Coronavirus disease 2019 may be driven by an overactivation of the renin-angiotensin-aldosterone system

SARS-CoV-2 enters the cell through the ACE2 receptor, which is considered one of the main inhibitors in the Renin-Angiotensin-Aldosterone System (RAAS).1 ,2 The virus has been shown to downregulate the ACE2 receptor, leading to a subsequent increase in the vasopressoragentangiotensinII.3 Evidently,criticalcoronavirusdisease2019(COVID-19)is thought to be due to a dysregulated immune response, causing a cytokine-release syndrome eventually leading to acute respiratory distress syndrome (ARDS).4 ,5 However, several reports on clinical laboratory features and case-descriptions of critically ill patients with COVID-19 show discrepancies compared to typical ARDS. Here, we show that infusing swines with angiotensin II induces a pathophysiological syndrome closely resembling that of patients with RT-PCR-positive COVID-19. By using multimodal clinical imaging of patients, comparing laboratory data and translational histological features, we show that it is highly likely that an increase in RAAS is one, if not the main, pathogenic feature in critical COVID-19. Furthermore, it is plausible that this large animal model can be used to screen for potential new treatments for patients with severe COVID-19 and that MRI lung perfusion can be used to evaluate the outcome of potential treatments targeting the pathophysiological syndrome. a cytokine-release syndrome.4 ARDS-pattern.5 the levels of pro-inammatory cytokines be lower than the levels that are expected during cytokine-release syndrome.7 a case series from Wuhan described that ANGII levels were markedly elevated and linearly associated with viral load and lung injury.8 found that infusion of ANGII in three sedated swine within hours a clinical syndrome similar to that observed in ICU patients in COVID-19, with histological changes in the lungs after only ve hours with severe thickening of the alveolar walls, possible hyaline membranes and clotting of vessels, as previously reported in humans in COVID-19.9–11 We present data in reverse-transcriptase–polymerase-chain-reaction (RT-PCR) conrmed COVID-19, showing high pulmonary artery (PA) pressures in one patient and extensive peripheral hypoperfusion of the lungs using magnetic resonance imaging (MRI) in another patient. We also present computed tomography pulmonary angiography (CTPA) results from 289 consecutive SARS-CoV-2 RT-PCR-positive patients and a total of 339 CTPA examinations; 17% of scans revealed pulmonary thrombosis/embolism and 60% of the scans had a wide PA diameter, suggestive of pulmonary hypertension, further supported by echocardiography ndings in a sub-cohort of 50 patients. We also relate our ndings to the results observed in swines: ANGII infusion leads to high PA pressures and a marked reduction in bleeding time. D-dimer increases were observed, as seen in COVID-19 patients. Paradoxically, brinogen increased, suggesting a possible link between ANGII infusion and brinogen synthesis. All three swines showed pulmonary thrombotic material without deep venous thrombosis, suggestive of local thrombotic events. We suggest that much of the clinical presentation in ICU patients with COVID-19 could be driven by a loss of the inhibition of the RAAS, causing supranormal concentrations of ANGII.


Introduction
SARS-CoV-1 and SARS-CoV-2 both enter human cells through cell surface angiotensin-converting enzyme 2 (ACE2) receptors.1 SARS-CoV-1 downregulates ACE2,3 with a subsequent increase in angiotensin II (ANGII) levels,6 which potentially creates a disruption akin to overactivating the Renin-Angiotensin-Aldosterone System (RAAS).2Current clinical management guidelines for the coronavirus disease 2019 (COVID-19) are mainly centred around the assumption that SARS-CoV-2 directly results in an acute lung parenchymal disease such as acute respiratory distress syndrome (ARDS).4,5 The possible underlying mechanisms are thought to be viral ARDS and/or high levels of cytokines leading to a cytokine-release syndrome.4However, the initial clinical presentation of COVID-19 in the intensive care units (ICU) at our hospital, as well as described by others, does not follow a typical ARDSpattern.5Further, the levels of pro-in ammatory cytokines can be lower than the levels that are expected during cytokine-release syndrome.7Moreover, a case series from Wuhan described that ANGII levels were markedly elevated and linearly associated with viral load and lung injury.8 We found that infusion of ANGII in three sedated swine within hours induced a clinical syndrome similar to that observed in ICU patients in COVID-19, with histological changes in the lungs after only ve hours with severe thickening of the alveolar walls, possible hyaline membranes and clotting of vessels, as previously reported in humans in COVID-19.9-11We present data in reverse-transcriptase-polymerasechain-reaction (RT-PCR) con rmed COVID-19, showing high pulmonary artery (PA) pressures in one patient and extensive peripheral hypoperfusion of the lungs using magnetic resonance imaging (MRI) in another patient.We also present computed tomography pulmonary angiography (CTPA) results from 289 consecutive SARS-CoV-2 RT-PCRpositive patients and a total of 339 CTPA examinations; 17% of scans revealed pulmonary thrombosis/embolism and 60% of the scans had a wide PA diameter, suggestive of pulmonary hypertension, further supported by echocardiography ndings in a sub-cohort of 50 patients.We also relate our ndings to the results observed in swines: ANGII infusion leads to high PA pressures and a marked reduction in bleeding time.D-dimer increases were observed, as seen in COVID-19 patients.Paradoxically, brinogen increased, suggesting a possible link between ANGII infusion and brinogen synthesis.All three swines showed pulmonary thrombotic material without deep venous thrombosis, suggestive of local thrombotic events.We suggest that much of the clinical presentation in ICU patients with COVID-19 could be driven by a loss of the inhibition of the RAAS, causing supranormal concentrations of ANGII.Based on these results, we believe that critical COVID-19 is primarily a vascular and coagulation disease that is driven by a viral disruption of the RAAS and that focus should, henceforth, be directed towards its understanding and therapeutic targeting.

Two clinical cases
We performed lung perfusion MRI in a SARS-CoV-2 RT-PCR-positive 61-year-old male on ICU day 22.Initial presentation to the ICU included a PaO2 of 7.6 kPa and PaCO2 of 3.6 kPa by arterial blood gas; serum measurements of IL-6 154 ng/l, TNF-α 10.2 ng/L, C-reactive protein 242 mg/L, ASAT 1.08 μkat/L and D-dimer 0.94 mg/L FEU.We show that the maximum contrast bolus concentration in the PA comes at 15 seconds after injection and that the maximum contrast bolus concentration reaches the aorta after 20 seconds (Fig. 1).Using these values, we calculated a time-to-peak (TTP) map of the lungs using a colour lookup table where blue and red corresponds to 14 and 21 seconds respectively.All red parts of the lung have a maximum contrast bolus concentration later than the aorta (sic).To further understand these paradoxical values, we measured regions of interest in the lungs without in ltrates, as identi ed by T2-weighted anatomical images.The most probable conclusion is that peripheral regions of the lungs do not receive a contrast bolus up to 10 seconds after the contrast has peaked in the aorta.In conclusion, using MRI perfusion, we demonstrate a lack of contrast bolus enhancement in the peripheral lung parenchyma.In the second patient, a SARS-CoV-2 RT-PCR-positive 58-year-old male, we placed a PA-catheter on ICU day 25 and recorded systolic and diastolic PA pressures of 51 ± 2.9 / 11 ± 3.2 mmHg (mean ± SD) and SvO2 of 62.4 ± 1.3 % (mean ± SD), both during the rst hour after placing the catheter.Upon admission to the ICU, the patient's clinical chemistry values were PaO2 of 8.0 kPa and PaCO2 of 5.1 kPa, IL-6 at 95 ng/L, TNF-α at 9.8 ng/L, C-reactive protein at 66 mg/L, ASAT 0.53 μkat/L, D-dimer of 0.26 mg/L FEU.

Computed tomography pulmonary angiography
To further con rm our ndings, we retrospectively analysed all consecutive CTPA examinations acquired at Karolinska University Hospital in Huddinge, Stockholm, Sweden, in patients with RT-PCR-con rmed COVID-19.In total, 339 examinations were performed in 289 adult patients with RT-PCR-con rmed COVID-19.Motion artifacts or other technical issues prohibited PA-diameter measurements in 47 examinations, as measured by the gold standard reader TG, resulting in 292 analysed scans.The inter-rater agreement was excellent (N=290, ICC=0.97,P<0.001).The mean PA-diameter was 29.2 ± 4.3 mm (mean ± SD), with 174 of 292 (60%) of examinations with COVID-19 demonstrating a PA diameter ≥28 mm, suggestive of pulmonary hypertension.12Clinical readings in the same cohort revealed that 56 of 336 examinations (17%, 3 scans were not interpretable) revealed pulmonary thrombosis/embolism.

Large animal physiology
To further understand our clinical ndings, we explored the systemic effects of the RAAS by infusing three sedated swines with ANGII.Within ve minutes, arterial and PA blood pressures started to climb (Fig. 2).Target arterial systolic blood pressures were set at 150 mmHg and the start dose of ANGII was adopted from the ATHOSIII trial.15Swine #1 was subject to 530 minutes of ICU care and reached PA-systolic (PAs) pressures exceeding 30 mmHg after 360 minutes.Swine #2 reached PAs > 30 mmHg after 30 minutes and died of acute right ventricular heart failure with extensive pulmonary thrombosis/embolism after 300 minutes.Swine #3 reached PAs > 30 mmHg in 390 minutes.ANGII infusion further induced a rapid decline in SvO2, falling below 50% in all swines.Swine #1 had a warm body temperature, which is a paradoxical nding when administering a vasopressor.Surface body temperature was measured in Swine #2 and #3 and skin temperature did not differ more than 2 degrees from 90 minutes into the experiment until death.Arterial blood gas sampling showed declining trends in PaO2 and trends towards a small increase in PaCO2 (Fig. 2).No lactate could be detected in the blood.

Large animal radiology and dissection
When performing an autopsy on Swine #1, we found a 13 cm long pulmonary thrombus (Fig 3c).To exclude embolisation, we performed ultrasonography of the hind legs to screen for deep venous thrombosis (DVT) every second hour in Swine #2 and #3, but none were detected.However, we did nd extensive pulmonary thrombosis formation in Swine #2 visualised by conventional angiography and later con rmed at autopsy (Fig. 3).Swine #3 had disseminated thrombotic material in the pulmonary arteries, as illustrated by a supplementary dissection videos.Ultrasonography without DVT of the hindlimb suggests local clot formation in the pulmonary circulation and not embolic material from the deep venous system.Macroscopically, we found wedge-shaped areas of red discoloration in the periphery of the lung (Fig. 3a).Samples for histology were acquired from these areas.

Large animal histology
The lung sample unexpectedly sank to the bottom of the formaldehyde solution (i.e.not oating as expected from healthy lung tissue).A red discoloration in the frozen sections was noted in the cryostat and the lung was surprisingly easy to cut, without the normal tendency of shattering when cutting snap-frozen lung tissue.After hematoxylin and eosin staining, we observed thrombotic material in both large and small vessels of the lungs.We found severe thickening of the alveolar walls and what might be hyaline membranes and/or debris in the alveolar sacs (Fig 3).

Large animal clinical chemistry data
We performed bleeding time assessments in Swine #2 and #3 with baseline bleeding times of 285 seconds and 255 seconds respectively.These were reduced to < 120 seconds after 90 minutes and remained low for the remainder of the experiment.D-dimer increased in Swine #1, doubling after 120 minutes, and almost doubled in Swine #2 after 120 minutes.Paradoxically, brinogen also increased or remained unchanged in spite of increased D-dimer, suggesting a possible unknown link between ANGII and brinogen synthesis, not caused by an acute phase reaction, since IL-6 and TNF-α remained below the level of detection in all swines during the experiment.Osmolality successively increased in all three swines by > 11 mosmol/kg through another unclear mechanism.We measured no changes from baseline values of Owrens PT (INR), APT-time, antithrombin, C-reactive protein, ferritin, albumin, creatinine, cystatin C, urea, ASAT, ALAT, γ-GT, LD, CK, myoglobin or triglycerides.

Discussion
Our results demonstrate a similar pattern between patients with COVID-19 admitted to the ICU and ANGII-infused swine.ANGII has previously been used in the ATHOSIII clinical trial.15Sub-analyses of this trial highlighted important thrombotic and infectious complications associated with ANGII.16 If SARS-CoV-1 disturbs the distribution of the ACE2-receptor to produce a subsequent increase in ANGII,3 it is likely that the same mechanism of receptor disruption in SARS-CoV-2 leads to the same effects.Further, a linear relationship seems to exist between ANGII levels and clinical outcome.7 Infusion of ANGII, resulting in an acceleration of the RAAS, induces the same clinical presentation as a loss of RAAS brake, or a loss of ACE2-receptors.The most appropriate dose of ANGII used in swines can be debated, however, injection of an ACE2 blocker (PD123319) in humans leads to an increase in mean arterial pressure of 12 mmHg after ve minutes.17 In the pulmonary circulation, using lung perfusion MRI, we demonstrated a reduction and, in some areas, likely cessation of blood ow, even in the absence of in ltrates.We further support these ndings by CTPA and echocardiography in a retrospective patient cohort.There are a number of methods to evaluate possible signs of increased pressure in the lung vasculature bed using CTPA.One of the most commonly applied methods in a clinical setting is measuring the PA diameter.Axial PA diameter measurements have been found to be a relatively sensitive parameter for detecting even borderline pulmonary hypertension with a sensitivity of 80% and a speci city of 62%, when using a cutoff ≥28 mm.12Given the characteristics of the measurement, it is thus likely that our measurements somewhat underestimate the fraction of patients with increased PA pressure, which is supported by our echocardiography ndings indicative of elevated PA pressures.In combination, our ndings provide multimodal evidence of an elevated PA pressure as a hallmark of COVID-19 and multiple lines of evidence in support of our two case reports and further our swine model data.
We believe our results indicate that COVID-19 patients are prothrombotic, as observed in the ICU, and that this may be caused by disruptions in at least two tenets of Virchow's triad, including: stasis in the pulmonary blood ow by vasoconstriction and a hypercoagulable state with increased brinogen synthesis, shortened bleeding times and rather distinct histological changes after only 500 minutes of ANGII infusion in our swine models.Other publications even suggest disruption of the third tenet of the triad, by endothelial dysfunction due to endothelitis.18Altogether, these ndings support our conclusion that severe COVID-19 is a vascular syndrome with a hypercoagulable state.Our results are translations between a large animal model and the clinical presentation of patients with COVID-19 in the ICU, therefore, these ndings may only be considered preliminary.The CTPA cohort is retrospective and lacks an internal control group.However, all the cumulative agreement of the individual ndings lends support to our hypothesis.Furthermore, the hypothesis of viral-ARDS or cytokine-release syndrome should probably be considered weakened by the observed pathophysiological effects in both large animals and clinical data.
If our assumptions are true, then many of the currently ongoing clinical trials may prove less successful than symptomatic targeting of vascular/coagulation disturbances or dysregulation of the RAAS.Our retrospective cohort of CTPA demonstrates a high incidence of pulmonary embolism in COVID-19 patients.Extended use of CTPA and treatment with anticoagulation therapy is therefore warranted in clinical guidelines.In a time of pandemic disease, we urge capable large animal laboratories to obtain ethical permissions to validate our ndings and encourage clinical centres with MRI capabilities to evaluate the potential clinical bene ts of lung perfusion MRI.We will assist any clinical site by answering questions about MRI protocols to the best of our abilities.We will continue to model severe COVID-19 as a vascular syndrome to advance our understanding of the pathophysiological mechanisms and use this model for large scale pre-clinical therapy evaluation.We will also use lung perfusion MRI to evaluate our patients when translating these preclinical ndings to our ICU patients.

Magnetic resonance perfusion
This case is submitted with written consent by next of kin.A 1.5 Tesla Philips Ingenia MRI scanner, software version R 5.4 (Philips Healthcare, Best, Netherlands), was used for perfusion analysis.An anatomical T2-weighted sequence was used to identify lung in ltrates with the following parameters: MultiVane, 2D eld-of-view 460×460 mm, voxel size 0.9×0.9×4mm, echo/repetition times at 100/2458 ms with a compressed sensing-sensitivity encoding factor 2, 1 average and 36 slices.For the dynamic contrast series, we used 4D time-resolved MRI angiography with a keyhole T1-weighted gradient-recalled-echo with eld-of-view 500×500 mm, voxel size 1.5×1.5×4mm, echo/repetition times at 1.8/3.78ms, a compressed sensing-sensitivity encoding factor 3.6, 1 average with 16 dynamic phases and 60 slices.A Max 3 contrast injector (Ulrich Medical, Ulm, Germany) was used to administer gadolinium-based contrast agent: gadobutrol solution, 1.0 mmol/ml, 2 ml, followed by 20 ml 0.9% saline solution with 5 ml/s.The rst phase acquired the entire K-space in 8.3 s and subsequent phases used a keyhole acceleration with 20% scanning, resulting in a temporal resolution of 1.7 s per phase.TTP maps were generated using the T1 MRI perfusion application in Philips Intellispace v10.1.3(Philips Healthcare, Best, Netherlands).

Computed pulmonary angiography
A consecutive cohort of all patients with RT-PCR-con rmed COVID-19 undergoing CTPA at Karolinska University Hospital in Huddinge, Stockholm, Sweden, between March 2nd ( rst patient admitted) and May 20th were retrospectively and independently evaluated by two raters (T.G., radiologist; J.A., 4th-year medical student) according to a standardised scheme,19 after consultation of a senior thoracic radiologist (M.K.).The ratings of the radiologist was considered the gold standard and the inter-rater agreement was assessed by intraclass correlation coe cient.All CTPA imaging was performed on Siemens SOMATOM De nition Flash (Siemens Healthineers, Erlangen, Germany), GE Discovery CT750 HD and Revolution CT (GE Healthcare, Milwaukee, USA).A ow chart describing the CTPA cohort can be found in Fig. 4. Echocardiography data was retrospectively analysed in a sub-sample (N=50) of patients undergoing CTPA imaging by a senior clinical physiologist (G.A.) to extract the TR-Vmax, SPAP and RVOT-AT, as previously described, with normative values from the litterature.13,14All echocardiography was performed on GE Vivid S70 (GE Healthcare, Milwaukee, USA).The CTPA and echocardiography sub-study was approved by the Swedish Ethical Review Authority (no.2020-01895); Informed consent was waived due to the retrospective nature of the study.

Clinical and preclinical measurement of PA
This submitted with written consent by next of A Swan-Ganz catheter (Edwards Lifesciences, Irvine, USA) for measurements of PA pressure was placed in the reported patient and in all of the swines.In the patient, it was placed from the jugular vein and in the swines from the femoral vein and connected to a HemoSphere and Vigilance monitor (Edwards Lifesciences, Irvine, USA) respectively.

Swine experimental setup
This study took place at the Karolinska Experimental Research and Imaging Centre, Karolinska University Hospital, Stockholm, between May 11th and May 15th, 2020.All animal studies were conducted according to Karolinska Institutet guidelines for animal experiments.The study was approved by the Regional Ethics Committee for Animal Research in Stockholm, Sweden (no.6716-2020).Three swines with weights between 36 and 39 kg were used in this study.Each animal fasted for 12 hours with free access to water before the procedure.They arrived sedated after premedication with intramuscular cepetor vet 1 mg/ml-zoletil 100 (Vetmedic/Virbac, Thirsk, UK) 0.8-1 mg/kg.Induction of anaesthesia was conducted with pentobarbital (Sandoz, Holzkirchen, Germany) 1-3 mg/kg and fentanyl (B.Braun, Melsungen, Germany) 2.5 μg/kg as an intravenous bolus dose.Maintenance of anesthesia was achieved with continuous infusion of pentobarbital (0.1-0.2 mg/kg/min) and morphine (Meda, Solna, Sweden) (0.1-0.25 mg/kg/h) titrated to a moderate depth of anesthesia.No muscle relaxants were used during the experiment.using an endotracheal tube (7.0) was performed after induction of anesthesia and during spontaneous breathing.The animals were normoventilated by pressure-controlled ventilation with Siemens Servoventilator 900C (Siemens Healthineers, Erlangen, Germany) with a total tidal volume of 10 mL/kg and with an inspiratory oxygen fraction (FiO2) of 0.21.End-tidal CO2 was monitored with a capnography incorporated in the surveillance system (Datex, GE Healthcare, Milwaukee, USA).Ventilation and respiratory frequency were adjusted to maintain end-tidal CO2 between 4.7-5.3kPa.A 5-lead electrocardiogram was recorded continuously as well as arterial pulse oximetry (SpO2) for peripheral oxygen saturation (Datex, GE Healthcare, Milwaukee, USA).The Pulse oximetry was placed on the tail.All animals received a urine catheter and hourly diuresis was recorded.Prior to skin incision, local anesthesia (lidocaine 10 mg/ml, Aspen Nordic, Ballerup, Denmark) <0.4 mg/kg) were given.A 7 F central venous catheter (Therumo, Tokyo, Japan) was inserted by open technique via the left internal jugular vein for administration of drugs.
ANGII infusion started at 20 ng/kg/min in Swine #1 and were stepwise elevated to 80 ng/kg/min in 60 minutes, followed by a stepwise increase to a maximum dose of 240 ng/kg/min after 515 minutes except for Swine #2 who suffered acute right ventricular heart failure at a dose of 100 ng/kg/min and died.This schedule was then used in subsequent swines.Bleeding time was assessed by cutting the skin of the ear for 10-15 mm and removing any blood every 15 seconds.When no further bleeding could be identi ed, this was determined to be the bleeding time.

Preclinical imaging
All conventional angiography was performed with a Philips XD20 angiographic system and 3DRA workstation (Philips Healthcare, Best, Netherlands).Visipaque 270 contrast agent (GE Healthcare, Milwaukee, USA) was used for all contrast-enhanced applications.Xper-CT images obtained were reviewed using the XD20 systems XperCT high dose program and soft tissue algorithms.
For endovascular access, we punctured the vessels using a micropuncture set (Merit Medical AB, Stockholm, Sweden) guided by ultrasonography with Siemens Acuson Sequoia 512 (Siemens Healthineers, Erlangen, Germany).Access was established with an introducer in the right femoral artery, the right femoral vein (Swine #1 and #3) and the left femoral artery (Swine #2).We used a 7 French introducer except for the PA catheter where we used an 8 F introducer (Terumo, Tokyo, Japan).We placed the 7.5 F PA catheter (Edwards Lifesciences, Irvine, USA) by uoroscopy guidance and the location was con rmed by invasive pressure.A distal access guide (Envoy 6 F, Cordis, Santa Clara, USA) was then placed in the proximal part of the descending aorta for continuous monitoring of aortic pressure and a 7 F pigtail catheter was placed in the right atrium of Swine #2 for pulmonary angiography.

Figures Figure 1 MRI
Figures

Figure 3 Macroscopic
Figure 3 Ultrasonography was performed every second hour during the experiment to exclude deep venous thrombosis in the hind legs.Pre-clinical chemistry analysisBlood samples were acquired at baseline and at 120, 240, 360, 435, 545, 665, 760 minutes or directly prior to death when systolic blood pressure dropped below 70 mmHg.Arterial blood gases were obtained at baseline, 75, 270, 345, 435, 545, 665, 750 min or when close to circulatory collapse.Citrated platelet-poor plasma and serum samples were analysed using proprietary assays at the Karolinska University Laboratory, accredited according to ISO 15189 by the Swedish Board for Accreditation and Conformity Assessment.Coagulation parameters were analysed on the Sysmex CS-5100 System (Siemens Healthineers, Erlangen, Germany), chemistry analyses were analysed on the Cobas 6000 Analyzer (Roche Diagnostics, Basel, Switzerland) and osmolality in serum was analysed with the Osmometer Advanced 2020 Multi-Sample (Advanced Instruments, Norwood, USA).The Mathworks Inc., Natick, USA) and IBM SPSS Statistics version 25 for Mac (IBM, Armonk, USA).A threshold of P<0.05 was considered to be statistically signi cant.Statistical analysis of manually drawn regions of interest in normal-appearing lung tissue on MRI perfusion were analysed using MATLAB.Extrapulmonary and extramediastinal tissues were manually masked.Other descriptive statistics on MRI perfusion were also generated in MATLAB.Inter-rater agreement of CTPA measurements were calculated using intraclass correlation coe cient of average measurements in SPSS.