Source: Mediterranee Infection,
Full Study Document in PDF: Natural history of COVID-19 and therapeutic options
Raoult, et al.: “After confirming in our cohort that zinc deficiency was associated with an unfavorable prognosis, and in view of a comparative study in favor of zinc, 15 mg of zinc three times daily was added.”
“Treatment with an oral combination of hydroxychloroquine, azithromycin and zinc may REPRESENT THE BEST CURRENT THERAPEUTIC OPTION in relation to its antiviral and immunomodulatory effects’’.
Abstract
Introduction
28 COVID-19 presents benign forms in young patients who frequently present with anosmia.
29 Infants are rarely infected, while severe forms occur in patients over 65 years of age with
30 comorbidities, including hypertension and diabetes. Lymphopenia, eosinopenia,
31 thrombopenia, increased lactate dehydrogenase, troponin, C-reactive protein, D-dimers, and
32 low zinc levels are associated with severity.
33 Areas covered
34 The authors review the literature and provide an overview of the current state of knowledge
35 regarding the natural history of and therapeutic options for COVID-19.
36 Expert opinion
37 Diagnosis should rely on PCR and not on clinical presumption. Because of discrepancies
38 between clinical symptoms, oxygen saturation or radiological signs on CT scans, pulse
39 oximetry and radiological investigation should be systematic. The disease evolves in
40 successive phases: an acute virological phase, and, in some patients, a cytokine storm phase;
41 an uncontrolled coagulopathy; and an acute respiratory distress syndrome. Therapeutic
42 options include antivirals, oxygen therapy, immunomodulators, anticoagulants and prolonged
43 mechanical treatment. Early diagnosis, care, and implementation of an antiviral treatment; the
44 use of immunomodulators at a later stage; and the quality of intensive care are critical
45 regarding mortality rates. The higher mortality observed in Western countries remains
46 unexplained. Pulmonary fibrosis may occur in some patients. Its future is unpredictable.
47
48 Keywords: SARS-CoV-2, COVID, Humans, Pathophysiology, Treatment,
49 Hydroxychloroquine, Azithromycin, Tocilizumab, Remdesivir, Care
Highlights
51 Diagnosis of COVID-19 should rely on PCR and not on clinical presumption.
52 Because of “happy hypoxia”, pulse oximetry and radiological investigation should be
53 systematic.
54The disease evolves in successive phases: an acute virological phase, and, in some
55 patients, a cytokine storm phase; an uncontrolled coagulopathy; and an acute
56 respiratory distress syndrome.
57Massive screening allowing early diagnosis, care (oxygen, anticoagulants), and
58 implementation of an antiviral treatment; the use of immunomodulators at a later
59 stage; and the quality of intensive care are critical regarding mortality rates.
60 Intravenous catheterization should be avoided, and early oral treatment in an
61 outpatient basis should be preferred.
62 Treatment with an oral combination of hydroxychloroquine, azithromycin and zinc
63 may represent the best current therapeutic option in relation to its antiviral and
64 immunomodulatory effects.
65Preventive anticoagulants should be prescribed in patients with coagulopathy (positive
66 D-dimers).
Introduction
68 COVID-19 has a pleomorphic clinical presentation including asymptomatic individuals and
69 patients with mild to severe involvement with several evolutionary stages [1-3]. Age and
70 comorbidities including, notably, hypertension, diabetes and coronary heart disease are the
71 main risk factors for evolving toward severe infections [1-3]. Schematically, after the
72 incubation period, two main clinical presentations can occur: upper respiratory tract infections
73 (URTIs) with severe headaches, anosmia, ageusia (or dysgeusia) and rhinitis, which are
74 mainly observed in young patients who then have a good clinical outcome; and lower
75 respiratory tract infections (LRTIs) with pneumonia symptoms that are observed more
76 frequently in patients with comorbidities and can be severe to fatal in older patients [1-3]. At
77 admission, prognosis can be assessed through the National Early Warning Score (NEWS-2), a
78 simple aggregate scoring system including respiration rate, oxygen saturation, systolic blood
79 pressure, pulse rate, level of consciousness or new confusion, and temperature. Age has been
80 added in a modified version of this score [4]
.
81 During the onset of the COVID-19 outbreak, olfactory and gustative disorders,
82 including anosmia and ageusia were described in infected patients [5]. In Marseille, 3,497
83 adults who underwent PCR between 24 March and 25 April 2020 were asked the following
84 question prior to being tested for SARS-CoV-2: “Have you lost your sense of smell or taste in
85 the past two months?” The prevalence of the loss of smell and/or taste in COVID-19 patients
86 was 356/673 (53%), and the positive predictive value (PPV) for the diagnosis of COVID-19
87 by PCR was 67% when smell and taste disorders were reported (submitted). Asking patients
88 and healthcare workers (HCWs) about loss of smell and taste could be useful in areas where
89 testing for SARS-CoV-2 is politically or technically limited or impossible. Interestingly,
90 “happy hypoxemia”, a hypoxia observed in patients who are SARS-CoV-2 positive yet
91 comfortable and without dyspnea emphasizes the need to perform a low-dose CT-scan on
most patients to detect pneumonia at an early stage [6]. Most COVID patients are definitively
93 cured, but extreme caution is needed in patients with comorbidities and/or biological
94 parameter abnormalities such as lymphopenia, eosinopenia, increased D-dimers, troponin,
95 lactate dehydrogenase (LDH) or C-reactive protein (CRP) [1, 3]. Venous thromboembolism is
96 relatively common [7], is mainly characterized by pulmonary embolism, and is found in up to
97 one-third of critical cases [8]. Acute respiratory distress syndrome (ARDS), pulmonary
98 embolism and bacterial superinfection may result in a fatal evolution [9, 10]. Finally, delayed
99 pulmonary fibrosis may occur in an as yet unknown proportion of patients [11].
100 At the beginning of the health crisis, the use of chest X-rays was restricted to patients in
101 intensive care units due to its low value in detecting ground-glass opacities. However, low
102 dose chest computed tomography (LDCT) appears to be a useful tool in the management of
103 patients with regards to diagnosing, assessing and quantifying disease severity and for
104 differential diagnosis. LDCT might be of interest in predicting lung fibrosis during healing
105 [12-15]. The main findings of COVID-19 pneumonia on chest CT include ground-glass
106 opacities, consolidation, and a crazy-paving pattern. These features are not specific, but the
107 distribution of lesions during COVID-19 pneumonia is more likely to be peripheral,
108 asymmetric and located in the lower lobes [16]. CT features revealed a good sensitivity and
109 specificity for COVID-19 diseases in centers where CT was used as a diagnostic tool [17].
110 Furthermore, Li et al. developed a deep learning algorithm able to discriminate COVID-19
111 pneumonia from community-acquired pneumonia, with good results (Figure 1) [18]. COVID
112 19 pneumonia is also characterized by the high prevalence of lung involvement in
113 paucisymptomatic patients. In our center, we decided to perform LDCT on all patients with a
114 positive PCR for COVID-19. Of the 2,065 LDCTs that were performed on COVID-19
115 patients, more than 70% revealed pneumonia. Of the 1,043 patients with a NEWS-2 score=0
116 who underwent LDCT, 628 (60.2%) had radiological abnormalities, including 494 (47.4%)
117 with minimal lung lesions, 118 (11.8%) with intermediate lesions and 11 (1%) with severe
118 lesions. Moreover, of the 1,370 LDCTs performed on patients without perceived dyspnea, 937
119 (68%) had pneumonia [3].
120 Symptoms at the time of diagnosis of COVID-19 pneumonia do not appear to be related
121 to prognosis [1]. A meta-analysis of 1,558 patients found that significant risk factors for
122 mortality in COVID-19 were hypertension, diabetes, chronic obstructive pulmonary disease,
123 cardiovascular disease, and cerebrovascular disease [19, 20]. A study on 1,591 patients in the
124 intensive care unit showed that the mortality rate was higher in patients over the age of 64
125 than in younger patients [20]. Furthermore, we showed that the percentage of lung
126 involvement quantified using the deep learning algorithm is an independent prognostic
127 marker, and the addition of lesion quantification significantly enhances the prediction model
128 based on comorbidities and NEWS-2 score (unpublished).
129 Reports of an increased incidence of acute pulmonary embolisms or intravascular
130 coagulopathy associated with COVID-19 have emerged in the literature [21]. The prevalence
131 of pulmonary embolism in COVID-19 has been reported to be approximately 20% in patients
132 with severe disease. Leonard-Lorant et al. found that a D-dimer threshold higher than 2,660
133 µg/L could detect all patients with a pulmonary embolus on chest CT after contrast injection
134 [8].
135 COVID-19 might lead to sequelae such as lung fibrosis during the healing phase [22]
.
136 The real prevalence and clinical impact of COVID-19 sequelae on the lungs, as well as on the
137 myocardium, requires further study.
138 Literature search methodology
139 A literature search was performed using the following keywords: SARS-CoV-2, COVID,
140 coronavirus, pathophysiology, natural history, treatment, and humans without restriction of
141 the date or language. Medline, Google, and Google Scholar were used alongside
142 crossreferencing.
143 2. SARS-CoV-2 epidemiology
144 The first COVID-19 cases were identified in late December 2019 in Wuhan, China, and the
145 disease turned into a pandemic within a few weeks. As of 21 July 2020, more than fourteen
146 million cases have been reported globally, with more than 600,000 deaths.
147
(https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423
148 467b48e9ecf6). The SARS-CoV-2 epidemic exhibits a bell-shaped incidence curve [23, 24]
149
(https://coronavirus.jhu.edu/data/new-cases; https://www.mediterranee-infection.com/covid
150 19/; Figure 2), which is a typical epidemic curve. In Western countries in the Northern
151 hemisphere, the outbreak decreased dramatically during the spring, as is the case for
152 epidemics arising from other respiratory viruses. Moreover, it affected people differently
153 according to their age. Very few cases and less severe outcomes were observed in children
154 [25-31], while SARS-CoV-2 infections have been more frequent and severe in the elderly. For
155 instance, a large study conducted in Iceland on the general population found that children
156 under the age of 10 years were half as likely to be diagnosed with SARS-CoV-2 than children
157 over the age of 10 years or adults [31]. In addition, targeted screening did not diagnose
158 infections in children under the age of 10 years. This predominance of cases among adults
159 differs from the age distributions observed with other respiratory viruses, including endemic
160 coronaviruses [32]. Interestingly, several studies have detected immune responses to SARS
161 CoV-2 in unexposed individuals. Thus, Grifoni et al. detected circulating SARS-CoV-2-
162 specific CD4
+
and CD8
+
T cells in ≈20-60% of SARS-CoV-2-unexposed individuals sampled
163 between 2015 and 2018 [33], and 10% of uninfected pregnant women exhibited IgG to
164 SARS-CoV-2 [34]. Conversely, increased IgG reactivity to endemic coronaviruses was
165 reported in SARS-CoV-2 infections, further suggesting crossimmunity [35]. As the four
166 coronaviruses 229E, NL63, OC43 and HKU1 circulate endemically worldwide and massively
167 infect humans during the first years of life [32], exposure to these viruses may have conferred
168 crossimmunity to SARS-CoV-2 in a preferential way to children.
169 3. Natural history of the disease (Figure 3)
170 3.1 Transmission route
171 The main transmission route for SARS-CoV-2 is human-to-human transmission via
172 respiratory droplets and skin contact, with incubation times of 2–14 days (mean incubation
173 time of approximately 6 days) [36, 37]. In addition to its presence in nasopharyngeal swabs,
174 bronchoalveolar lavage, and sputum, SARS-CoV-2 has been detected using molecular tools in
175 saliva, stools, urine, blood, tears, and conjunctival secretions [36, 38-40]. Live viruses have
176 been detected in feces, including nondiarrheal feces, implying that SARS-CoV-2 could be
177 transmitted through feces [36]. Its molecular detection around the toilet (doorknob, surface of
178 the toilet bowl, internal sink bowl) in the room of a patient who did not have diarrhea but who
179 did have positive stool samples for SARS-CoV-2, supports the hypothesis that fecal viral
180 shedding may be involved in transmission [41]. Live viruses have also been observed in saliva
181 [38]. Since saliva can be released through coughing and the influenza virus can be present in
182 respiratory droplets even during normal breathing, SARS-CoV-2 could be transmitted by
183 saliva directly or indirectly, even in patients without a cough or other respiratory symptoms
184 [38]. It has already been reported that asymptomatic carriers can spread SARS-CoV-2 [42].
185 Although SARS-CoV-2 has been detected in the tears and conjunctival secretions of COVID
186 19 patients with conjunctivitis, its transmission through the conjunctival route is debated [39].
187 SARS-CoV-2 can survive outside the body for long periods of time; it can remain viable
188 in aerosols for up to three hours [43]. Viable SARS-CoV-2 was detected up to 72 hours after
189 experimental application to plastic and stainless steel. On copper and cardboard, no viable
190 SARS-CoV-2 was measured after four hours and 24 hours, respectively. Thus, transmission of
191 SARS-CoV-2 by aerosols and fomites is likely. Investigations of a COVID-19 cluster in a
192 shopping mall support the hypothesis that the rapid spread of SARS-CoV-2 could result from
193 spread via fomites (elevator buttons, bathroom taps) or virus aerosolization in a confined
194 public spaces (bathrooms, elevators) [44].
195 Nosocomial transmission including outbreaks of SARS-CoV-2 have occurred. Although
196 direct transmission is the most common route of transmission, contaminated surfaces that are
197 frequently contacted in healthcare facilities are a potential source. An extensive
198 environmental study was performed in an intensive care unit (ICU) and general ward (GW) at
199 Wuhan Huoshenshan Hospital [45]. Nucleic acids from SARS-CoV-2 were observed on
200 surfaces that were frequently touched, such as computer mice, rubbish bins, and handrails on
201 patients’ beds, but were also found on floors and were sporadically observed on doorknobs
202 and on the sleeve cuffs and gloves of medical staff. Nucleic acids were also detected in the air
203 from a patient’s room, with an estimate that the transmission distance of SARS-CoV-2 could
204 be around four meters in a GW. However, since the detection of nucleic acids does not
205 indicate the amount of viable virus and the minimum infectious dose is unknown, the distance
206 of aerosol transmission cannot be strictly determined. These data imply a risk of infection for
207 medical personnel and other close contacts. However, appropriate precautions and adherence
208 to hand and environmental hygiene can effectively prevent infection since no member of the
209 hospital staff was infected with SARS-CoV-2 as of 30 March 2020 [45].
210 3.2 Cellular life cycle of SARS-CoV-2 (Figure 4)
211 The SARS-CoV-2 is an enveloped RNA surrounded by spike glycoproteins [46-49].
212 The virus enters cells through membrane fusion. The first step of the SARS-CoV-2 replicative
213 cycle is the attachment of the virus to the angiotensin-converting enzyme 2 (ACE2)
214 glycoprotein. The receptor-binding domain (RBD) amino acid sequences present in the S1
215 spike protein interact with the N-terminal region 30-41 and 82-93 of ACE2 that contains
216 several sites for N-glycosylation. A cell surface protease, TMPRSS2, is responsible for spike
217 cleavage allowing the appropriate conformation for the S2 spike to expose the hidden fusion
218 peptide for insertion into the cellular membrane lipid bilayers. The viral nucleocapsid is thus
219 delivered into the cytoplasm through the endocytic vesicle. After acidification of the late
220 endosome, the action of cathepsin enables the uncoating of the genomic RNA and the
221 enzymes necessary for its replication. The genomic RNA is used as a template by the
222 replicase to synthesize the negative sense genomic RNAs (anti-genome), which are used as
223 templates to synthesize the progeny positive sense genomes and subgenomic RNAs. Similarly
224 to SARS-CoV, the 5′-proximal two-thirds of the SARS-CoV-2 viral genome are translated
225 into polyproteins that give rise to several nonstructural proteins (Nsps) following
226 autoproteolytic processing. Among the Nsps, Nsp12 is an RNA-dependent RNA polymerase,
227 Nsp3 and Nsp5 are proteinases, Nsp13 is a helicase, Nsp14 and Nsp15 are ribonucleases, and
228 Nsp14 is a methyltransferase (involved in RNA cap formation). The 3′-proximal third
229 sequence serves as template for several subgenomic mRNAs that encode the viral structural
230 (the spike/S, the envelope/E, the membrane/M, and the nucleocapsid/N) and accessory
231 proteins. The S, E, and M proteins are synthesized and anchored on the endoplasmic
232 reticulum (ER) with the N protein translated in the cytosol. Posttranslational modifications of
233 viral proteins occur within the endoplasmic reticulum and trans-Golgi network vesicles. After
234 assembly in the ER-Golgi intermediate compartment (ERGIC), where the E protein plays an
235 essential role in virus assembly and the mature M protein shapes the virus, mature virions are
236 released from smooth-walled vesicles by exocytosis.
237 3.3 Immune response
238 Immune responses shape the clinical course of COVID-19. The hallmark of the disease
239 is the occurrence, in 10–20% of patients, of a sudden deterioration 7–10 days after the onset
240 of symptoms, increasing the risk of acute respiratory failure, organ support and ultimately a
241 fatal outcome [27]. Early publications described lymphopenia mostly affecting T and B cells,
242 neutrophilia, and decreased eosinophil and monocyte counts [10, 27]. The degree of
243 eosinopenia, lymphopenia and neutrophilia, the latter sometimes expressed as the neutrophil
244 to-lymphocyte ratio, was associated with and was predictive of clinical severity. In severe
245 cases, deficient induction of type I interferons at the initial stage, increased levels of some but
246 not all pro-inflammatory cytokines, most notably interleukin (IL)-6, increased levels of the
247 regulatory cytokine IL-10, an autoimmune signature, and an inconsistent specific antibody
248 response to SARS-CoV-2 were reported [10, 27, 50-52]. Tissue investigations showed that
249 SARS-CoV-2 pulmonary tropism targeted alveolar cells. Severe pulmonary lesions were
250 associated with interstitial mononuclear infiltrates dominated by lymphocytes, CD4
+
and
251 CD8
+
T-cells, pulmonary edema, hyaline formation and pneumocyte desquamation without
252 histopathological evidence of sequestrated eosinophils [53]. Resident or recruited
253 immunosuppressive and inflammatory monocytes and macrophages (CD14
+HLA-DRlo/neg)
254 tend to replace protective resident alveolar macrophages. Severe respiratory failure may arise
255 through two distinct pathways: either an atypical macrophage activation syndrome, or an
256 immune dysregulation status characterized by impaired monocyte activation and antigen
257 presentation [54]. The systemic passage of high levels of cytokines produced in the lungs may
258 contribute to a multivisceral failure syndrome [10]. Neutrophilia is associated with the
259 increased formation of extracellular traps (NETosis), contributing to inflammation, cytokine
260 dysregulation and autoimmune and thrombotic manifestations [55, 55]. SARS-CoV-2 was
261 reported to infect T lymphocytes and macrophages, resulting in impaired antigen presentation,
262 increased IL-6 production, and possibly in lymphocyte apoptosis in lymphoid organs [54, 56].
263 Th1 polarization is involved in the efficient control of SARS-CoV-1 but not SARS-CoV-2
264 infection [57]. Crossreactive T-cell recognition between circulating seasonal human
265 coronavirus (HCoVs), the SARS-CoV and SARS-CoV-2 and crossreactive antibodies
266 between SARS-CoV-2 and SARS-CoV have been reported [58-60]. Interestingly, CD4+ T
267 cells from COVID-19 patients targeted the N and C terminal regions of the S protein equally,
268 whereas CD4+ T cells from noninfected patients only targeted the C terminal region, which is
269 highly homologous with the seasonal HCoVs S protein [59]. Recently, Grifoni et al., used
270 HLA class I and II and predicted peptide megapools detected circulating SARS-CoV-2-
271 specific CD4
+
and CD8
+
T cells in 100% and ≈70% of patients with resolved COVID-19,
272 respectively [33]. Interestingly, such cells were also detected in ≈20–60% of SARS-CoV-2-
273 unexposed individuals collected between 2015 and 2018. It should also be noted that all 20
274 SARS-CoV-2-unexposed individuals exhibited IgG to HCoV-OC43 and HCoV-NL63.
275 Another study showed the presence of anti-SARS-CoV-2 humoral responses elicited by
276 patients previously infected with endemic coronaviruses [34].
277 It is thus hypothesized that past infection with HCoVs causing upper respiratory infection
278 may confer SARS-CoV-2 cellular immunity as the result of CD4
+
T cellular crossreactivity.
279 Regarding the antibody response, between 47% and 100% of COVID-19 patients seroconvert
280 within fifteen days after the onset of symptoms [61, 62, 62]. Recovery after mild or moderate
281 forms is associated with neutralizing antibodies, whereas patients with severe forms of the
282 disease present early and high levels of non-neutralizing and possibly deleterious antibodies
283 [61, 63-65].
284 Thrombotic phenomena and lung vessels obstructive thrombo-inflammatory syndrome
285 play a role in severe and critical COVID-19 presentations. Similar to SARS-CoV-1, an
286 autoimmune signature including antiphospholipid antibodies, e.g., anticardiolipin IgA, anti-β2
287 glycoprotein 1 IgA, IgG and IgM, is present in SARS-CoV-2 infected patients [65]. It is
288 assumed that autoimmune markers are associated with the occurrence of thromboembolic
289 phenomena, but direct evidence is lacking. Lupus anticoagulant has also been reported as a
290 possibly frequent finding in up to 25 of 56 patients (45%), but again, the causative or
291 predictive relation to thrombotic events during COVID has not been assessed [50].
292 To summarize, our current understanding of immune responses to SARS-CoV-2 is that of an
293 early choice between two distinct pathways. An efficient monocyte-macrophage, CD4
+
and
294 CD8
+
T cellular response accompanied by a controlled inflammatory response enables virus
295 control and swift recovery. Conversely, a status of SARS-CoV-2-induced immune
296 dysregulation associated with low levels of type I interferons, an IL-6-driven inflammatory
297 status with immunosuppressive and inflammatory monocytes and macrophages, a defective
298 antigen presentation, an extensive organ immunopathogenesis and a prominent anti-SARS
299 CoV-2 and autoimmune antibody response is associated with severe forms of the disease.
300 Investigation of the SARS-CoV-2 immune response has already given insights into multiple
301 immune modulating therapies, from repurposed molecules (antimalarials, chlorpromazine,
302 antibiotics) to antivirals to monoclonal antibodies directed at cytokines (anti-IL-6, anti-IL-1)
303 or innate pathways (C5/C5a). Whether the crossreactivity of CD4+ T-cell lymphocytes
304 epitopes and antibodies with seasonal HCoVs can protect against SARS-CoV-2 and SARS
305 CoV needs to be further explored.
306 3.4 Immunopathogenic phase: the cytokine storm
307 On approximately the tenth day of infection, COVID-19-associated pneumonia may
308 evolve toward acute respiratory failure due to ARDS requiring ICU admission and high-flow
309 oxygen or mechanical ventilation, with a severe prognosis [2]. The underlying mechanisms of
310 these complications are immunological rather than due to the virus itself, which in most cases,
311 is no longer detectable at this stage.
312 Persistence of viral RNA detection in respiratory specimens has been reported even
313 three weeks after disease onset [66] and in 10-20% of patients at day 10 [3]. However, disease
314 in some patients became aggravated while the virus was no longer detectable using
315 conventional detection methods. It cannot be ruled out that, although not detected, the virus
316 was still present in cellular reservoirs not accessible to detection and can continue to activate a
317 pro-inflammatory response. However, it is likely that the virus is at the origin of the triggering
318 of a pro-inflammatory reaction that then self-amplifies.
319 Elevated circulating inflammatory cytokine concentrations have been reported in
320 patients, notably interleukin (IL)-1β, IL-6, MCP-1, IP-10, MIP-1α, IL-2, IL-10, revealing a
321 so-called “cytokine storm”, described in other inflammatory diseases such as macrophage
322 activation syndrome (MAS) or systemic inflammatory response syndrome [67, 68].
323 Strikingly, and consistently with immunopathophysiology, IL-6 levels are a near perfect
324 predictor of subsequent acute respiratory failure [69].
325 In vitro experiments and animal models of other SARS-CoV infections have shown that the
326 virus induces alveolar epithelial cell necrosis, the release of viral particles and of cytoplasmic
327 proinflammatory danger-associated molecular patterns (DAMPS), including ATP, nucleic
328 acids, or IL-1α [70]. In the meantime, SARS-CoV inhibits or delays IFNα/β production by
329 alveolar epithelial cells, which normally constitute the initial anti-viral defense [71].
330 Alternatively, SARS-CoV induces a robust but delayed IFNα/β response by the plasmacytoid
331 dendritic cells and macrophages, which, together with DAMPS, trigger chemokine production
332 and the recruitment of inflammatory monocytes/macrophages into the lungs [71]. In
333 inflammatory monocytes/macrophages, the virus induces NF-κB activation and the
334 transcription of several pro-inflammatory cytokines, notably IL-1β and IL-18 precursors, and
335 the constituents of the NLRP3 (for NOD-like receptor family, pyrin domain containing 3)
336 inflammasome [72, 73]. NLRP3 assembles in the cytoplasm after a second danger cell signal,
337 consisting in K+ efflux, ATP, lysosome degradation or production of mitochondrial reactive
338 oxygen species (ROS) [74]. Once assembled, NLRP3 activates caspase 1 and the processing
339 of biologically inactive IL-1β and IL-18 precursors into biologically active cytokines [74].
340 Importantly, SARS-CoV behaves as a membrane K+ channel and thus directly activates
341 NLRP3 assembly [73, 75]. In SARS-CoV animal models, the excessive NLRP3 activation
342 and IL-1β or IL-18 production induce a cytokine storm, ARDS and death, whereas blocking
343 NLRP3, the IL-1 receptor type 1 or IL-18 are protective [72]. IL-1β is a major pro
344 inflammatory cytokine known to induce fever via prostaglandin E2 secretion, neutrophilia via
345 GCSF production, liver acute-phase protein synthesis and a Th-17 immune response through
346 IL-6 secretion [76]. IL-18 is known to induce IFNγ production by Th-1 lymphocytes and NK
347 cells [77]. By binding to its receptor on inflammatory monocytes/macrophages, IL-1β induces
348 a harmful cytokine loop consisting of excess NLRP3, IL-1β/IL-6, IL-18/IFNγ synthesis,
349 initiating the cytokine storm [57]. NLRP3 also induces pyroptosis, a programmed cell death
350 process leading to the uncontrolled release of IL-1β, IL-18 and SARS-CoV2 particles [74].
351 At this stage of COVID-19, patients usually present with high-grade fever (>38.5°C),
352 moderate increased neutrophils, and elevated CRP concentrations (>150 mg/L), which are the
353 hallmarks of an IL-1/IL-6 signature [78]. Later, hyperferritinemia, diffuse coagulopathy and
354 cytopenia may appear, constituting the hallmarks of an IL-18/IFNγ signature [78]. Thus, the
355 absence of or delayed primary immune defenses against SARS-CoV2 encourages the
356 persistence of the virus, which in some individuals stimulates an uncontrolled self-stimulating
357 inflammatory loop and a cytokine storm that are initially located in the lungs but may diffuse
358 systemically and induce multi-organ failure.
359 3.5 Coagulopathy and thrombosis
360 Evidence of abnormal coagulation parameters associated with COVID-19 appeared in
361 early reports from China [1]. The most common hemostatic abnormalities observed are mild
362 thrombocytopenia and increased fibrinogen and D-dimers. Elevated D-dimers upon admission
363 are associated with increased mortality [79, 80].
364 Although the SARS-CoV-2 virus does not appear to have intrinsic procoagulant effects, the
365 significant increase in cytokines, also known as a cytokine storm, induces the activation of
366 coagulation and thrombin generation leading a state of major hypercoagulability. Initial
367 pulmonary findings during autopsy showed the presence of diffuse alveolar lesions with
368 infiltration of macrophages and CD4
+
T lymphocytes around thrombosed small vessels with
369 significant hemorrhages. The formation of microthrombi in the pulmonary microvessels
370 appears to be involved in the pathogenesis of the respiratory picture observed in patients with
371 SARS-CoV-2 infection [9]. This form of thrombosis involving the immune cells is referred to
372 as immunothrombosis [81]
. In addition, hypoxia induced by respiratory impairment may
373 cause thrombosis not only by increasing blood viscosity but also by increasing hypoxia
374 inducible transcription factors [82]. Moreover, the tropism of the virus for ACE2 receptors
375 could induce the activation of endothelial cells, disruption of their natural antithrombotic
376 properties, and apoptosis. Endothelialopathy may also contribute to the pathophysiology of
377 microcirculatory thrombosis [83]. Based on the literature currently available, the
378 coagulopathy and vasculopathy mechanisms are uncertain. Furthermore, patients with severe
379 COVID-19 present a hypercoagulability that predisposes them to thrombotic events. This
380 condition explains the presence of high levels of D-dimer. Despite coagulopathy, bleeding
381 manifestations have not been described. In later stages of COVID-19, sepsis-induced
382 coagulopathy and disseminated intravascular coagulation have been reported [83].
383 Many reports describe a high risk of venous thromboembolism. Notably, a high frequency of
384 pulmonary embolism (20 from 27%) was reported in ICU patients while patients were
385 receiving a standard dose of venous thromboembolism (VTE) prophylaxis [84, 85]. However,
386 a discrepancy between pulmonary embolism and deep venous thrombosis has been observed.
387 In view of these data, the use of anticoagulants could reduce the vicious circle of
388 inflammation-coagulation observed in patients with a severe form of the infection [86].
389 Moreover, a protective effect of heparin in patients with most severe COVID-19 infections
390 and increased D-dimers has recently been reported [87].
391 4. Therapeutic options (Tables 1 & 2)
392 Therapeutic options include antiviral and nonantiviral molecules. Potential antiviral drugs
393 according to in vitro results and chemical structure are detailed in Table 1. Three chemical
394 classes are particularly represented: 4-aminoquinolines, phenothiazines, which are chemically
395 related to methylene blue, and ribonucleic analogues. Zinc is of particular interest because it
396 is a nontoxic micronutrient with direct antiviral activity on RNA-dependent RNA polymerase
397 of Nidovirales with demonstrated activity in vitro and in a clinical study with the synergistic
398 association combining hydroxychloroquine (HCQ) and azithromycin (AZ). Interferon has
399 been administered in several Chinese studies, but its efficacy remains unclear. Nonantiviral
400 molecules are mainly represented by immunomodulators with corticosteroids, anti-interleukin
401 6, and hydroxychloroquine since the latter molecule has both antiviral and anti-inflammatory
402 effects. Increasing evidence stresses the critical importance of early anticoagulants in patients
403 with coagulopathy (positive D-dimers), and this is further supported by the fact that lung
404 embolism may be the direct cause of death in up to 30% of cases at autopsy [9].
405 4.1 Drugs active against SARS-CoV-2
406 Testing for molecules that are potentially active against SARS-CoV2 has been based on
407 different approaches. These included the selective testing of molecules previously shown to
408 be active against SARS-CoV and/or MERS-CoV or suspected to have a broad enough range
409 of activity to merit testing on SARS-CoV-2; high throughput testing of drug libraries; and in
410 silico prediction followed by confirmation of drug activity in vitro. Based on previous studies
411 on SARS-CoV, chloroquine (CQ) and hydroxychloroquine were among the earliest tested
412 molecules against SARS-CoV-2 [88-91]. Both compounds were shown to be efficient but
413 HCQ exhibited a less toxic profile [90]. Of 20 drugs previously demonstrated to have in vitro
414 antiviral activity against SARS-CoV and MERS-CoV, several were found to be effective in
415 vitro on SARS-CoV-2, including antitumoral drugs, which are likely not to be easily
416 applicable to treatment in humans, and antimalarial drugs such as amodiaquine and, again,
417 CQ and HCQ [91]. Of the tested drugs that are suspected to display a broad enough range of
418 activity to be active on SARS-CoV-2, there were two antimicrobial agents, ivermectin and
419 teicoplanin [92, 93] and the antiviral remdesivir [89] (Table 1). In another study using a more
420 systematic approach, testing of anti-SARS-CoV-2 activity in 1,520 approved drugs [94],
421 identified 90 molecules with potential efficacy. Of these, opipramol, quinidine and
422 omeprazole showed significant activity, and this was again the case for CQ and HCQ. From
423 the antibacterial agents, these authors also identified several fluoroquinolones and the
424 macrolide azithromycin (AZ). Interestingly, in a unique study associating AZ with HCQ (both
425 at 5 µM), a relative viral inhibition of 99% was observed, suggesting a synergistic effect [95].
426 In another study, the authors cloned, tagged and expressed 26 of 29 SARS-CoV-2 proteins
427 and studied their interactions with human proteins as inhibitors of these interactions [96]. Of
428 the 69 suggested compounds, 29 FDA-approved drugs were identified as being able to inhibit
429 SARS-CoV-2 replication in vitro, including HCQ. In conclusion, many drugs have been
430 identified as efficient, including many FDA-approved and well known drugs that have been in
431 use for decades. However, it would be illusory to imagine using psychotropic, antitumoral or
432 anabolic steroids that were identified in high throughput screenings. Therefore, drugs such as
433 CQ, HCQ, antibiotics or antihistamines used for allergies, and proton pumps inhibitors used
434 to treat gastroduodenal ulcers, alone or in association with other treatments are likely to
435 represent the most promising drugs to be tested in clinical trials against COVID-19 (Table 1).
436 Finally, another approach uses in silico structure-based virtual drug screening. This consists
437 of identifying candidate drugs potentially active on SARS-CoV-2, explaining the activity of
438 drugs or trying to explain observed activities [97, 98]. This approach does not require, as is
439 the case of in vitro testing of antiviral activity, a cell culture platform, viral strains, or a P3-
440 security level laboratory. It only predicts interactions and the inhibition of viral replication
441 using bioinformatic approaches based on structures, biochemical interactions, structural and
442 molecular modeling analyses, in silico docking models, or protein-protein interaction
443 networks. For instance, 69 compounds including HCQ were identified as targeting 66
444 druggable human proteins or host factors [96]. Compounds were also described as interacting
445 with the viral S-protein and angiotensin-converting enzyme 2 (ACE2)-host cell receptor [99]
446 while three viral polymerase inhibitors, zidovudine, tenofovir and alovudine, were suspected
447 to inhibit SARS-CoV-2 RNA polymerase [100]. In addition, structural and molecular
448 modeling analyses made it possible to propose a mechanism of action of CQ and HCQ on
449 SARS-CoV-2, by inhibiting the binding of the viral S protein to gangliosides, which are
450 present on the host cell surface and are linked with sialic acids that are used by the virus for
451 its entry, in addition to ACE2 [101].
452 4.2 Antiviral properties of chloroquine
453 CQ is a synthetic 4-aminoquinoline related to quinine. As far back as the mid-1960s, it
454 was demonstrated that CQ inhibited the mouse hepatitis virus 3 and the encephalomyocarditis
455 virus. Since these pioneering results, a large number of in vitro studies have confirmed that
456 CQ and HCQ appear to be large spectrum bioactive agents, which possess antiviral activities
457 against numerous viruses including rabies virus, poliovirus, human immunodeficiency virus,
458 hepatitis viruses, influenza viruses, arboviruses and Ebola virus among others [102].
459 However, the demonstration of in vitro activity cannot anticipate the in vivo efficacy of CQ.
460 Although there was evidence of an antiviral effect of CQ/HCQ in humans infected by
461 hepatitis C virus, more disappointing results were reported on other infectious diseases, in
462 particular in the treatment of Chikungunya virus, where CQ seemed to worsen symptoms
463 [103]. With regard to human coronaviruses, CQ was reported to inhibit the in vitro replication
464 of HCoV-229E, SARS-CoV, HCoV-O43 coronavirus, and MERS-CoV coronavirus [104].
465 Recently, it was reported that CQ/HCQ actually inhibited the in vitro replication of SARS
466 CoV-2 [89]. With HCoV-O43 it was reported that lethal infections in newborn mice could be
467 prevented by administering CQ through the mother’s milk [105]. However, CQ apparently
468 failed to cure MERS-CoV in vivo [106].
469 CQ is known for its multiple in vitro mechanisms of action on viruses. Coronavirus cell
470 entry occurs mainly through the endolysosomal pathway. A recent study reported the antiviral
471 activities of CQ/HCQ against SARS-CoV-2 in Vero E6 cells treated for one hour with the
472 drugs before exposure to the virus [90]. The virus yield in the cell supernatant was quantified
473 by qRT-PCR, while vesicle-containing virion was investigated by confocal microscopy. The
474 authors reported that CQ/HCQ significantly inhibited viral entry. Indeed, CQ/HCQ may
475 possibly inhibit at least five steps of SARS-CoV-2 replication. Recently, the results of a
476 molecular modeling approach suggesting that CQ binds to sialic acids and gangliosides with
477 high affinity were reported [101]. They hypothesized that both CQ and HCQ inhibit the
478 attachment of the amino acid region 111-158 of the viral spike of SARS-CoV-2 to
479 gangliosides. A second mechanism of action of CQ/HCQ on the same step requires the drug
480 to enter target cells to modulate the activity of cellular enzymes. CQ is likely to inhibit the
481 biosynthesis of sialic acid found at the extremity of sugar chains of glycoproteins. The potent
482 effects of CQ observed in vitro in cultures of Vero cells exposed to SARS-CoV was
483 considered attributable to a deficit in the glycosylation of cell surface receptor ACE2, which
484 is the first target of the virus [107]. The second step of the SARS-CoV-2 replicative cycle that
485 could possibly be inhibited by CQ is the pH-dependent viral endocytosis. This was previously
486 demonstrated for several viruses. The protonated form of CQ increased the pH of endosomal
487 compartments, thereby blocking the release of the infectious nucleic acid and enzymes
488 necessary for its replication [108, 109]. The third step of the SARS-CoV-2 that could be a
489 target for CQ/HCQ is transcription. Indeed, CQ could modulate the activity of the SARS
490 CoV-2 RNA-dependent RNA-polymerase through its function as an ionophore [110] favoring
491 the intracellular transport of the mineral zinc, which inhibits the activity of the polymerase
492 [111, 112]. The fourth step of the SARS-CoV-2 replicative cycle that is likely to be impaired
493 by CQ/HCQ deals with the posttranslational modifications of viral proteins within the
494 endoplasmic reticulum and trans-Golgi network vesicles, possibly by impairing the
495 maturation of its M protein [113]. A fifth level of CQ action could be its effect on cell
496 signaling, in particular through MAPK [114].
497 Since CQ/HCQ have well-characterized immunomodulating activities, in particular anti
498 inflammatory properties [115], the use of these drugs in the treatment of COVID-19 was also
499 suggested to possibly protect patients from the cytokine storm that marks the most severe
500 forms of COVID-19 disease.
501 4.3 Clinical use of chloroquine derivatives in COVID-19 patients (Table 2)
502 An early observational prospective controlled open label trial was performed in France
503 at the very beginning of the French epidemic and reported a dramatic beneficial effect on viral
504 shedding [116], although no clinical outcomes were investigated. In this context, we recently
505 sought to clarify its clinical efficacy through a meta-analysis of comparative studies
506 conducted on COVID-19 patients [117]. The first findings were that no meta-analysis could
507 be performed due to major discrepancies in the direction of effect. We therefore investigated
508 which moderator variables could explain such significant heterogeneity while in vitro studies
509 consistently evidenced an activity of the drug on SARS-CoV-2. Several parameters were
510 tested, including CQ versus HCQ, dosages, duration and timing of treatment, severity of the
511 disease (mild versus severe), combination with an antibiotic (notably AZ), design of studies
512 (prospective or retrospective, multicentric or monocentric, randomized controlled trials,
513 clinical studies or big data analyses conducted on medical records, etc.), comparable groups at
514 baseline, diagnostic approach (PCR, CT scan, clinical), and combination with other antivirals
515 (notably oseltamivir, lopinavir/ritonavir, ribavirin, umifenovir and alpha-interferon).
516 Strikingly, we observed that the main parameter determining the direction of effect was a
517 study design of big data versus clinical studies, and big data studies were associated with
518 country (USA), a conflict of interest and the absence of a detailed treatment protocol. When
519 analyzing big data separately (data were extracted from electronic record files by analysts
520 who did not treat COVID-19-infected patients) and clinical studies (performed by physicians
521 who treated COVID-19-infected patients), heterogeneity was controlled, and the directions of
522 the effects became consistent in each group. In big data studies, CQ derivatives were
523 associated with either a null [118-121] or a deleterious effect [122]. Strikingly, the latter big
524 data study reporting a deleterious effect in 96,000 electronic medical records was
525 subsequently retracted [123], confirming that clinical studies could be more reliable than big
526 data studies.
527 In clinical studies, a favorable effect was observed for the duration of fever, duration of
528 cough, clinical cure, death and/or ICU transfer, and viral shedding [117]. In clinical studies,
529 three [124-126] out of four [124-127] randomized controlled trials reported a significant
530 beneficial effect. Overall, CQ derivatives were beneficial and improved survival in COVID
531 19 infection. However, a standardized therapeutic protocol is required with an adequate
532 dosage (between 400 and 1,000 mg/d). Studies with higher dosages were associated with
533 significant toxicity [128]. In our experience, a dosage of 600 mg/day for 10 days is adequate.
534 Indeed, we previously proposed this dosage in Q fever endocarditis for at least 18 months and
535 in Whipple’s disease for 12 months. This has been subsequently recommended by the
536 American CDC [129]. We recently updated this meta-analysis [130] confirming a significant
537 beneficial effect on mortality and viral shedding including two large observational studies
538 from the USA [131, 132] and two from China [133]. Since then, two studies [134, 135],
539 including the RECOVERY trial [134], have been published with a diagnosis not confirmed by
540 PCR, which does not allow for a conclusion to be drawn. In the RECOVERY trial, cases
541 could be “clinically suspected” without laboratory confirmation (approximately 10% negative
542 tests and 10% without any testing), and 2,400 mg of hydroxychloroquine was administered
543 during the first 24 hours, which corresponds to a toxic dose. Here, we reported a last update of
544 our meta-analysis after the inclusion of three very recent studies from Brazil (RCT) [136],
545 Spain [137] and Taiwan [138] and focused on mortality (Figure 5) and persistent viral
546 shedding (Figure 6). Notably, Cavalcanti et al. reported a better effect with the HCQ-AZ
547 combination than with HCQ alone, demonstrating the importance of the synergy
548 demonstrated in vitro [136]. Despite substantial heterogeneity, a significant beneficial effect
549 could be confirmed for both mortality and viral shedding.
550 However, debate is currently rife in the scientific community and among government
551 decision makers as to the risk-benefit of using HCQ in the treatment of COVID-19 patients.
552 Recently, the WHO decided to ban the use of HCQ for COVID-19 patients, although various
553 clinical trials suggest that the benefits of using this molecule in combination with
554 azithromycin outweigh any harmful effects [126].
555 4.4 Azithromycin and COVID-19
556 AZ is a well-known and safe macrolide antibiotic with immunomodulatory properties. It
557 has a long half-life and a large volume of distribution [139]. It has excellent tissue penetration
558 in the lung and is widely prescribed for the treatment of respiratory infections. Its mechanism
559 of immunomodulation includes decreased production of pro-inflammatory cytokines (IL-6,
560 IL-8 and TNFα) and inhibition of neutrophil activation [140]. Although AZ has not been
561 labeled for the treatment of antiviral infections, it has been studied in vitro and in clinical
562 trials for activity against several viruses [139]. Numerous investigations have reported the in
563 vitro antiviral activity of AZ against viral pathogens, including SARS-Cov-2, at
564 concentrations that are physiologically achievable with the usual doses used for the treatment
565 of bacterial respiratory infections [139]. The precise mechanism of antiviral activity is not
566 known [139]. However, the intracellular accumulation of AZ, a weak base in endosomal
567 vesicles and lysosomes intracellularly may result in an increase in endosomal and/or
568 lysosomal pH and limit viral replication, through the lack of an optimal acidic environment in
569 the intracellular milieu. Interestingly, HCQ is also a weak base, and this could explain how
570 the two drugs work together to inhibit viral replication. Recently, our group demonstrated that
571 the combination of HCQ and AZ had a synergistic effect in vitro on SARS-CoV-2 at
572 concentrations compatible with that obtained in the human lung [95].
573 Many clinical studies on the efficacy of AZ alone or in combination with other drugs
574 against various viral infections have been observational, single
‐arm, non-randomized studies
575 or retrospective evaluations and have mainly focused on viral load as an end point.
576 Collectively, however, they present preliminary evidence that the inclusion of AZ in various
577 treatment regimens can influence the course of viral infection and clinical outcomes [139,
578 141, 142].
579 4.5 Clinical use of zinc in COVID-19 patients
580 Zinc has both antiviral and immune properties [143]
. Known to inhibit the
581 multiplication of several viruses, in vitro models have demonstrated that low zinc
582 concentrations combined with ionophores block the elongation of the SARS-CoV-1 RNA
583 dependent RNA polymerase. [112]. Interestingly HCQ and CQ are ionophores that enhance
584 zinc uptake, thereby increasing its concentration into the lysosomes [110]. The potential
585 synergistic effect of a combination with CQ/HCQ on SARS-CoV-2 replication remains,
586 however, to be demonstrated. Zinc is also involved in antiviral immunity through several
587 mechanisms including the modulation of interferon response [143]. As zinc supplementation
588 over a short period is not harmful to health, it has been proposed in combination with HCQ in
589 COVID-19 patients, with promising results [144].
590 4.6 Clinical uses of remdesivir on COVID-19 patients
591 To date, eight studies (1,773 patients) have reported the use of remdesivir in COVID-19
592 patients. Case reports showed that remdesivir has no clinically relevant efficacy. A
593 compassionate uncontrolled study supported and funded by Gilead reported that 68% of
594 treated patients saw clinical improvements, but there was considerable missing data, and the
595 outcomes of 9/61 patients were still under evaluation at the time of publication and were not
596 reported [145]. One randomized controlled study (RCT) included 236 patients (158/78) from
597 10 hospitals in Wuhan. The mean age, sex ratio, delay from onset to enrolment, comorbidities,
598 enrolment criteria (O2<95%), and radiologically confirmed pneumonia, were comparable in
599 the two arms. The primary clinical endpoint was the time to clinical improvement within 28
600 days after randomization, and 100% of patients enrolled were evaluated in the intention-to
601 treat analysis. No significant differences were noted between the two groups. Serious adverse
602 events or events leading to administration of the drug being stopped were reported in 18% and
603 12% of patients, respectively, in the remdesivir group compared to 6% and 5%, respectively,
604 in the placebo group, demonstrating the poor safety profile of the drug [146]. Another RCT
605 conducted on 1,063 patients (541/522) argued for a significant benefit of remdesivir on time
606 to-recovery and mortality in the intention-to-treat analysis; however, at the time of
607 publication, only one-third of enrolled patients had received complete treatment and had been
608 analyzed. Such a loss to follow-up poses serious threats to the validity of this study [147]. The
609 last released paper compares five days to 10 days of treatment with remdesivir with no
610 significant difference in mortality rates or improvement in clinical status. Serious adverse
611 events were reported in 27.7% of treated patients, including 4.7% acute kidney injuries. In
612 7.3% of patients, adverse events led to the treatment being stopped [148]. In addition,
613 remdesivir is currently only available intravenously, making early treatment at home
614 impossible and exposing patients to the risk of intravenous catheter complications
615 (lymphangitis, thrombophlebitis, endocarditis).
616 4.7. Use of nonantiviral therapy in COVID-19 patients
617 During the immunopathogenic phase of COVID-19, serum IL-6 concentrations have
618 been found to be consistently elevated (90 to 160 pg/ml) and correlated with severity [1, 67].
619 IL-6 is produced by various cells, mainly in response to IL-1, and is the main inducer of acute
620 phase-protein synthesis by the liver, notably CRP. It plays a role in B lymphocyte
621 differentiation and encourages mononuclear cell inflammation and Th-17 response [149, 150].
622 IL-6 acts through binding to a ligand receptor, IL-6R, then complexes with signaling receptor
623 gp130. Although gp130 is widely expressed, IL-6R expression is limited to leukocytes and
624 hepatocytes but can be shed from these cells in a soluble form, which binds to IL-6 and then
625 to gp130, rendering IL-6R-negative cells sensitive to IL-6, a mechanism named trans
626 signaling [151]. Through this mechanism, IL-6 has been involved in various inflammatory
627 diseases, such as rheumatoid arthritis (RA) and lung fibrosis [149] [152].
628 Tocilizumab is a humanized anti-IL-6R monoclonal antibody (mAb) that is able to bind
629 both to the membrane and soluble IL-6R and to inhibit IL-6 functions. It is currently used in
630 the treatment of various chronic diseases, notably rheumatoid arthritis, giant cell arteritis, and
631 systemic juvenile idiopathic arthritis (sJIA), and has a good safety profile despite its long half
632 life [149]. sJIA is often complicated by macrophage activation syndrome characterized by a
633 cytokine storm, and tocilizumab has been shown to be effective in the treatment of this
634 disease [153]. Moreover, cancer treatments using CAR-T cells may be complicated by a
635 cytokine release syndrome, which is efficiently treated by tocilizumab [154]. Tocilizumab
636 was, therefore, used to treat 21 severe or critical COVID-19 patients in China and was
637 retrospectively reported to be effective [155]. The patients received a single 400 mg infusion.
638 Their body temperature returned to normal within 24 hours, oxygen saturation improved
639 rapidly in 75% of cases, and CRP returned to normal in 80% of the patients after five days.
640 This observation was confirmed in Italy [156]. Tolerance to treatment appeared to be good in
641 all studies, but acute hypertriglyceridemia and pancreatitis were recently reported in two
642 patients [157].
643 Together, these results are promising and have justified several ongoing trials in Italy,
644 France and the US (using sarilumab, another anti-IL-6R mAb) [158]. However, other
645 therapeutic strategies may be effective in order to treat the cytokine storm associated with
646 COVID-19, such as IL-1 receptor antagonists (anakinra), which act upstream of IL-6 and have
647 a short half-life, anti-IFNg (emapalumab), NLRP3 pharmacological inhibitors (dapansutrile)
648 or JAK kinase inhibitors, which may block IL-6 and IFNs signaling downstream of their
649 receptors. Recent retrospective case series or small cohorts conducted in Italy [159] and in
650 France [160] have shown that anakinra may be an effective treatment in arresting
651 inflammatory respiratory deterioration.
652 In addition, preliminary results from the RECOVERY trial suggest that dexamethasone
653 either by mouth or by intravenous injection improves prognosis in patients on oxygen and
654 those in intensive care [161]. However, final detailed results are not published. Moreover, no
655 benefit was observed among those patients who did not require respiratory support. In this
656 context, hydroxychloroquine, with its strong anti-IL6 activity [162], remains an attractive
657 option as an early nonantiviral therapy.
658 4.8 Prophylactic use of hydroxychloroquine against COVID-19
659 4.8.1 Pre-exposure prophylaxis
660 In a case-controlled study in India on symptomatic healthcare workers (HCWs)
661 including 378 positive cases (symptomatic HCWs with SARS-CoV-2 positive PCR, including
662 172 with HCQ prophylaxis) and 373 controls (symptomatic HCWs with SARS-CoV-2
663 negative PCR including 193 with HCQ prophylaxis), the administration of at least four HCQ
664 doses was associated with a significant, fifty percent lower risk of COVID-19 (adjusted odds
665 ratio [aOR] 0.44, 95% confidence interval 0.22–0.88) [163]. A dose-dependent relationship
666 was observed with an adjusted odds ratio decreasing from 0.44 (0.22–0.88, p = .02) for four–
667 five doses and 0.04 (0.01–0.16, p < .001) for six doses and more. One to three doses had no
668 effect while six or more prophylactic doses had a remarkably high (>80%) protective effect.
669 Of the 365 HCWs who reported taking HCQ, nausea, headache and diarrhea were reported in
670 approximately 5%; only individual case reported palpitations.
671 4.8.2 Postexposure prophylaxis
672 In South Korea, a COVID-19 exposure event occurred in a long-term care hospital, and
673 HCQ postexposure prophylaxis was provided to 211 individuals [164]. None of them
674 developed the disease, and none had a positive PCR during follow-up. However, since no
675 subjects who did not receive prophylaxis became positive, effectiveness could not be truly
676 evaluated. In the USA, a double-blinded randomized controlled trial was performed with 414
677 subjects receiving HCQ and 407 receiving folate as a placebo within four days of exposure
678 [165]. No arrhythmias or deaths occurred. Furthermore, 49 (11.8%) developed a disease
679 compatible with COVID-19 in the HCQ group compared to 58 (14.3%) in the control group
680 (p = 0.35). The authors thus concluded that HCQ did not prevent the disease. However, most
681 of the people included did not have a definitive diagnosis of COVID-19 by PCR. In addition,
682 the delay in postexposure chemoprophylaxis strongly impacts the efficacy, with a fifty percent
683 decrease in the risk of infection when HCQ was administered on day 1 and decreasing
684 efficacy over time. For influenza, postexposure chemoprophylaxis has only been
685 recommended when antivirals can be initiated within 48 hours of exposure [166]. This
686 highlights the need to initiate chemoprophylaxis rapidly after exposure.
687 4.9. Global care of COVID-19 patients
688 4.9.1 Outpatient care
689 In the context of a brutal and deadly pandemic (global mortality rate of approximately
690 6% and up to 18% in France), no randomized trial investigating early versus delayed
691 treatment has been conducted. Barbosa Esper et al. reported that early telemedicine treatment
692 with HCQ and AZ was associated with a significant decrease in hospitalization rates [167].
693 Zev Zelenko, a general practitioner in the suburbs of New York, USA, reported on the early
694 treatment of 405 patients with HCQ 200 mg 2/d, AZ 500 mg 1/d and zinc sulphate 220 mg
695 1/d for five days at a total cost of $20 [168]. All patients with dyspnea or with risk factors
696 (regardless of clinical status) were treated. As of 26 April, he had reported one death in a
697 patient with leukemia (1/405 (0.2%)). The only adverse events reported were nausea and
698 diarrhea in 10% of cases but no cardiac toxicity. Guerin et al. reported on a French study on
699 outpatient treatment of 88 patients with HCQ-AZ, HCQ and controls [169]. Early HCQ or
700 HCQ-AZ treatment was associated with a significantly shorter time to clinical recovery. There
701 was no significant difference between HCQ alone and HCQ-AZ. No cardiac toxicity was
702 observed.
703 Finally, massive screening and early treatment have been implemented in our center
704 (Marseille, France), with more than three-quarters of the 1,061 reported cases managed in a
705 day care hospital for initial evaluation and treatment [141]. The therapeutic protocol included
706 at least an ECG, potassium measurement, D-dimers, low-dose CT scan, HCQ (200 mg three
707 times per day in the absence of ECG repolarization disorder and abnormal kalemia), and AZ
708 (500 mg on the first day then 250 mg per day for four additional days). Correction of
709 hypokalemia was needed in approximately 15% of our cohort and was associated with
710 severity [170]. This is critical since hypokalemia could increase the risk of cardiac
711 arrhythmia. After confirming in our cohort that zinc deficiency was associated with an
712 unfavorable prognosis, and in view of a comparative study in favor of zinc [144], 15 mg of
713 zinc three times daily was added. Any patient with positive D-dimers was evaluated for
714 pulmonary embolism, and treatment with low molecular weight heparin was systematically
715 discussed on an outpatient or inpatient basis. At least one patient was diagnosed with a
716 pulmonary embolism in our day care hospital after systematic D-dimer assessment without
717 any clinical signs. In the presence of a clinical (NEWS score ≥ 5) or biological sign of
718 severity, or if the treatment became difficult, patients were systematically hospitalized.
719 4.9.2 Inpatient care
720 Under this global management strategy, based on early massive nonselective screening,
721 day care hospitals, inpatient and early resuscitation care, no deaths under the age of 60 years
722 have been observed. In the 60+ age group, the mortality regardless of treatment was 5.0%,
723 which was similar to the mortality in the same age group reported in China (6.0%) but lower
724 than that reported in Italy (12.3%) [171]. On the other hand, among patients aged 60 years and
725 over who had at least three days of dual HCQ-AZ therapy in our center, the mortality was
726 3.1%, which was much lower than that reported in China and Italy for the same age group.
727 Based on the currently available literature, we would recommend measuring D-dimers,
728 prothrombin time, and platelet count in all patients with COVID-19 infection. The optimal
729 dosage of low molecular weight heparin is currently the subject of much discussion within the
730 medical community. Although the majority suggests prophylactic daily LMWH, a minority
731 considers intermediate or therapeutic doses [83]. In all cases, the optimal doses for each
732 patient need to take account specific patient thrombotic risk factors, such as active cancer
733 (treatment within the last six months), recent personal history of thromboembolic events
734 ( 30 kg/m 2 ) [172].
735 Regarding possible drug interactions, oral anticoagulants prescribed for long-term use should
736 be replaced by curative heparin therapy.
737 4.9.3 Intensive care
738 Approximately 15% of COVID-19 patients require admission to an ICU. The biggest
739 reason for admission is acute respiratory failure [173]. Gattinoni et al. recently described two
740 phenotypes of ICU patients: phenotype L and phenotype H. Phenotype L corresponds to a
741 clinical picture of ARDS with low elastance, which is unexpected in this syndrome.
742 Phenotype H for high elastance is in line with the common picture of ARDS, representing 20-
743 30% of COVID-19 patients [174]. A CT scan is a perfect tool to differentiate the two
744 phenotypes: the nonaerated lung is close to 0 in phenotype L patients, while posterior
745 condensations are found in phenotype H patients [174]. Most patients evolve from phenotype
746 L to phenotype H, which may be due to COVID-19 pneumonia evolution or to the adverse
747 effects of high-pressure mechanical ventilation. The hospital mortality rate of COVID-19
748 patients admitted to the ICU ranged from 26% [20] to 50% [175] and was even higher in
749 elderly patients [173]. However, in our center, the in-ICU mortality was approximately 18%
750 (unpublished data) in a context of a global strategy including early testing, early care and
751 early treatment [141].
752 While COVID-19 was initially documented as isolated acute respiratory failure, patients
753 admitted to the ICU develop multiple organ failure in approximately 50% of cases [173].
754 Among organ failures, an exacerbated activation of coagulation in COVID-19 patients
755 resulted in a large number of thromboses and pulmonary embolisms. The activation of
756 coagulation is a well-known host response in patients with septic shock. Specific treatments
757 were developed to counteract this response, but successive studies failed to demonstrate
758 efficiency. However, disruptions of coagulation strongly affect the outcomes of ICU patients
759 with COVID-19. In a cohort of 184 ICU patients, 27% of patients developed deep vein
760 thrombosis, and pulmonary embolism was found in 80% of cases [84]. Anticoagulation
761 treatment should be widely administered to reduce the risk of pulmonary embolism, and
762 protocols should include curative anticoagulation in most mechanically ventilated patients.
763 Monitoring coagulation factors, including D-dimers and fibrinogens, provides relevant
764 information to decide between prophylactic or curative anticoagulation. Those with an
765 increased body mass index are at a very high risk of thrombosis.
766 In ICU patients, an intense inflammatory response has been described in the late phase
767 of the disease. The production of pro-inflammatory cytokines has been reported to be
768 exacerbated in patients with negative SARS-CoV-2 PCR. This response profile encouraged a
769 few teams to use steroids, notably for patients developing ARDS [68]. In a cohort series, the
770 use of steroids in ARDS patients was associated with improved outcomes, within the
771 limitations of this study based on multivariate analysis [176]. However, the pro-inflammatory
772 profile of septic patients in the ICU was previously described, while recent trends revealed
773 profound immunosuppression. Recent data do not confirm the pro-inflammatory response in
774 COVID-19 patients in the ICU (personal data) as the profile is time dependent. Introducing
775 steroids to patients with viral infections and immunosuppressed patients can be high risk,
776 resulting in an increase in superinfections. Thus, it seems prudent to restrict the use of steroids
777 in those severe patients who usually have a severe lymphopenia.
778 4.10 Traditional Medicine
779 Traditional Chinese Medicine (TCM) therapy has played a role in treating epidemic
780 diseases in China’s long history [177]. In combination with Western medicine, TCM therapy
781 was widely used in China for the management of COVID-19, with up to 91.5% of patients
782 with a confirmed diagnosis having reportedly received TCM [178]. Several studies support
783 that the current practice of TCM has a favorable impact in the management of COVID-19 and
784 can shorten of the course of fever, course of the disease, and length of hospital stay, and
785 reduce the number of severe patients and death rate [177, 179]. Three patented TCM
786 medicines (Lianhua Qingwen capsules, Jinhua Qinggan granules, and Xuebijing injection)
787 and three TCM decoctions (Qingfei Paidu, Huashi Baidu, and Xuanfei Baidu) have mainly
788 been highlighted for the management of COVID-19 [177, 180]. Lianhua Qingwen is a
789 formulation prepared from the classic compounds of ancient China, commonly used in the
790 treatment of colds and flu. Jinhua Qinggan is a formulation developed for the treatment of
791 influenza A (H1N1) in 2009. The use of Jinhua Qinggan and Lianhua Qingwen is suggested
792 for medical observation of COVID-19 [181]; they could also be used for patients with mild
793 and moderate symptoms. Xuebijing injection was developed for the treatment of SARS in
794 China and is indicated for systemic inflammatory response syndrome induced by infections
795 [180]. Xuebijing is used for severe and critical COVID-19 infection [181]. Qingfei Paidu,
796 Huashi Baidu, and Xuanfei Baidu decoctions are three new prescriptions specifically designed
797 for COVID-19 [181]. Their use is suggested in the prevention and mild and moderate
798 infections of SARS-CoV-2 [181]. The action mechanisms of Chinese medicines are still
799 unclear. Most of them are founded on pharmacology-based predictions [177]. The
800 composition, potential active ingredients, predicted targets, signaling pathways and
801 mechanisms of these Chinese herbal medicines have been recently detailed by Huang et al. in
802 a review article [177]. The use of TCM at different stages of COVID-19 is among the
803 recommendations in China and South Korea [177, 182]. Ho et al. have summarized the
804 guidelines of the National Health Commission and the National Administration of Traditional
805 Chinese Medicine of the People’s Republic of China for the management of COVID-19,
806 which, in addition to the abovementioned formulas, includes compositions adapted according
807 to the patient’s condition [182]. The reported duration of COVID-19 treatment with herbal
808 medicines usually varied from 5 to 15 days [179]. Regarding potential side effects, no severe
809 discomfort or abnormal liver or kidney function has been identified [179]. Although basic and
810 clinical research is still needed to understand the mechanisms of action and lead to evidence
811 based medicine, it is worth paying attention to TCM. Indeed, it must be remembered that the
812 use of herbal medicine against malaria has allowed to the discovery of artemisinin, a plant
813 extract of Artemisia annua used as a standard treatment against Plasmodium falciparum
814 worldwide [183].
815 5. Conclusions
816 COVID-19 is a newly emerging disease that requires us to adjust our medical approach as we
817 receive new observational evidence about the virus and its behaviors. Indeed, the observation
818 of symptoms usually rarely associated with respiratory infections, such as anosmia or a lack
819 of perceived dyspnea with documented hypoxemia and severe radiological impairment on CT
820 scan, and viral thrombotic disease with up to 30% of critical cases developing pulmonary
821 embolism, led us to adapt our clinical approach toward SARS-CoV-2-associated pneumonia.
822 Early treatment of patients through the rapid implementation of specific PCR diagnostic tools
823 has been, in many cases, key to successfully controlling the disease. In most affected
824 countries, the epidemic appears to spontaneously resolve over time, but COVID-19 morbidity
825 and mortality indicators are, paradoxically, worse in the rich countries of Western Europe and
826 the US than in developing countries. The debate about the usefulness of chloroquine
827 derivative-based treatment, initially proposed by Chinese physicians, and that of new
828 antivirals such as the orphan drug remdesivir, has reached unprecedented levels of
829 aggressiveness. This culminated in the recent retraction of COVID-19 publications from the
830 world’s two most prestigious medical journals, The Lancet and The New England Journal of
831 Medicine. Thus, lessons from the COVID-19 pandemic go far beyond the disease itself and
832 question our responsiveness toward a disease with medical, societal, economic, and political
833 consequences that may lead to major editorial pitfalls challenging the credibility of the main
834 actors in the field of medical literature.
835 6. Expert opinion (Figure 7)
836 The COVID-19 epidemic is an unprecedented crisis, not in terms of mortality, but in terms of
837 emotion and the measures taken to fight it. It is not the worst epidemic we have experienced
838 in recent decades, but it has led to changes in the organization of care, including lockdown
839 measures, which are unprecedented. However, it is likely that in most countries, COVID
840 related mortality will go unnoticed in the annual mortality statistics. This is particularly
841 because it has mostly affected older people. The epidemic has also revealed the
842 disorganization of political decisions with totally different strategies from one country to
843 another and paradoxically higher mortality in the richest countries, especially those in
844 Western Europe and the US. No consensus has been reached around the implementation of
845 basic strategies including diagnostic tests, isolation of contagious patients, patient care and
846 potentially safe and effective therapies. Finally, unsupported decisions taken by governments
847 and the WHO have increased confusion. This has led to doubtful publications being retracted
848 from the best journals in the world and has given rise to a previously unknown state of
849 nervousness. When studying the disease seriously, a few things are worth noting. It does not
850 present itself as common flu but commonly presents with anosmia. The patient may be free of
851 fever, cough, and shortness of breath, despite significant hypoxia and CT lesions. This
852 warrants an expanded strategy of diagnostic testing, measuring oxygen saturation and
853 performing low-dose CT scans to detect specific lesions. Early therapeutic management in
854 terms of care (anticoagulants, oxygen therapy) and drugs that have proven or will prove to be
855 effective at this stage (HCQ and AZ and possibly remdesivir in the early stages) will prevent
856 progression to respiratory failure, resuscitation and death. A number of markers have been
857 associated with this pejorative evolution including a lymphopenia below 500 being be the
858 most predictive, decreased zinc levels, eosinopenia, increase in polymorphonuclear cells and
859 increase in D-dimers. Subjects with hypertension are more at risk as a result of hypertension
860 or antihypertensive drug use. The disease appears to unfold in three stages: a purely
861 virological stage, a stage associating the virus with the immune response, and a final stage in
862 the most severe forms, which appears essentially as an immune response without the virus.
863 This explains why antiviral treatment attempts in patients with very severe disease are usually
864 ineffective, and samples taken to look for the viral load at this stage are often negative. A
865 cytokine storm occurs against which no drug has proven to be effective. This is something
866 that is common in adult acute respiratory distress syndromes, where treating the cause is less
867 critical than properly managing it. Immunity as evidenced by antibodies against the virus
868 appear around the tenth day and is extremely marked in the most seriously affected subjects.
869 In our experience, deceased subjects had the highest antibody levels. Concerning the
870 epidemiology of the disease, it presents in the form of a bell curve, which is highly typical of
871 respiratory viral infections. It appears that part of the population is naturally immune, which
872 may be related to endemic circulating coronaviruses, which cause 10 to 20% of respiratory
873 infections, especially in children. Indeed, it is remarkable that the incidence of both the virus
874 and the disease in children has been low. Under these conditions, the use of a vaccination
875 before this point is likely to present more risks than benefits. The future of this disease is
876 unknown.
877 What lessons has this crisis taught us? First, in 2020, infectious disease diagnosis can no
878 longer be based solely on clinical criteria without microbiological testing. Indeed, at the peak
879 of the epidemic in our center, only 22% of symptomatic patients were PCR positive. Many
880 viruses have cocirculated, including endemic coronaviruses. Second, general
881 recommendations cannot be made until the disease is observed and known. It is only recently
882 that the four stages of the disease have been identified: 1) viral, 2) viral and dysimmune, 3)
883 dysimmune, and 4) lesional (ARDS) stages. Each stage corresponds to a specific treatment
884 (Figure 3). Then, when evaluating the efficacy of a treatment, the molecule should not be
885 considered in isolation. A treatment is not a molecule, it is a therapeutic protocol with one or
886 more molecules (synergistic effect of the HCQ+AZ+zinc combination [144]), indications,
887 contraindications, precautions for use, and a precise dosage and duration. For example, for
888 several decades we have been administering HCQ at 200 mg three times a day for 18 to 24
889 months in Coxiella burnetii endocarditis in association with doxycycline [184]. This dosage is
890 necessary to achieve therapeutic blood levels [184]. This protocol has been used according to
891 the recommendations of the American CDC for the same indication [129] and in the most
892 well-known professional sites (https://www.uptodate.com/contents/q-fever-endocarditis). In
893 rheumatic diseases such as lupus or rheumatoid arthritis, dosages of 400 to 600 mg/d are used
894 (www.uptodate.com). In COVID-19, extreme dosages of 2,400 mg/d were used in the
895 RECOVERY trial, i.e., four times the usual dose, while in the Discovery trial, low doses of
896 400 mg/d were used. Overall, standardized therapeutic protocols associated with a
897 comprehensive strategy including early massive screening would appear to be critical in
898 responding to the COVID-19 pandemic
Related:
COVID-19 and rheumatic autoimmune systemic diseases: report of a large Italian patients series
Hydroxychloroquine and azithromycin tolerance in haemodialysis patients during COVID-19 infection