COVID-19 trials in Italy: A call for simplicity, top standards and global pooling

Abstract

The novel coronavirus disease, affecting ~9 million people in the past five months and causing >460,000 deaths worldwide, is completely new to mankind. More than 2,000 research projects registered at ClinTrials.gov are aiming at finding effective treatments for rapid transfer to clinical practice. Unfortunately, just few studies have a sufficiently valid design to provide reliable information for clinical practice.

In Italy - the first western country affected by the pandemia – 35 studies have been approved by the Italian Drug Agency. We here summarise the study protocols and critically appraise their design, assumptions and endpoints. Currently, about one in seven approved Italian studies has a sufficiently valid design to provide reliable information on the benefit/risk profile of the proposed treatment. Because most treatments proposed to date represent nonspecific repurposing of available compounds, sensational results cannot be expected; rather, small to moderate possible favourable effects. For this reason, large, simple, randomised trials using highest research standards are advocated. Additionally, systematic descriptions of national protocols may allow global pooling of trial data with common designs.

The novel coronavirus disease (COVID-19) has affected ~9 million people in the past five months, overwhelmed frontline professionals with patient care and risk of infection, and caused over 460,000 deaths worldwide. Less than a year ago, COVID-19 was unknown to man. Over 2,000 research projects registered at ClinTrials.gov are aiming at finding effective preventive or curative treatments for rapid transfer to clinical practice. Authoritative observers, however, have noted flaws in many of these projects, advocating higher research standards [[1], [2], [3], [4]].

In Italy, the first European country affected by COVID-19, the need for research was perceived as very urgent. Appropriately, the Health Ministry simplified bureaucracy by entrusting project evaluation to the Italian Drug Agency (AIFA) and final approval to a single national Ethics Committee. Laudably, transparency was ensured by publication of the approved protocols on AIFA’s website [5]. We appraised the 35 studies approved between March 11 and May 22, 2020, assessing design, assumptions, endpoints and sample size (Table 1 ).

Table 1.

COVID-19 studies approved in Italy up to June 22, 2020.

Study	Treatment	Patients	Type of study	Blind or Open label	Primary endpoint	Assumption of benefit	Sample size (subjects)	Authors’ overall appraisal
GILEAD GS-US 540-5773	Remdesevir 5 vs 10 days vs SOC	COVID-19 + Hospitalised SpO2 ≤94%	Randomised	Open	Normal body temperature and SpO2 at 14 days	45% for 5 day Rx 60% for 10 day Rx	400	No control group. Soft endpoint. Optimistic assumption of efficacy. Probably underpowered
GILEAD GS-US 540-5774	Remdesevir 5 vs 10 days vs SOC	COVID-19 + Hospitalised SpO2 ≤94%	Randomised	Open	% discharged at 14 days	25% increase with Remdesevir	600	Intermediate endpoint. Adequately sized
TOCIVID	Tocilizumab	COVID-19 + Hospitalised	Observational		Death at 14 and 30 days	ARR 10%	330	No control group
Sobi-IMMUNO-101	Emapalumab vs Anakinra vs SOC	COVID-19 + Hospitalised SpO2 ≤93% Hyperinflammation	Randomised	Open	% without invasive ventilation or ECMO	60% increase vs SOC	54	Optimistic assumption of efficacy. Probably underpowered
Sarilumab COVID-19	Sarilumab vs Placebo	COVID-19 + Hospitalised	Randomised	Double blind	Not reported	Not reported
RCT-TCZ-COVID-19	Tocilizumab + SOC vs SOC + Tocilizumab in case of clinical deterioration	COVID-19 + Hospitalised Pneumonia	Randomised	Open	Occurrence of >1: -death -invasive ventilation -respiratory decline	50% reduction in primary endpoint occurrence	398	Optimistic assumption of efficacy. Probably underpowered
Tocilizumab 2020-001154-22	Tocilizumab vs Placebo	COVID-19 + Hospitalised Pneumonia SpO2 ≤93%	Randomised (2:1)	Double blind	Clinical status on a 7-category ordinal scale	2-day difference between treatment groups in time to >2 category improvement	330	Intermediate endpoint
Hydro-Stop-COVID19 Trial	HCQ 400 mg bid vs SOC	COVID-19 + Out-patients	Randomised	Open	Negative test at 8 days	From 15 to 60% (i.e., 400% increase) vs SOC	216	Soft endpoint. Optimistic assumption of efficacy.
SOLIDARITY WHO	5 arms: Remdesevir, CQ or HCQ, Lopinavir-Ritonavir, Lopinavir-Ritonavir + Interferon, SOC	COVID-19 + Hospitalised	Randomised adaptive design	Open	In-hospital mortality	15-20% reduction	>10,000	Hard endpoint. Adequately sized
COLVID 19	Colchicine vs SOC	COVID-19 + Hospitalised Pneumonia	Randomised	Open	Death or mechanical ventilation or ICU at 1 month	50% reduction	308	Optimistic assumption of efficacy. Probably underpowered
ColCOVID19	Colchicine vs SOC	COVID-19 + Hospitalized Pneumonia	Randomised	Open	Two-category improvement on 7-category scale at 14 days	50% improvement	310	Soft end-point. Optimistic assumption of efficacy. Probably underpowered
INHIXACOVID19	Enoxaparin	COVID-19 + Moderate/severe disease	Observational		Death at 30 days	Not defined	100	No control group. Inconclusive for the primary endpoint
BARICIVID-19	Baricitinib vs SOC	COVID-19 + Hospitalised SpO2 ≤93%	Randomised	Open	Invasive ventilation at 7 and 14 days	60% reduction	126	Optimistic assumption of efficacy. Probably underpowered
COPCOV	CQ or HCQ vs Placebo	Healthcare or other frontline workers	Randomised	Double blind	Symptomatic COVID-19 infection Symptom severity	23% reduction	40,000 (20,000 in Asia, 20,000 in Europe)	Hard endpoint. Adequately sized
COVID-SARI	Sarilumab	COVID-19 + Hospitalised Pneumonia Elevated D-Dimer	Observational		≥30% decrease in O2 requirement compared to baseline	Not defined	40	No control group
X-Covid 19	Enoxaparin vs SOC	COVID-19 + Hospitalised	Randomised	Open	Venous thromboembolism	33% reduction	2,712	Adequately sized
PROTECT	HCQ vs SOC	Prevention: Healthy subjects or workers in contact with COVID-19 pts Treatment: COVID-19 + Outpatients	Cluster randomisation (2:1)	Open	Prevention: rate of COVID-19 + at 30 days Treatment: rate of COVID-19 at 14 days	Prevention: 30% reduction Treatment: 50% improvement	Prevention: 3,000-4,000 Treatment: 600	Complex design. Prevention arm: adequately sized. Treatment arm: optimistic assumption of efficacy
ESCAPE	Sarilumab vs SOC	COVID-19 + Hospitalised Pneumonia	Randomised (2:1)	Open	Two-category improvement on 7-category scale at 14 days	37% reduction	171	Intermediate endpoint
XPORT-CoV-1001	Selinexor vs SOC	COVID-19 + Hospitalised SpO2 ≤94%	Randomised	Single blind	Time to clinical improvement	34% reduction	230	Intermediate endpoint
AMMURAVID	7 arms: HCQ, HCQ + Tocilizumab, HCQ + Sarilumab, HCQ + Siltuximab, HCQ + Canakinumab, HCQ + Baricitinib, HCQ + Methylprednisolone	COVID-19 + Hospitalised Pneumonia Elevated D-Dimer or hsCRP	Randomised adaptive design	Open	Severe respiratory failure (PaO2/FiO2 <200 mmHg) at day 10	Not defined	350	Exploratory study
HS216C17	Favipiravir vs Placebo	COVID-19 + Pneumonia	Randomised	Double blind	Time to clinical recovery	56% improvement	256	Soft endpoint. Optimistic assumption of efficacy. Probably underpowered
FibroCov	Pamrevlumab vs SOC	COVID-19 + Hospitalised Pneumonia Supplemental O2	Randomised	Open	% not on ventilatory support ≤15 days	60% improvement	68	Optimistic assumption of efficacy. Probably underpowered
AZI-RCT-Covid19	HCQ vs HCQ + Azithromicyn	COVID-19 + Hospitalised Pneumonia SpO2 <93%	Randomised	Open	Clinical recovery at 10 days	29% improvement	144	Probably underpowered
CAN-Covid	Canakinumab vs Placebo	COVID-19 + Hospitalised Pneumonia SpO2 ≤93%	Randomised	Double blind	Survival free of invasive ventilation at day 29	15% absolute improvement Between 30 to 75% relative risk improvement	450	Optimistic assumption of efficacy. Probably underpowered
ARCO-Home	4 arms: Darunavir-Cobicistat, Lopinavir-Ritonavir, Favipiravir, HCQ	COVID-19 + Outpatients	Randomised adaptive design	Open	Virologic endpoint: Negative test at 7 days Clinical endpoint: % hospitalized at 14 days	Virologic endpoint: 100% improvement Clinical endpoint: 50% improvement	From 175 to 435	Optimistic assumption of efficacy. Probably underpowered
DEFI-IVID 19	Defibrotide	COVID-19 + Hospitalised Pneumonia SpO2 ≤92%	Observational		Respiratory-failure rate	20% reduction	50	No control group
COMBAT-19	Mavrilimumab	COVID-19 + Hospitalised Pneumonia SpO2 ≤92% Elevated hsCRP	Randomised	Double blind	% not on O2 supplementation at day 14	100% increase	50	Optimistic assumption of efficacy. Probably underpowered
PRECOV	HCQ	Health professionals COVID-19 negative	Randomised	Open	% with positive test at day 28	50% reduction	1,000	Optimistic assumption of efficacy
DEF-IVID 19	Defibrotide	COVID-19 + Hospitalised Pneumonia SpO2 ≤92%	Observational		% respiratory failure rate	20% reduction	50	No control group
EMOS-COVID	Enoxaparine low vs high dose	COVID-19 + Hospitalised Pneumonia PaO2/FiO2<250 Elevated D-Dimer	Randomized	Open	% mortality or respiratory failure	33% reduction	300	All patients treated with enoxaparine
STAUNCH	3 arms: steroids and unfractionated heparin vs steroids and LMWH vs LMWH alone	COVID-19 + Positive pressure ventilation >24h and invasive mechanical ventilation <96h P/F ratio <150 D-dimer and hsCRP >6 x upper limits	Randomised	Open	Death at 28 days	25% reduction	210	Probably underpowered. Very high mortality assumption for LMWH alone
TOFACOV-2	Tofacitinib + HCQ vs HCQ alone	COVID-19 + Hospitalised Interstitial pneumonia	Randomised	Open	% needing mechanical ventilation	75% reduction	116	Optimistic assumption of efficacy. Probably underpowered
CHOICE-19	Colchicine vs SOC	COVID-19 +	Randomised	Open	% hospitalised at 30 days	50% reduction	438	Optimistic assumption of efficacy. Probably underpowered
COVID-19 HD	LMWH high vs low dose	COVID-19 + Hospitalised Pneumonia SpO2 ≤93% D-dimer >4 x upper limit	Randomised	Open	In-hospital clinical worsening	50% reduction	300	Optimistic assumption of efficacy. Probably underpowered
IVIG/H/Covid-19	Intravenous polyvalent immunoglobulin	COVID-19 + Hospitalised Pneumonia	Observational		Survival at 3 and 6 months	Pilot study: not defined	30	No control group

ARR=absolute risk reduction, COVID-19=2019 coronavirus disease, CQ=cloroquine, no.=number, ECMO=extracorporeal membrane oxygenation, hsCRP=high sensitivity C-reactive protein, HCQ=Hydroxychloroquine, ICU=intensive care unit, LMWH= low molecular weight heparin, O2=oxygen, pts=patients, SOC=standard of care, Rx=treatment, SpO2= percutaneous oxygen saturation, vs=versus.

Most Italian studies focus on severe hospitalised COVID-19 patients and on antiviral, anti-inflammatory or antithrombotic treatments. Only a minority deals with outpatients or disease prevention. Twenty-nine (83%) are randomised, but 22 are open label, and more than half of these are susceptible to biased endpoint evaluation. Nineteen of the 29 randomised studies (66%) are small, based on over-optimistic assumptions of benefit, with a high risk of inconclusive results, even for potentially favourable treatments. Six studies (17%) are observational without appropriate control groups. Only 5 (14%) show a sufficiently adequate overall design to provide reliable results for application in clinical practice (Table 1).

Current COVID-19 study drugs represent nonspecific repurposing of available compounds [6]. Nonspecific treatments cannot be expected to yield sensational benefits; rather, small to moderate ones. For COVID-19, on the other hand, even small-to-moderate treatment effects leading to even small relative mortality reductions could have an enormous impact on the absolute number of survivors. A reliable demonstration of moderate treatment benefits and of potential subgroup effects (e.g., by age, sex, comorbidities or disease severity) requires testing in thousands of patients. Currently, about 1 in seven approved Italian studies has a sufficiently valid design to provide reliable information on the benefit/risk profile of the proposed treatment.

The European Medicines Agency recently called for adequately sized COVID-19 trials to produce decision-relevant results [7 ]. A systematic description of all national trials might show overlapping designs across countries that, if valid, might allow pooling of individual patient data. While waiting for an effective vaccine, the crucial question is: will current trial results produce sufficiently reliable evidence on effective and safe preventive/therapeutic approaches to face, potentially next autumn, a relapse of the infection? The answer is hopefully yes, but only thanks to the currently few adequately designed large-scale randomised trials [[8], [9], [10]].

Author contribution

All authors contributed to the critical evaluation of the studies approved in Italy and to the whole content of the manuscript.

Funding

None.

Declaration of Competing Interest

APM received personal fees from Bayer, Fresenius, Novartis for the participation in study committees outside the present work. FA has received personal fees from Amgen, Bayer, BMS-Pfizer and Daiichi Sankyo outside the present work. GDP has received personal fees for lectures from Boheringer Ingelheim, Bayer, BMS-Pfizer and Daiichi Sankyo outside the present work. CG has received personal fees from MSD, ViiV, Gilead and Janseen Cilag for services outside the present work.