoble, will include 330 patients to provide at least 280 analysable patients. The inclusion period may be three to six months depending on the recruitment rate. Advertising, approved by the ethics committee, will be displayed to improve the recruitment rate. The treatment period is 14 days per patient and the last patient visit will be at 28 days (+/- 2 days) from the start of treatment.
Randomisation will be stratified by centre and by severity factors (1 versus ≥2). It will be performed according to a 1:1 ratio and balanced by blocks; the block size was random. The random numbers will be generated by computer. The software used to collect the data from the electronic report form allocated the patients automatically. Randomisation was performed according to the Interactive Web Response System.
Subjects are randomised into blocks as the allocation progresses, with a block being a sub-group of a predetermined size which contains a random allocation of patients. The software used for the randomisation is R version 9.4. An email alert will notify the investigator’s team and the pharmacist that a patient has been randomised. For the pharmacist, the allocation will be revealed. The randomisation key is known only to the biostatistician and the data managers, to make it impossible for the investigator to assign a particular treatment. It should be noted that the biostatistician who carries out the randomisation is different to the biostatistician who will perform the statistical analysis.
The study population is made up of male or female patients over 45 years of age, suspected of having COVID-19 with one or more risk factors for worsening the disease. They may be included in the study after a positive SARS-CoV-2 PCR test performed at the hospital. We chose 45 years as the minimum age in this study because this is the low end of the range for COVID-19-positive patients likely to be admitted to intensive care units . In order to be eligible, as already mentioned, these patients, older than 45, must have one or more risk factors for worsening the disease as described by HCSP : 70 years of age or older, cardiovascular history (stroke, coronary artery disease, complicated hypertension, cardiac surgery, NYHA class III or IV heart failure), poorly controlled insulin-dependent diabetes or complicated secondary diabetes, respiratory disease likely to undergo decompensation due to viral infection, patients with chronic renal failure on dialysis, and cancer patients under treatment. We have also added BMI (BMI>30) as a risk criterion, based on recent data from the literature [11–13,32]. Exclusion criteria are vulnerable populations and contraindication of the use of doxycycline. Box 1 presents all of the inclusion/exclusion criteria.
Furthermore, patients cannot be included in any other interventional research, unless they are hospitalised, and provided there is no drug interaction with any other experimental treatment. Patients who are hospitalised will be considered as a failed patient for the primary objective. These patients will then be followed up only for the management of this SAE. If hospitalisation occurs between D2 and D7, the PCR will not be performed at D7 and the corresponding objective will not be performed for this patient.
The plan for the study described in this section is presented in Figure 1 and the flowchart in Figure 2. The screening is performed by COVID-19 emergency facilities or the COVID-19 unit of the investigating centre.
A patient suspected of having COVID-19, with one or more characteristic clinical signs, and having one or more risk factors for worsening the disease is sent to hospital for a SARS-CoV-2 test. In the waiting room of the department specially dedicated to SARS-CoV-2 PCRs, a poster presenting DYNAMIC will be displayed. If the PCR is positive, either the patient who has seen the poster will contact the DYNAMIC team or the investigation team will contact the patient because of the PCR result. The hospital investigator explains the DYNAMIC study to the patient orally, checks the eligibility criteria, provides information on DYNAMIC and obtains the patient’s signed informed consent. If the version of the informed consent changes during the patient's participation in the study, they will be informed of this immediately during the telephone calls held during the trial, and the new consent will be signed at the next consultation. The Supplementary Material file contains the French informed consent form that patients sign prior to their inclusion in the trial (the updated protocol is version 1.8 as of 28 October 2020). Furthermore, blood samples for serology and immunological markers will be collected, stored and kept for Dynamic study if the patient signed in addition the biocollection informed consent. This second informed consent form states that the samples may be used for scientific research. This biocollection and its consent procedure have been registered with French ethics committee CPP (Comité de protection des Personnes) Ouest IV under number DC-2011-1399. After a clinical examination, patients will be randomised either in the doxycycline arm or in the placebo arm. The patients, the investigators and their teams will be blind and will not know the assigned treatment. As Karanicolas et al. pointed out in their article, the biostatisticians will also be blinded until the analyses are performed to reduce the study bias .
On D3 and D14, a telephone call will be made by the investigator or their team. A clinical assessment of the patient will be carried out using the questionnaire shown in Figure 3.
Furthermore, the patient will answer daily this questionnaire and if their answers indicate warning signs they will call the investigators or the emergency department. If necessary, a face-to-face consultation will be conducted immediately after this call to accurately assess the extent to which the disease has worsened.
On D7, a face-to-face follow-up visit will be performed. After a clinical examination, a SARS-CoV-2 qualitative PCR test will be carried out. On D28, the face-to-face end-of-study visit will be performed. A blood sample will be kept for serology and other immunological markers. The patient will return any unused therapeutic units and the empty blister packs.
An early end-of-study visit will be carried out corresponding to the end-of-study visit (D28 visit), with the exception of serology if the patient leaves the study before 14 days. For included patients hospitalised before 28 days, the experimental treatment is stopped as soon as they are hospitalised and they are considered a failure. Clinical data related to their hospitalisation will be reported in the CRF. For patients hospitalised after 28 days and up to 90 days post-treatment, they will be followed up until discharged. This follow-up will include the collection of vigilance data and various hospitalisation data (date of end of hospitalisation, length of hospitalisation in an intensive care unit, duration of mechanical respiratory support, possible death).
During the trial, any drug judged necessary for the patient’s well-being, which should not interfere with the evaluation of the test drug, may be administered at the investigator’s discretion. Absorption of doxycycline is impaired by antacids containing aluminium and calcium, as well as magnesium, bismuth subsalicylate and iron- and zinc-containing preparations. The study drug should therefore be taken at a different time to antacids. Other prohibited treatments are those prohibited by the Summary of Product Characteristics of doxycycline and are listed in the exclusion criteria (see Box 1). However, in the absence of a change to the therapeutic management of a patient arriving at hospital according to their treatment arm, a blinding procedure is not required in this protocol. Treatment will be stopped if hospitalisation occurs.
The embedded testing strategy will be used to successively test:
- The decrease in the percentage of patients requiring hospitalisation related to SARS-CoV-2 infection after at least 48 hours of treatment.
- A decrease in the percentage of patients requiring mechanical ventilator support if the first null hypothesis is rejected.
An intermediate futility analysis will be carried out after 100 patients have been evaluated, with the objective of stopping the clinical trial if there is no probability that the null hypothesis will be rejected at the end of the full study. The sole response to this analysis is whether or not to stop the trial because of a lack of efficacy. The Peto method will be used with an alpha risk of 1 per 1,000 at the intermediate analysis and of 5% at the final analysis . As already quoted, the biostatistician will be part of the blinded team. They will give the results of this intermediate analysis to the Data and Safety Monitoring Committee (DSMC), which may or may not unblind depending on the results.
The analysis will be stratified by centre and number of severity factors (1 versus ≥ 2). The severity factors are ≥ 70 years, BMI > 30, cardiovascular history (stroke, coronary artery disease, complicated hypertension, cardiac surgery, NYHA class III or IV heart failure), poorly controlled insulin-dependent diabetes or complicated secondary diabetes, respiratory disease likely to undergo decompensation due to viral infection, patients with chronic renal failure on dialysis, and cancer patients under treatment at the time of inclusion.
A modified intention-to-treat analysis will be applied. It will include all eligible randomised patients who provided their consent to participate in the trial. It will exclude patients who withdrew their consent prior to the assessment of the first primary endpoint, randomised patients who were incorrectly included (inclusion or exclusion criteria not met) and patients who did not receive any dose of treatment.
A per-protocol analysis will also be performed, including all patients evaluated in their treatment group who received more than 75% of the total treatment dose.
For the main outcome (percentage of patients), we will use a Mantel-Haenszel test (stratified on centres and on the number of severity factors) to compare the rate of success (success means no hospitalisation). If the null hypothesis is rejected, we will compare the rate of mechanical ventilator between the two arms. [35,36]. For the quantitative secondary criteria, a linear model stratified on hospital centres and on the number of risk factors will be developed  if the minimum number of recruitments per centre allows, otherwise the model will be stratified across centres and adjusted for the number of risk factors. The statistical software used will be SAS and R, version 3.8 or later. The significance thresholds will be set at 5% in a bilateral situation.
It is estimated that 25% of the eligible patients in the study will require hospitalisation. In the first wave, hospitalisation rates ranged from 18.1%  to 31.4% , we chose the mean. It is hypothesised that with doxycycline treatment, this rate would decrease to 12%. Indeed, the article concerning the effect of doxycycline on dengue patients: "Dengue Patients Treated With Doxycycline Showed Lower Mortality Associated with a Reduction in IL-6 and TNF Levels" , shows that the group on doxycycline had a lower mortality rate than those in the untreated group (11.2% [13/116] vs 20.9% [24/115], respectively, p=0.05). In addition, doxycycline administration resulted in a significant drop in IL-6 and TNF. These results therefore confirm our hypothesis of reducing the hospitalisation rate to 12%.
An embedded testing strategy is used for the primary objective, and as the report of the European Medicine Agency stated, no adjustment is needed for multiplicity .
For the second embedded hypothesis, it is estimated that 23% (https://www.data.gouv.fr/fr/datasets/donnees-hospitalieres-relatives-a-lepidemie-de-covid-19/) of hospitalised patients are admitted to intensive care units. With the recruitment level we have, we will have a power of 80% to show a decrease from 23% to 10%.
Under these hypotheses, with a 5% alpha risk in a bilateral situation and a power of 80%, we would need to evaluate a total of 280 patients in both arms. If the first hypothesis is rejected, the power of the test of the second hypothesis would then be 83.5%. Three hundred and thirty (330) patients will be included to compensate for study dropouts and non-assessable patients.
Adverse event management
For doxycycline, the main AEs expected are skin disorders (photosensitivity reaction, rash), immune system disorders (urticaria, rash, pruritus, angioedema, anaphylactic reaction) or digestive disorders (nausea, epigastralgia, diarrhoea, anorexia, glossitis, enterocolitis, and anal or genital candidiasis). There have been no reports of overdose. AEs reported for other tetracyclines following renal impairment (hepatic toxicity, hyperazotemia, hyperphosphataemia, acidosis), are unlikely to occur with doxycycline due to the lack of a change in blood levels in relation to the functional value of the kidney. For the placebo, the main expected effects are digestive disorders (excess gas, feeling of abdominal bloating, abdominal cramps and pain, and diarrhoea). As regards the pathology, patients with an uncomplicated viral infection of the upper respiratory tract may have non-specific symptoms such as fever, fatigue, cough (with or without discharge), anorexia, malaise, muscle aches, sore throat, dyspnoea, nasal congestion or headache. Rarely, patients may also experience diarrhoea, nausea and vomiting [6,41], and loss of taste or smell . The disease can worsen to pneumonia, acute respiratory distress syndrome , sepsis and septic shock  .
All SAEs, whether expected or unexpected, require the completion of an SAE report. The investigator must ensure that the information entered in this report is accurate and clear. The SAE should be reported to the sponsor immediately (within 24 hours of being highlighted by the investigator). After receiving an unexpected SAE report, the sponsor will notify the authorities. Furthermore, a DSMC has been set up; it is a consultative committee responsible for reviewing the safety of a study on behalf of the study sponsor and coordinator/principal investigator. Members of the committee who are competent in the field of clinical trials (pathology methodology and pharmacovigilance) are not involved in the study.
The DSMC is a referral point for pharmacovigilance if an SUSAR (Suspected Unexpected Serious Adverse Reaction) or SAE poses particular analytical difficulty, or if a doubt arises concerning the risk/benefit of the study. It will also make a decision on the outcome of the futility analysis. If the study is ended early following a decision of the DSMC or the study sponsor, the regulatory authorities and the ethics committee will be informed by post within a maximum of 15 days. In any event, written confirmation will be sent to the coordinating investigator of the study (specifying the reasons for early termination) and to the principal investigator of each centre, if applicable. All patients involved in the study will be informed and will be required to attend their early discharge visit.
- Ethical, regulatory and dissemination aspects
The clinical study will be conducted in accordance with the relevant versions of the French Public Health Code, national and international good clinical practice guidelines, and the Declaration of Helsinki.
In accordance with French law, the study protocol was submitted to French regulatory authority the ANSM (Agence Nationale de Sécurité du Médicament et des Produits de Santé). This clinical study was submitted to and approved by the ethics committee of Boulogne-Billancourt (CPP Ile de France VIII) on 13 May 2020. Requests for substantial modifications should be addressed by the sponsor for approval or notification to the ANSM and/or the ethics committee concerned in compliance with Law 2004-806 of 9 August 2004 and its implementing decrees. The amended protocol will be a dated and updated version. If necessary, the information form and consent form will be amended.
All of the submissions/declarations were made by the Sponsor Department at CHU Nantes, which manages the quality of the data collected. The data collected during the study will be processed electronically in accordance with the requirements of French Data Protection Authority the CNIL (in compliance with French Reference Methodology MR001). Data will only be shared between the investigators. However, the datasets analysed during this study will be available from the corresponding author upon reasonable request.
An electronic Case Report Form (eCRF) shall be drawn up for each included patient. Subjects will be identified by the first letter of their family name, the first letter of their first name, the centre number and the inclusion number. This code should be the only information featured on the eCRF enabling a retrospective link to the patient. The investigator, or their team, shall also encode the patient data on any documents that may be in their possession (imaging, biology test reports, etc.) attached to the eCRF. At the end of the study, database reconciliation is carried out between the CRF database and the safety database. This reconciliation is performed before database locking. Similarly, an annual reconciliation is carried out when updating the Annual Safety Report.
As required, the sponsor has provided an insurance policy to cover the financial consequences of its civil liability in accordance with the regulations. This protocol was created thanks to a Scientific Committee. The Scientific Committee is coordinated by Professor B Dréno and its members include external and internal experts in COVID-19, experts in clinical trials and a methodologist. An inspection or audit may take place as part of this study, performed by the sponsor and/or the regulatory authorities. Inspectors will check the documents, logistics, records and any other resources the authorities consider to be associated with the clinical trial and that may be located at the trial site itself.
The trial results will be published in international, medical and scientific journals and presented at national and international conferences. The investigators will follow the rules and guidelines of the International Committee of Medical Journal Editors . In practice, the Scientific Committee will be among the authors of the publication, as will the investigators who included the patients in the trial.