key: cord-0944655-z9a54f6g
authors: Gaibazzi, N.; Martini, C.; Mattioli, M.; Tuttolomondo, D.; Guidorossi, A.; Suma, S.; Dey, D.; Palumbo, A.; De Filippo, M.
title: Lung disease severity, Coronary Artery Calcium, Coronary inflammation and Mortality in Coronavirus Disease 2019.
date: 2020-05-06
journal: nan
DOI: 10.1101/2020.05.01.20087114
sha: 9209355df61d976fbf9bd29d1ec12b0153690491
doc_id: 944655
cord_uid: z9a54f6g

IMPORTANCE The in-hospital mortality rate of the coronavirus disease 2019 (COVID-19) is higher in case of myocardial injury, but the underlying mechanism is not known and might depend on pre-existing coronary artery disease (CAD), coronary inflammation or others. OBJECTIVE To determine the association of the extent of lung disease or coronary artery chest computed tomography (HRCT) variables, the Agatston coronary calcium score (CCS) and peri-coronary adipose tissue attenuation (PCAT), representing CAD and coronary inflammation, with mortality in patients with COVID-19. DESIGN Retrospective case series. SETTING Single academic institution, Parma University Hospital, Italy, between March 5, 2020 and March 15, 2020. Final follow-up: March 30, 2020. PARTICIPANTS 500 consecutive patients with suspected COVID-19 who underwent HRCT as a gatekeeper were initially selected and the subgroup with laboratory-confirmed SARS-CoV-2 infection formed the final study group. EXPOSURES SARS-CoV-2 infection by real-time reverse transcriptase polymerase chain reaction (RT-PCR) assay of nasopharyngeal swabs. MAIN OUTCOMES AND MEASURES In-hospital mortality was the end point. Demographic, clinical, laboratory and HRCT data were collected from hospital electronic records, and HRCT features (CCS and PCAT) were measured post-hoc from HRCT images. RESULTS Among 500 patients with suspected COVID-19, 279 had laboratory-confirmed COVID-19 and formed the study group. Among them, 170 patients (61%) were discharged alive and 109 (39%) died. Comparing patients discharged alive with patients who died, the median age was 65 vs 77 (p<0.001), with males 56% vs 68% (p=0.061), prior cardiovascular disease 9% vs 24% (p=0.001), median D-dimer 723 vs 1083 ng/ml (p<0.001), median C-reactive protein 78 vs 148 mg/L (p<0.001), the mean CCS 17 vs 189 (p<0.001) and the median PCAT -76.4 HU vs -68.6 HU (p<0.001). In multivariable analysis, only age (p<0.001), D-dimer (p=0.041), C-reactive protein (p=0.002), extent of lung disease (p=0.002), and PCAT (p<0.001), remained associated with in-hospital death. CONCLUSIONS AND RELEVANCE Increased age, D-dimer, C-reactive protein and the HRCT image features of extent of lung disease and coronary inflammation by PCAT (but not the CCS) were independently associated with mortality in hospitalized COVID-19 patients. Our study suggests that higher mortality in COVID-19 may be at least partly mediated by coronary artery inflammation.

This retrospective, single-centre, observational case series study was performed at the Parma University Hospital, Parma, Italy, which has been designated to treat patients with COVID-19 in this area in the North of Italy, serving slightly less than half-million citizens.

The first consecutive 500 patients, who presented at the dedicated hospital triage with clinically suspected COVID-19, between March 05, 2020 and March 15, 2020, and were indicated for fasttrack HRCT were first selected; the subgroup of such patients who subsequently had SARS-CoV-2 viral nucleic acid detected in nasopharyngeal swabs formed the final study group. (12) This group was analyzed to evaluate the association of independent variables with in-hospital mortality.

Definition of clinically suspected . Clinical variables considered at triage to define 'clinically suspected COVID-19' were a) recent/current fever for at least 2 days, b) dry cough, c) respiratory distress during mild exercise or at rest. O 2 saturation<96% (<94% for patients with chronic pulmonary disease) using pulse oximetry was defined reduced O 2 saturation. The presence of at least one of the abovementioned clinical criteria + reduced O 2 saturation qualified the subject as clinically suspected COVID-19 and mandated fast-track HRCT, as the gatekeeper to admit patients to dedicated COVID-19 wards if showing interstitial pneumonia, even if SARS-CoV-2 viral nucleic acid laboratory results were pending. All patients were admitted to hospital after HRCT, either to COVID-19 specialized or general wards, based on the presence or absence of HRCT findings suggestive of COVID-19 pneumonia.

Definition of laboratory-confirmed COVID-19. The detection of SARS-CoV-2 viral nucleic acid in nasopharyngeal swabs was used as the only reference to adjudicate laboratory-confirmed COVID-19. This study complied with the edicts of the 1975 Declaration of Helsinki and was approved by the institutional ethics board of Parma University Hospital with expedited approval for COVID-19 studies (371/2020/OSS/AOUPR). Written informed consent was waived by the ethics commission due to the emergency state caused by COVID-19.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10.1101/2020.05.01.20087114 doi: medRxiv preprint

The electronic discharge letter and other available hospital electronic records were reviewed by a team of physicians working in Parma University Hospital. Clinical and laboratory information was collected on admission and during hospitalization by attending physicians and was then checked and recorded by the researchers. Patient data including demographics, medical history, comorbidities, available laboratory examinations, nasopharyngeal swabs tested with SARS-CoV-2 RT-PCR assay, or other intensive care measures and outcomes were collected.

Patients were scanned with a 128-slice Somatom Definition Edge scanner (Siemens Medical Solutions, Forchheim, Germany), equipped with an integrated high-resolution circuit detector.

Examinations were performed with patients in supine position, at full inspiration, without the use of contrast medium. Scan parameters were: collimation 128×0.6 mm; rotation time 0.33 msec; pitch 1.2; tube voltage 120 kVp; tube power 150 reference mAs; tube current modulation on. Scan data were reconstructed with novel model based iterative reconstruction algorithm (ADMIRE) keeping the noise at a low level and preserving intrinsic low-contrast details.

Interpretation of HRCT. Each scan was clinically evaluated by the radiologist in charge, all having more than 10-year experience, in particular using the following classification (similar to the recent Consensus Statement by the Radiological Society of North America (13) . In our centre, the three following summary categories were used: a) presence of COVID-19 typical interstitial pneumonia, according to the presence of ground-glass opacities, bilateral multifocal patchy consolidation, and/or interstitial changes with a peripheral distribution, b) indetermined appearance, in case of absence of typical features and presence of multifocal, diffuse, perihilar, or unilateral ground-glass opacities with or without consolidation lacking a specific distribution and non-rounded or nonperipheral, c) negative for pneumonia. The report also focused per-protocol on the estimated semiquantitative percentage of lung parenchyma involved (5 visually-assessed classes, <10%, 10-19%, 20-49%, 50-79%, >80%).

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10.1101/2020.05.01.20087114 doi: medRxiv preprint Coronary calcium was classified using a threshold of 130 Hounsfield units (HU) involving 3 contiguous voxels for identification of a calcific lesion resulting in a minimum lesion area of 1.02 mm 2 . The lesion score was calculated using the area density method, by multiplying the lesion area by a density factor derived from the maximal HU within the area, as described by Agatston (14) .

The CCS was calculated with commercially available software (CaScore; Siemens, Germany).

To measure PCAT, we used a software package clinically validated for this use (AutoPlaque TM Version 2.5, Cedars-Sinai Medical Center). (15) Since in the current study we used high-quality and fast HRCT machines, but the scans were not ECG-triggered nor contrast medium used, we customized the methods otherwise typically used for PCAT measurement; we manually identified the coronary artery lumen and adjusted the vessel contours. The left anterior coronary artery (LAD) was the only coronary artery vessel reliably visualized in most patients (less motion artifacts compared with right coronary artery) in the non-contrast, non ECG-triggered HRCT datasets for the aim of PCAT measurement. Figure 1 shows the technical steps to measure PCAT attenuation. We traced the proximal 40-mm segment of the LAD and 3-dimensional layers within radial distance from the outer coronary wall equal in thickness to the average diameter of the vessel were constructed automatically, using the abovementioned semiautomated software Within the predefined volume of interest, voxels with tissue attenuation ranging from -190 up to -30 HU were considered as adipose tissue. We quantified the mean PCAT attenuation based on the attenuation histogram of peri-coronary fat within the range -190 HU to -30 HU. (10,15,16)

The primary endpoint was incidence of in-hospital death in laboratory-confirmed COVID-19 patients. Successful clinical course leading to hospital discharge comprised relieved clinical symptoms, normal body temperature, normal O2 saturation with no need of O2 therapy and at least partial decrease of inflammation as shown by C-reactive protein testing.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10.1101/2020.05.01.20087114 doi: medRxiv preprint

No statistical sample size calculation was performed a priori, and sample size was equal to the number of patients enrolled during the study period. Categorical variables are expressed as number of patients (percentage) with 95% CIs, and continuous variables as mean (SD) or median (interquartile range [IQR]) as appropriate. The means for continuous variables were compared using independent group t tests when the data were normally distributed, otherwise, the Mann-Whitney test was used. Proportions for categorical variables were compared using the χ 2 test, although the Fisher exact test was used when data were limited. Stepwise multiple logistic regression was used to assess the relationship between the demographic and clinical variables, laboratory variables and, as a final step, HRCT variables and the primary end point of death. Variables that were not normally distributed (such as CCS, C-reactive protein, D-dimer) were log-transformed, as is standard. We considered an increase in PCAT attenuation in steps of 1 HU and 5 HU; the results were similar and increase by 5 HU were reported for multivariable analysis. All variables with p<0.1 on univariable analysis were considered for the inclusion into multivariable logistic regression models. A 2-sided p<0.05 was considered statistically significant. All statistical analyses were performed with Stata statistical software, version 15.0 (StataCorp LLC, USA).

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10.1101/2020.05.01.20087114 doi: medRxiv preprint

Suspected COVID-19 overall group (n=500): Among the initially 500 consecutive patients with clinically-suspected COVID-19 who underwent HRCT, n=145 died (29%) during their index hospitalization, at a median of 6 (lower-upper quartile, 4-10) days after admission, while n=355 were discharged home alive after a median of 6 (lower-upper quartile, 2-13) days. n=279 (55.8%) were found positive for RT-PCR viral nucleic acid detection in the nasopharyngeal swab obtained either at presentation, or repeated in the next 2 days if the first was negative, and they received the diagnosis of laboratory-confirmed COVID-19.

In this initial cohort n=399 patients had typical findings of COVID-19 interstitial pneumonia on HRCT, n=66 were negative for COVID-19 pneumonia and n=35 were indeterminate cases. Figure 2 shows the mortality rate in the overall n=500 population, based on HRCT findings of typicality for COVID-19, but also based on laboratory detection of SARS-CoV-2 nucleic acid in nasopharyngeal swabs. HRCT findings for COVID-19 demonstrated the following prevalence of SARS-CoV-2 nucleic acid detected in patients' nasopharyngeal swabs: 248/399 (62%) in COVID-19 typical HRCT, 19/35 (54%) in indeterminate HRCT, 12/66 (18%) in negative HRCT. Figure 2 shows that the visual semi-quantitative estimate of diseased lung parenchyma, as reported by the radiologist in charge (here summarized in 3 classes, 0-19%, 20-49% and ≥ 50%) was associated with mortality, in laboratory-confirmed COVID-19 patients but also in patients with negative report for SARS-CoV-2 nucleic acid in the nasopharyngeal swabs.

These patients with laboratory-confirmed COVID-19 formed the final study cohort, described and tested for univariable and multivariable association between demographic, laboratory, standard and novel variables on HRCT and inhospital mortality.

In this cohort, n=170 (61%) patients were discharged home alive at a median (lower-upper quartile) of 8 days (5-16) after admission, while n=109 (39%) died during hospitalization, at a median (lower-upper quartile) of 6 days (4-10). Table 1 shows the baseline demographic, clinical, . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. . . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. . 1 1

The first model with clinical+laboratory variables showed a receiver-operating-characteristic (ROC) curve area of 0.808 (95%CI 0.756-0.859), which was significantly increased by the addition of HRCT variables to the model, with a resulting ROC area of 0.861 (95%CI 0.817-0.905), p=0.004. Figure 4 shows the comparison between two paradigmatic cases, one with lower PCAT indicating no coronary inflammation, and the other one showing higher PCAT (closer to 0 HU), indicating coronary inflammation.

Coronary artery calcium, measured in the HRCT scans as a proxy for pre-existent subclinical coronary artery disease, was not associated on multivariable analysis with the primary end point of mortality, apparently excluding that the cardiac contribution to increased mortality in COVID-19

patients is mediated by a demand-supply mismatch mechanism, caused by pre-existent CAD.

Peri-coronary adipose tissue attenuation, which correlates with coronary artery inflammation, was independently associated with in-hospital mortality, as it was also the case for age, D-dimer, Creactive protein and extent of diseased lungs on HRCT. The semi-quantitative assessment of lung disease, a simple HRCT estimate available in real-time by the radiologist in charge was significantly and independently associated with mortality. Coronary inflammation was assessed using the recently described PCAT variable, a non-invasive volumetric measurement of x-ray attenuation of the adipose tissue in direct, close contact with the coronary arteries, which is an indirect 'thermometer' of coronary inflammation (10, (15) (16) (17) . Figure 3 shows the output PCAT images of a comparison between two paradigmatic, opposite cases.

Before the results of the current study we could not exclude that the prolonged hypoxemia typical of symptomatic COVID-19 represented a stress condition leading to myocardial injury, particularly in the elderly patients with asymptomatic CAD, through a demand-supply mismatch. Since the CCS is able to indirectly measure pre-existent CAD burden, and it was not independently associated with . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10.1101/2020.05.01.20087114 doi: medRxiv preprint in-hospital death, we may reasonably conclude that pre-existent (known or unknown) CAD has no or limited role in worsening the outcome of COVID-19 patients. The finding that a higher coronary inflammation status in COVID-19 patients is associated with mortality is instead a novel finding and may shed some light on the possible mechanisms responsible for the frequently-found myocardial injury, associated with higher mortality in COVID-19 patients.(4-7) However, the current study did not address the incidence of myocardial injury (cardiac troponins were not routinely measured in our COVID-19 cohort), so that while we can affirm that coronary inflammation is associated with in-hospital death, we cannot speculate on the relation between coronary inflammation and myocardial injury. The finding of a disproportionate increase in Creactive protein, together with higher PCAT in patients with fatal outcome, supports the existing hypothesis that inflammation in COVID-19 is a key driver of outcome, being a possible mechanism of coronary plaque destabilization, as well as a potential trigger for cardiac arrhythmias.(18) This opens the possibility that the potent anti-inflammatory drugs already in testing phase in prospective trials may help decrease mortality.

Although the demonstration of SARS-CoV-2 nucleic acid in the nasopharyngeal swab remains the reference for COVID-19 diagnosis (12) , it has several limitations, which may limit its diagnostic sensitivity to lower than ideal. The use of HRCT as an initial gatekeeper, as performed in the current study in presence of symptoms and signs suggestive of the disease, may prove useful during massive outbreaks, as the one the world is facing. In an outbreak setting the inherently limited specificity of HRCT (which aims at the detection of a typical interstitial pneumonia, and not specifically at the detection of the virus) is sufficient to produce a moderate-to-high positive predictive value, due to the high prevalence of disease (high pre-test probability). In turn, HRCT has a very high sensitivity and negative predictive value, which may be preferable in the gatekeeping setting of highly contagious diseases. (19) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10.1101/2020.05.01.20087114 doi: medRxiv preprint COVID-19 prognosis by HRCT. The current study also demonstrated that HRCT is capable to accurately risk-stratify patients with acute respiratory distress and laboratory-confirmed COVID-19 through simple semi-quantitative estimation of the extent of lung parenchyma affected by pneumonia, which represents the pulmonary key clinical manifestation of COVID-19. Interestingly, such risk-stratification capability, although to a lesser degree, was demonstrated also in patients in whom viral nucleic acid was not detected in at least two nasopharyngeal swabs (Figure 2 ). Whether this is due to RT-PCR on swabs being insufficiently sensitive or to HRCT being able to stratify also non-COVID-19 patients would require an independent reference method for COVID-19 disease, which is not available at this time.

Our study has limitations. First, we cannot unequivocally classify those patients with HRCT findings typical of COVID-19 who tested negative to viral nucleic acid detection in nasopharyngeal swabs, since the direct virus detection remains the WHO recommended standard for COVID-19 diagnosis. (12) The mounting clinical experience during the outbreak suggests that most such patients may be affected by COVID-19, since the nasopharyngeal detection of the virus is limited in sensitivity and the optimal timing to perform the swab unknown. Second, as a retrospective study, specific information regarding the prevalence of optional variables, for example the presence of myocardial injury by high-sensitivity troponins, were not available, owing to the difficult clinical conditions in the isolation wards and the urgency of containing the epidemic. Another limitation was that the presence of obesity was adjudicated as a binary variable by the caring physicians (body mass index>29 kg/m2), but individual granular data are not available.

Third, though the methods used to calculate the CCS in non-contrast, non ECG-triggered HRCT scans have been previously validated (11, 20, 21) , the HRCT protocol is not the gold-standard to measure the CCS, or for the measurement of PCAT. To our knowledge, ours is the first study to report such PCAT measurement and its clinical results. The absolute value of fat attenuation in . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10.1101/2020.05.01.20087114 doi: medRxiv preprint Hounsfield units might potentially differ between an ideal ECG-triggered and contrast computed tomography angiogram and a non-contrast, non-ECG triggered HRCT as available in the current study; however, since all patients underwent exactly the same modality of HRCT scans, the relative comparison between alive and dead patients in our study should be unbiased by this limitation.

Further we measured PCAT only in one coronary artery (LAD), because it was the one technically best visualized in the given conditions of HRCT scans. Prior literature robustly supports that PCAT does not differ significantly among coronary arteries in a given individual, and right coronary artery or LAD may be used alternatively as inflammatory markers of overall coronary inflammatory status. (15-17) 

While the current study confirms prior data that increased age, D-dimer and C-reactive protein are associated with in-hospital mortality in COVID-19, we also report that the simple extent of lung disease and the novel PCAT parameter of coronary artery inflammation in HRCT scans are also independently associated with mortality. This may hypothetically lend support to therapeutic strategies aimed at lowering inflammation during COVID-19, which would possibly exert their positive effects also on the inflamed coronary arteries.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10.1101/2020.05.01.20087114 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. PCAT, peri-coronary adipose tissue attenuation; LAD, left anterior descending.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. . Figure 1 .

.

It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 6, 2020. 

China Medical Treatment Expert Group for Covid-19. Clinical characteristics of coronavirus disease 2019 in China

Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak in China: summary of a report of 72 314 cases from the Chinese Center for Disease Control and Prevention

Case-Fatality Rate and Characteristics of Patients Dying in Relation to COVID-19 in Italy

Association of Coronavirus Disease 2019 (COVID-19) With Myocardial Injury and Mortality

Association of cardiac injury with mortality in hospitalized patients with COVID-19 in Wuhan

Cardiovascular Implications of Fatal Outcomes of Patients With Coronavirus Disease 2019 (COVID-19)

COVID-19 and the cardiovascular system

Cardiac involvement in a patient with coronavirus disease 2019 (COVID-19)

Coronary calcium score improves the estimation for pretest probability of obstructive coronary artery disease and avoids unnecessary testing in individuals at low extreme of traditional risk factor burden: validation and comparison of CONFIRM score and genders extended model

Detecting human coronary inflammation by imaging perivascular fat

Coronary artery and thoracic calcium on noncontrast thoracic CT scans: Comparison of ungated and gated examinations in patients from the COPD Gene cohort

Clinical management of severe acute respiratory infection when novel coronavirus (nCoV) infection is suspected: interim guidance

Radiological Society of North America Expert Consensus Statement on Reporting Chest CT Findings Related to COVID-19. Endorsed by the Society of Thoracic Radiology, the American College of Radiology, and RSNA. Radiology: Cardiothoracic imaging

Quantification of coronary artery calcium using ultrafast computed tomography