To the best of our knowledge, this is the first report describing lung MRI T1 and T2 mapping findings in hospitalized patients with COVID-19 pneumonia and healthy controls. Native T1 and T2 values of the lung parenchyma were higher in the patients than in the controls, either when using a regional or global measurement approach. The T2 value of the affected regions retained the best diagnostic accuracy to differentiate the patients against the controls, whereas a higher reproducibility was obtained when assessing the global lung native T1 and T2 values. Myocardial native T1 was higher in the COVID-19 patients than in the controls.

MRI mapping techniques are increasingly used to characterize myocardial changes in a variety of clinical conditions [7,8,9], including COVID-19 [10], and this is consistent with our observation of the increased myocardial native T1 values detected in the patient group. On the other hand, few studies are currently available regarding tissue characterization when using a lung MRI, of which none of them include COVID-19 patients. Gargani et al. [3] reported data from a cohort of patients with systemic sclerosis, observing an increased signal by using T2-weighted STIR sequences, consistent with the presence of edema and, thus, inflammation. Previous studies investigating lung T1 mapping in pediatric patients with cystic fibrosis found reduced values as compared to controls, which is possibly explained by the presence of reduced blood flow due to fibrosis or vasoconstriction in the affected areas [4]. Together, these studies demonstrate the ability of MRI in detecting abnormalities within lung tissues in diverse clinical conditions, mainly due to the different fluid distribution, either blood or edema. In our study, we observed the concomitant increase of native T1 and T2 values in the affected lung areas of the patients with COVID-19 pneumonia. This finding should be read as secondary to the presence of extensive edema, since the presence of water influences both the rise of the specific marker T2 as well as the native T1 [11]. The presence of edema is consistent with acute pneumonia and ongoing lung inflammation, which is also one of the therapeutic targets during the acute phase of COVID-19 [12]. Results from our study suggest that native T1 and T2 mapping could be potentially used not only for the identification but also parametric quantification of the ongoing inflammation of lung parenchyma. To date, scarce data are available in the literature regarding the optimal approach for the tissue mapping evaluation of the lungs. If, on the one hand, the regional assessment could be more informative regarding local pathologic processes within the affected areas in the course of pneumonia, this is achieved at the expense of a lower reproducibility of the measure, as indicated by the results of the blinded re-reading of the images in our sample. Translating the evidence obtained in the field of cardiovascular diseases by using cardiac MRI, it might be assumed that better results with tissue mapping techniques should be obtained when applying them to diffuse rather than regional diseases, when the entire architecture of the organ is affected [13]. Notwithstanding this, the ROC curves from our study suggested a potential diagnostic role for the global lung mapping assessment too, that could be explained by the relatively diffuse nature of COVID-19 pneumonia as compared to other subtypes, such as those of bacterial origin. One limitation of this report was the small sample size that prevented a comprehensive description of any existing association between native T1 and T2 with pneumonia severity, systemic inflammation and outcome. Furthermore, where active inflammation is likely the cause of the detected MRI abnormalities, further studies are needed to assess whether lung MRI T1 and T2 mapping could be effectively used to image the long-term consequences of COVID-19 pneumonia after recovery, including fibrosis [14]. T1 and T2 mapping techniques can provide an accurate tissue characterization; however, they are limited by the need for a standardized setup for the imaging and definition of normal values [15], with variable results obtained when using different field strengths, machine vendors and/or post-processing software, which may limit the generalizability and immediate transferability of the absolute quantitative values reported here.