Prevalence and risk factors for lung involvement on low-dose chest CT (LDCT) in a paucisymptomatic population of 247 patients affected by COVID-19

In December 2019, a series of pneumonia cases caused by a novel coronavirus occurred in Wuhan, Hubei, China [1‐3]. The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was named by the International Committee on Taxonomy of Viruses [4]. Coronavirus disease 2019 (COVID-19) spread worldwide from Asia to Europe and is now in the ascending phase of the epidemic in America [2]. The World Health Organization (WHO) declared it a world pandemic situation on March 11, with over 110,000 cases, and the number is still increasing [5].

Involvement of the disease has a wide variety of clinical features, from cough to pulmonary failure [6, 7]. Moreover, a large number of patients remain totally asymptomatic, allowing the pandemic to spread even more easily [8, 9]. Reverse -transcription polymerase chain reaction (RT-PCR) is the main tool for diagnosis but does not allow for the assessment of disease severity. Chest X-ray is not recommended in the initial phase of the disease due to its low value to detect ground-glass opacities. Low-dose chest computed tomography (LDCT) appears to be a useful tool in the management of patients during the COVID-19 epidemic. LDCT is very sensitive for diagnosis, quantification of disease severity and identification of a differential diagnosis. Furthermore, after recovery, LDCT might be of interest in the prediction of lung fibrosis during healing [10‐13]. Kang et al. emphasized the role of LDCT in the diagnosis of COVID-19, especially in the early stages of the disease. Furthermore, Burian et al. showed that the proportion of lung involvement could be a risk factor for hospitalization in the intensive care unit, suggesting that the extent of lung involvement is clinically relevant [14]. The goal of the study was to (1) determine the prevalence and risk factors for lung involvement on LDCT according to clinical symptoms and comorbidities in all consecutive patients with a positive RT-PCR test over a short period of time (12 days) and (2) evaluate whether LDCT is able to detect other abnormalities in an incidental way that requires care or just medical follow-up.

In a predominantly paucisymptomatic patient cohort of 247 consecutive COVID-19 patients, we found a prevalence of 138/247 (56%) patients with lung involvement. A total of 109/247 (46%) had a normal CT. Risk factors for lung involvement were age > 54 years and diabetes in multivariate analysis on comorbidities and only age > 54 years in multivariate analysis including comorbidities and symptoms. Rhinitis and anosmia appeared to be protective factors against lung involvement. Incidental findings were noticed in 47/247 patients (19%), but only 14/247 (0.6%) needed follow-up.

These results show that age > 54 years and diabetes were risk factors for finding at least moderate lung involvement on LDCT, which is in line with other reported data about comorbidities on COVID-19 patients’ risk [20, 21]. A meta-analysis of 1558 patients found that significant risk factors were hypertension, diabetes, chronic obstructive pulmonary disease, cardiovascular disease, and cerebrovascular disease. Another Italian retrospective study from Grasselli et al. on 1591 patients in the intensive care unit showed that the mortality rate was superior in patients with ages greater than 64 years compared with younger patients [22, 23]. Since no clinical symptoms were risk factors for severe impairment on LDCT, probability to have an abnormal chest CT should be reliably based on comorbidities rather than symptoms. Another study on mortality likewise showed that no symptoms were risk factors for mortality [24] that is consistent with no symptoms were risk factor to find abnormalities on LDCT. This prevalence is in line with previous study on paucisymptomatic population, asymptomatic cases from the cruise ship “Diamond Princess” were 73%, with 54% of them showing lung abnormalities on CT, whereas 27% were symptomatic and 79% of them had abnormal findings on CT [25]. Yang et al. in a letter to the editor explained that the prevalence of pneumonia on CT is very different according to different study and he recall that prevalence rely on the population explored, a significant number of paucisymptomatic patient might have lung involvement [26].

Initial LDCT in paucisymptomatic cases could be helpful to better assess fibrosis development after recovery. After SARS-CoV-1 infection, a study showed that fibrosis features on CT more than 4 weeks after initial symptoms included reticular opacities, architectural distortion and bronchial dilatation in the same area as initial involvement [27]. More than 4 months after recovery, approximately 10% of 258 SARS-CoV-1 patients had permanent sequelae, especially older patients and those who required intensive care units for acute distress respiratory syndrome [28].

A pitfall of LDCT is the absence of contrast injection. COVID-19 disease has been shown to be a risk factor for pulmonary embolism [29] especially in the intensive care unit [30]. During data analysis, there was no evidence of a link between COVID-19 and the risk of thrombosis. The population studied was predominantly a paucisymptomatic population. Initially, D-dimers were not among biological dosages, and most of the patients had a low risk for thrombosis according to usual scores such as Wells or modified Geneva score. Following the evidences of recent publications, D-dimers have been systematically tested, and if the result is positive, chest CT angiography is performed instead of LDCT [31]. Currently, during follow-up, in cases of clinical worsening, unexplained tachycardia or persistent inflammatory syndrome, chest CT angiography is performed to ensure the absence of pulmonary embolism.

The published CT-SS has been chosen to classify disease extension on LDCT because this is a visual score permitting the quantification of lung involvement [18]. These advantages are that this method is easy to use, quick, wide and gradual, ranging from 0 to 40, and can differentiate mild from severe cases. Many scores have been suggested using quantitative or qualitative indicators [32, 33]. The CT score from Francone et al. has shown that a high score > 17/25 is highly predictive of patient mortality in short-term follow-up and that the parenchyma assessment is more accurate than inflammatory biomarkers in a multivariate analysis to evaluate the short-term outcome. A CO-RADS score was developed by Prokop et al. in the same way as TI-RADS or BI-RADS, but for suspected COVID-19 disease than to quantify the extent of the disease. In the presented population, due to positive RT-PCR results for all 247 patients, LDCT patients were all CO-RADS 6 [34].

Regarding the interest of the use of LDCT in COVID-19, a low-dose protocol has been recently recommended [35], but no significant results in routine clinical practice have been published to date. Since many of the patients should be referred for follow-up, LDCT is probably the best option, especially for younger patients [36]. LDCT may be repeated when clinical worsening occurs to detect a pneumothorax or if an additional bacterial infection occurs to look for lung abscess. This approach would ideal for patients in ambulatory care with a negative result for RT-PCR but with a high clinical suspicion for COVID-19 pneumonia to identify false negative nasal swab tests. Some studies have evaluated that a patient affected by COVID-19 could have 6 LDCTs during the course of the disease, and non-COVID-19 individuals could have 2 LDCTs to ensure that they are not affected [37].

To compare imaging features and prevalence of lung involvement for other viruses, such as influenza, parainfluenza and respiratory syncytial virus have been explored in symptomatic populations. The features encountered in these viruses are similar to those found for COVID-19 pneumonia. In a literature review, GGO was systematically associated with influenza lung involvement [38]. Several papers showed that the prevalence of lung abnormalities on chest CT was between 57 and 65% when pooling the data for all different viruses in a symptomatic population [39, 40]. In the presented population, the prevalence of abnormal findings was 54%, which could be considered high in paucisymptomatic patients. These results are very close to those found with severe acute respiratory syndrome coronavirus 1 (SARS-CoV-1) [41].

We reported incidental imaging findings in 47 patients (19%) and 14 patients (0.6%) will require follow-up. This low rate might be explained because our population is mostly young and has few comorbidities.

One limitation of this study is the short delay we observed in some of our patients between the first symptoms and LDCT, since chest CT may be normal in patients with COVID-19 within the 3 days following the appraisal of symptoms [42]. However, this percentage of patients was low in our population (25/247, 11%).

In this population of predominantly paucisymptomatic COVID-19 patients, the prevalence of lung involvement was 54%. Neither clinical symptoms nor signs were predictors of lung involvement, but age > 54 years and diabetes were risk factors for having a CT-SS ≥ 10. Rhinitis and anosmia appeared to protect against lung involvement. Longer follow-up is required to define what type of lesion or which patients might evolve towards pulmonary fibrosis sequelae.