Late Complications of COVID-19: A Morphologic, Imaging, and Droplet Digital Polymerase Chain Reaction Study of Lung Tissue

Material from autopsies of decedents at our institution, as well as surgical lung specimens including explants and wedge biopsies, needle core biopsies, transbronchial forceps biopsies, and cryobiopsies not performed for suspected neoplasm (May 2020–April 2021) in patients who tested positive for SARS-CoV-2 by nasopharyngeal or oropharyngeal swab, were retrieved either from clinical archives or from the authors' consultation files. Patients younger than 18 years were excluded. The onset of symptoms of COVID-19 or the date of positive testing, whichever date was earlier and/or known, was more than 28 days before specimen acquisition, the time span that was incorporated into the COVID-19 treatment guidelines by the National Institutes of Health as “long COVID.”^16,17  Clinical information was abstracted from medical records or information provided in the consultation letter. Clinical severity of disease was defined according to the National Institutes of Health COVID-19 treatment guideline as mild (typical infectious or constitutional symptoms without respiratory symptoms), moderate (infectious/constitutional symptoms with respiratory symptoms or mild chest infiltrate [<50%], with oxygen saturation on room air >94%), severe (respiratory rate >30, oxygen saturation on room air <94%, or >50% bilateral chest infiltrates), or critical (intensive care unit [ICU] admission with severe respiratory failure, additional organ dysfunction, or severe sepsis) at the time of presentation, corresponding with diagnostic SARS-CoV-2 PCR testing.¹⁶  As duration of acute illness along with definitions of late COVID are not standardized and vary in the literature, we defined acute COVID-19 illness as date of onset of symptoms or date of positive SARS-CoV-2 PCR testing, whichever date was earlier and/or known, to date of the following if hospital admission occurred: home discharge if not admitted to ICU, transition to floor status if admission to ICU without mechanical ventilation, extubation if mechanically ventilated, or transition to extracorporeal membrane oxygenation (ECMO). This definition attempted to define the direct effects of acute infection rather than prolonged illness from initial organ injury. Along the same lines, we defined clinical late COVID-19 as recurrence of typical infectious or respiratory symptoms more than 21 days from initial positive SARS-CoV-2 testing to distinguish it from prolonged illness or organ injury after suspected infection has waned or resolved.

Findings from a subset of autopsies have been previously reported.^3,18 

For all autopsy cases, the legal next of kin provided consent and permission for research. This study was approved by the institutional review board (21-002940).

All lung specimens were independently reviewed by at least 2 thoracic pathologists (A.C.R., J.M.B., M.C.A., Y.C.L., Y.M.B., J.J.M., B.T.L., H.D.T., A.K., M.L.S., E.S.Y., M.C.B.) blinded to radiologic findings and clinical history. Ancillary testing was performed in selected cases based on morphologic findings, including Grocott methenamine silver, Verhoeff–van Gieson, sulfated Alcian blue, Congo red, and Masson trichrome stains. Morphologic features that were recorded are detailed in Supplemental Table 1 (see supplemental digital content, containing 2 tables, at https://meridian.allenpress.com/aplm in the July 2022 table of contents). Major disagreements were resolved by rereview of the case by a third thoracic pathologist. Consensus morphologic features were used for analysis.

Formalin-fixed, paraffin-embedded (FFPE) tissue blocks were cut at 4 μm. Slides were stained with antibodies to CD68 (clone KP1, DAKO, Glostrup, Denmark), keratin (clone AE1/A3, DAKO), calretinin (clone CAL6, Leica Biosystems Newcastle Ltd, Newcastle, United Kingdom), and cytokeratin 5 (clone XM26, Leica Biosystems) in selected cases.

FFPE tissue with the most severe morphologic findings of ALI was selected for RT-ddPCR. RT-ddPCR was performed as previously described.¹⁸  Positive cases were defined as those with more than 4 droplets of the N1 and/or N2 target.

Any available CT imaging studies of the chest since onset of COVID-19 were reviewed by a thoracic radiologist (T.F.J.) blinded to clinical information and morphologic findings. Time to CT scan was calculated as days from COVID-19 onset to the most recent CT scan, which was regarded as the scan closest to specimen acquisition. Days between specimen acquisition and CT scan were also calculated. Patterns on CT scans were categorized into acute, subacute, late, or atypical for COVID-19 based on the expert consensus statement by the Radiological Society of North America along with historical literature on chronic lung injury evaluated with imaging.¹⁹  The acute category included findings of ground-glass opacity (GGO) and/or consolidation without evidence of bronchial dilation or bronchiectasis.²⁰  The subacute category consisted of GGO and/or consolidation along with findings of early organization such as mild architectural distortion or early bronchial dilation.²¹  The late category included imaging findings of prominent bronchiectasis, reticular opacities, and architectural distortion.²² 

Data were summarized with frequencies and percentages or medians and ranges, as appropriate. Selected comparisons were performed between groups using Fisher exact tests for categorical variables and Kruskal-Wallis or Wilcoxon rank sum tests for ordinal or continuous variables. Spearman correlation was used to estimate the association between days from COVID-19 onset and RT-ddPCR copy number. The date of COVID-19 onset was defined as the date of onset of symptoms or the date of the first positive SARS-CoV-2 test, whichever was earlier and/or available. All P values are 2-sided, and P < .05 was considered statistically significant. All analyses were performed using SAS version 9.4 (SAS Institute Inc, Cary, North Carolina).

Forty-four specimens from 43 patients who had onset of symptoms or had tested positive for SARS-CoV-2 more than 28 days prior to specimen acquisition or death were enrolled in the study. One patient had 2 specimens. Patient demographics, pertinent clinical findings, and origin of the specimen (Mayo Clinic versus consultation files of the authors) are summarized in Table 1 and detailed in Supplemental Table 1. Severity of acute illness, available for 23 patients, was dominated by severe or critical disease, with the majority of patients presenting with respiratory failure. Among these 23 patients, only 3 met our clinical definition of late COVID, manifesting as recurrence of infectious or respiratory symptoms greater than 21 days after initial infection. No patient was vaccinated at the time of the initial positive SARS-CoV-2 testing, with a majority of study patients diagnosed prior to 2021 when COVID-19 vaccination became more widely available. Five patients (of 23; 21.7%) had suspected secondary or superimposed infections on presentation that were distinct from hospital- or ventilation-acquired pneumonias.

Median time between COVID-19 onset and specimen acquisition was 97 days (range, 30–359 days). The majority (32 of 43; 74.4%) of patients had one or more comorbidities (Table 1). Connective tissue diseases included psoriasis/psoriatic arthritis (n = 2), rheumatoid arthritis (n = 1), and systemic lupus erythematosus and rheumatoid arthritis (n = 1). Treatment of patients is summarized in Table 1. COVID-19–related therapies included hydroxychloroquine, steroids, antibiotics, remdesivir, and convalescent plasma. Lung transplantation was performed because of sequelae of COVID-19 in 2 patients. Among the 12 patients in whom an autopsy was performed, 10 (83.3%) died of complications of COVID-19 (Supplemental Table 1).

CT imaging studies were available in 17 patients; the results are summarized in Table 2 and detailed in Supplemental Table 1. The time between COVID-19 infection and CT scan that was performed closest to the time of specimen acquisition was significantly shorter in patients with consolidation (median, 43 days; range, 7–101 days) versus patients without consolidation (median, 87.5 days; range, 14–233 days) (P = .02, Wilcoxon rank sum test). On the other hand, reticulations were associated with a longer time between COVID-19 onset and CT scan (median, 82 days [range, 43–150 days] versus median, 23.5 days [range, 7–70 days]; P = .006, Wilcoxon rank sum test) (Figure 1, A and B). There was no significant difference in the time between CT scans with regard to presence versus absence of GGOs, crazy paving, pleural effusion, or lymphadenopathy (P = .41, P = .45, P = .27, and P = .34, respectively; Wilcoxon rank sum tests). Furthermore, there was no significant association between radiologic pattern of COVID-19 and occurrence of a COVID–19-associated histologic ALI (P = .28, Fisher exact test), although the number of patients was very small for this analysis. In 3 patients, the CT scans were interpreted as atypical for COVID-19 pneumonia and lungs showed focal ALI.

All 3 patients who had an appearance that was atypical for COVID-19 pneumonia by CT had a time between CT and biopsy of 76 to 89 days, whereas all other patients with available CT scan had a time of 1 to 29 days.

In 5 patients, CT studies were performed within 7 days of specimen acquisition (Table 2). In all 5 patients, CT studies showed subacute or late COVID-19 pattern, and whereas all 5 lung specimens had morphologic features of ALI, they also showed features of fibrotic DAD.

Types of specimens and number of sections are detailed in Table 1 and Supplemental Table 1. Most of the specimens (37 of 44; 84.1%) were obtained by autopsy (n = 12; 27.3%) or explantation (n = 2; 4.5%) or as wedge biopsy (n = 23; 52.3%), which facilitates the macroscopic evaluation of some pathologies such as consolidation or hemorrhage and decreases sampling bias.

Gross findings of lungs at autopsy or time of explantation from 14 patients are summarized in Table 3 and detailed in Supplemental Table 1. Combined weights were increased, ranging from 1120 to 2700 g (expected, 650–850 g). Consolidations were noted in 10 patients (of 14; 71.4%) (Figure 2, A and B).

Histologic findings in lungs of patients with COVID-19 for more than 28 days are summarized in Table 3 and detailed in Supplemental Table 1. Most specimens (39 of 44; 88.6%) showed at least a component of ALI, either diffuse or focal, including acute and/or organizing DAD (Figure 2, C through E), organizing pneumonia (OP), acute fibrinous and organizing pneumonia, and/or acute fibrinous pneumonia. Thirty-three (of 44; 75.0%) specimens harbored fibrosis, including 22 specimens with findings resembling a fibrotic NSIP pattern; in the context of COVID-19 and ALI, this pattern was interpreted as fibrotic DAD (Figure 3, A and B). Cicatricial (fibrosing) OP was seen in 12 specimens (Figure 3, C through E). A subset of specimens harbored parenchymal cysts (n = 4). The parenchymal cysts were of various sizes and commonly lined by a discontinuous layer of CD68-positive macrophages or multinucleated giant cells surrounded by predominantly mature fibrosis (highlighted by Masson trichrome) with focal fibroblast proliferation (Figure 4, A through E). Keratin AE1/AE3, cytokeratin 5, and calretinin did not reveal any cyst-lining cells. At least a subset of these cysts may have resulted from cicatricial OP (Figure 4, F). None of the 4 specimens with cysts showed any morphologic findings suggestive of centrilobular emphysema; in 3 of these patients with available imaging there was also no suggestion of emphysema, and all 4 patients were nonsmokers. In 2 of the specimens the cysts appeared in the vicinity of a bronchovascular bundle; in the other 2 cases the cysts seemed randomly distributed. Two of the patients with cysts had been intubated, for 11 (patient 7) (randomly distributed cysts) and 67 (patient 34) (cysts in the vicinity of bronchovascular bundle) days; the intubation status was unavailable in another case. In addition, in one patient (patient 7) scattered thin-walled pulmonary cysts in the setting of multiple myeloma had been described in a CT scan 1 year prior to COVID-19. In this patient, although smaller cysts were surrounded by fibrosis and lined by macrophages as described above, a larger cyst was discontinuously lined by keratin AE1/AE3–positive, cytokeratin 5– and calretinin-negative epithelial cells (Figure 5, A through D). Only rare CD68-positive macrophages were scattered along the cyst lining. Furthermore, there was no fibrosis in the wall of these cysts, as also evaluated on Masson trichrome.

Patients with diffuse ALI had a shorter time between COVID-19 onset and specimen acquisition (median, 61.5 days; range, 30–279 days) than patients with focal (median, 140 days; range, 30–298 days) or no ALI (median, 130 days; range, 106–35 days) (P = .009, Kruskal-Wallis test). Time between COVID-19 onset and specimen acquisition was longest for patients with morphologic diffuse interstitial fibrosis of any pattern (median, 106 days; range, 30–279 days) when compared with patients with focal fibrosis (median, 101.5 days; range, 30–359 days) or no fibrosis (median, 83 days, range, 35–299 days), although this difference was not statistically significant (P = .96). Similarly, time between COVID-19 onset and acquisition of specimens that showed fibrotic DAD was not significant among these 3 groups (P = .58).

Twelve patients (of 43; 27.9%) had a lung disease that was either known or suspected before contracting COVID-19 based on clinical and/or radiologic features. The clinical diagnoses of these lung diseases included pulmonary fibrosis (n = 3), emphysema/chronic obstructive pulmonary disease (n = 2), undefined ILD (n = 2), possible sarcoidosis (n = 2), emphysema and pneumonia 1 year prior to COVID-19, chronic pulmonary thromboembolism, and bronchiolitis obliterans syndrome after lung transplantation for NSIP (n = 1 each). The microscopic findings in these patients are detailed in Supplemental Tables 1 and 2. Emphysema was microscopically identified in 6 patients. In contrast, patient 9 was unlikely to have had an underlying fibrotic ILD prior to COVID-19. Although that patient was thought to have had rheumatoid arthritis 12 years prior to COVID-19 based on symptomatology, there had been no recurrence of symptoms noted since then, and the patient did not have any lung-related complaints prior to acquiring COVID-19. Furthermore, the patient went into respiratory failure that was temporally associated with the diagnosis of COVID-19 and required mechanical ventilation, eventually leading to ventilator-associated pneumonia and demise of the patient. Almost one-third (12 of 44; 27.3%) of specimens showed recent thrombi (Figure 5, E). Only 4 specimens (of 44; 9.1%) revealed old recanalized thrombi.

SARS-CoV-2 RT-ddPCR was positive in 80% (16 of 20) of cases (Table 3). All positive results were in specimens that showed diffuse (n = 11) or focal (n = 5) features of ALI, including type II pneumocyte hyperplasia (n = 15), organizing DAD (n = 13), OP (n = 13), fibrotic DAD (n = 7), acute DAD (n = 7), acute fibrinous pneumonia (n = 6), organizing fibrinous pneumonia (n = 5), and acute fibrinous and organizing pneumonia (n = 4). Positive results were observed in specimens from patients between 30 and 174 days after COVID-19 onset. Negative results were found in specimens from patients between 106 and 252 days. There was a slight negative correlation between days from COVID-19 onset and RT-ddPCR copy number (correlation coefficient, −0.281). However, copy numbers of more than 10 000 were identified in lung tissue of patients at days 37, 83, and 97 following COVID-19 onset. These 3 patients had either focal or diffuse ALI on histopathology. A correlation between presence of histopathologic ALI and RT-ddPCR could not be performed, as RT-ddPCR results were not available in the 5 cases that did not have features of ALI.

In our systematic study of morphologic, imaging, and viral findings in lungs of patients with onset of COVID-19 at least 28 days prior, we found that ALI can be seen beyond the early postinfectious phase for up to 10 months. We had previously shown that RT-ddPCR for SARS-CoV-2 was positive in lung tissue up to 55 days after COVID-19 onset.¹⁸  However, in the current study, we found that RT-ddPCR can identify SARS-CoV-2 viral RNA in lung tissue up to nearly 6 months after onset. We also found that three-quarters of patients had morphologic features of pulmonary fibrosis. Although in a subset of patients the fibrotic background was most likely due to a preexisting fibrotic ILD, in most patients, the fibrosis appeared to be a new finding and was attributable to prior COVID-19. Morphologic features that were regarded as sequelae of a COVID-19–associated ALI included a fibrotic NSIP-like pattern that was interpreted as fibrotic DAD and cicatricial OP. In addition, our study showed that recent thrombi were still present in a quarter of patients with onset of COVID-19 at least 28 days prior. However, these thrombi seemed likely to resolve, as recanalized thrombi were observed in only less than 10% of our patients. Furthermore, we showed that CT imaging studies may have limitations in predicting COVID-19 infection. However, discrepancies between imaging and morphologic findings may, at least in part, be due to a relatively long interval between CT scanning and tissue acquisition in our study.

Only a few case reports and series have described morphologic lung findings in patients with COVID-19 beyond the early postinfectious phase. For instance, Bharat et al¹⁴  studied explanted lungs of 12 patients who underwent transplantation for sequelae of COVID-19–associated ARDS and described extensive ongoing ALI, specifically DAD, with lung fibrosis. Some patients also had acute bronchopneumonia or cavities with necrosis. The authors described interstitial fibrosis with honeycomb change and bronchiolization of alveoli, and, in a subset of patients, thrombi with or without recanalization in small and intermediate-sized vessels. Similarly, in our study, almost 90% of patients had morphologic findings of ALI, most commonly OP, organizing DAD, and acute DAD; acute bronchopneumonia was also present in a third of our patients. In contrast to our study, in which only a quarter of patients were intubated and less than 10% were on ECMO, all patients in the study by Bharat et al¹⁴  had prolonged intubation and ECMO. This raises the possibility that ALI and thrombi may have, at least in part, been due to ECMO treatment. Indeed, we have previously shown that 60% and 47% of patients treated with ECMO have findings of ALI and thromboembolism, respectively, which was more common than in non-ECMO patients.²³  In addition, microthrombi can be seen in association with ALI, specifically DAD and OP, regardless of etiology and are not specific to COVID-19.²⁴  Alternatively, survival in the face of prolonged end-stage critical illness may lead to thromboemboli. Bharat et al¹⁴  also described severe pleural adhesions in all of their patients. Although we also identified moderate to severe pleural adhesions, we found them in only about a third of patients. These discrepancies are likely because all patients in the study by Bharat et al¹⁴  had severe lung disease requiring transplantation, whereas our patients may or may not have had severe COVID-19–related symptoms. Interestingly, 1 of the 2 patients who underwent transplantation in our study had only focal pleural adhesions.

In a subset of our patients, we observed morphologic features of DAD and OP simultaneously. OP features can occur as part of DAD,^24,25  and DAD in general is responsible for the severity of the disease. However, in patients with later stages of COVID-19, conceivably patients may have had OP initially and developed DAD later or vice versa.

In a series of 3 patients who were treated with ECMO, NSIP-like pattern of fibrosis with honeycomb change, with or without superimposed ALI or extensive infarction of the lung with DAD, was identified in tissue acquired 38 to 126 days after COVID-19 infection.¹³  In that study, thrombotic complications were not found, although one patient did show extensive infarction.

It has been speculated that fibrosis may develop at least in a subset of patients who have recovered from COVID-19–related ARDS. However, systematic morphologic studies have been lacking. In our study of 43 patients who survived COVID-19 for at least a month, fibrotic changes were thought to be either due to a preexisting fibrotic ILD in at least 6 patients or, more commonly, due to sequelae of COVID-19. Fibrotic patterns due to COVID-19 included fibrotic DAD that had features similar to fibrotic NSIP. These findings were not surprising, as DAD is thought to evolve predominantly within the interstitium, where hyaline membranes are either resolved or replaced by fibroblasts, which eventually either resolve or are replaced by collagen fibrosis, leading to an NSIP-like pattern.²⁶  Pulmonary fibrosis following ALI has also been described in patients with severe acute respiratory syndrome (SARS)–CoV and Middle East respiratory syndrome (MERS)–CoV.^27,28  In a study of 10 patients with SARS-CoV, 3 patients showed morphologic signs of fibrosis, including interstitial and airspace organization and dense septal and alveolar fibrosis.²⁹  An autopsy study of 7 patients with SARS-CoV also described interstitial fibrosis, mild to moderate (n = 6) or severe (n = 1); the latter also showed honeycomb changes.³⁰  Imaging studies reported fibrosis on follow-up imaging in patients diagnosed with MERS-CoV.²⁸ 

In addition to a fibrotic DAD pattern, we identified features of cicatricial OP in almost a third of our specimens. Cicatricial (fibrosing) OP is characterized by dense intra-alveolar collagenous fibrosis associated with plugs of proliferative fibroblasts or possibly linear fibrous bands.^31–33  Dendriform ossifications may also be seen in cicatricial OP. This pattern has been described in association with fibrotic ILD,³²  cryptogenic OP,^31,33  Ehlers-Danlos syndrome, aspiration, and underlying connective tissue disease.³¹ 

Lungs from patients with COVID-19 for at least 28 days continued to show morphologic features of ALI; indeed, ALI could be seen for up to 10 months after COVID-19 onset. Diffuse ALI was more likely noted earlier after infection, whereas focal ALI was identified both early and later in the disease course. We also were able to detect SARS-CoV-2 viral RNA for up to nearly 6 months after infection using RT-ddPCR from lung tissue. Our results suggest that ALI may be, at least in part, due to active viral infection in the lung, likely in concert with secondary phenomena such as local and systemic cytokine production. However, treatment-related lung injury may also contribute to ALI, given that more than half of our patients had been treated with anti–COVID-19 therapy, including remdesivir, convalescent plasma, hydroxychloroquine, and/or antibiotics. Indeed, in a recent clinical trial including 397 patients with COVID-19 who were treated with remdesivir, 6% and 11% of patients who received the drug for 5 or 10 days, respectively, experienced respiratory failure.³⁴  In addition, a quarter of our patients were intubated for various periods of time and a few underwent ECMO, a treatment that by itself has also been associated with ALI.²³  Our results further confirm that testing for SARS-CoV-2 RNA using RT-ddPCR is a highly sensitive method of detection of SARS-CoV-2 RNA in FFPE tissue¹⁸  and may be helpful in the diagnosis of COVID-19 and COVID-19–associated ALI as long as some ALI is present, although detection of viral RNA does not necessarily confirm that the virus is currently active/replicating. Whether viral RNA can be identified in FFPE lung tissue when no morphologic findings of ALI are seen remains unclear, as no tissue from any of our patients without ALI tissue was available for RT-ddPCR.

Thromboemboli have been frequently described in early COVID-19.^3,5,8,35  In our population with later stages of COVID-19, recent thrombi were noted in the lungs of a quarter of patients; however, recanalized thrombi were seen in less than 10% of patients. In only one patient were arterial infarcts identified associated with recanalized thrombi. This finding is similar to that of Aesif et al,¹³  who did not report any thrombotic complications, although a large infarct was described in one patient. Bharat et al¹⁴  reported recanalized thromboemboli in some patients. Reasons for the relatively low number of organized or recanalized thrombi in late-stage disease are not entirely clear, but may be attributable to the fact that more than half of our patients were treated with thrombolytics.

In 4 patients, we identified scattered cysts. In all 4 patients the cysts showed a fibrotic wall and were lined by macrophages; in 2 patients the cysts were located in the vicinity of a bronchovascular bundle at least suggestive of a lymphangitic distribution. These findings suggest that the cysts represent pulmonary interstitial emphysema, although a sequela of cicatricial OP may also be a possibility in at least a subset of the cysts. Pulmonary interstitial emphysema has most commonly been observed in premature infants with ARDS but has also been described in adults and has been shown to correlate with ventilation and/or biopsy.³⁶  Indeed, 2 of the 4 patients had been intubated. Interestingly, in one of the patients we also observed larger cysts that did not have a fibrotic wall and were lined by epithelial cells rather than macrophages. These cysts were likely of a different etiology. Interestingly, in this patient, thin-walled cysts were described on CT imaging studies a year prior to COVID-19, and therefore most likely represented a preexisting condition.

In our study, imaging findings of acute, subacute, and late COVID-19 were not significantly associated with morphologic findings of ALI. Although this discrepancy could be due to a relative inability of imaging studies to predict morphologic findings in these patients, another reason may be the relatively long interval between imaging and specimen acquisition. The latter is further supported by our findings in the small subset of patients with CT scans within 7 days of specimen acquisition, which all showed a CT pattern consistent with subacute or late COVID-19 and for which specimens revealed fibrotic DAD on microscopy. Studies that compared radiologic with histologic features were largely performed in the early post–COVID-19 infection phase. A study comparing CT imaging with histologic features in patients who died between 5 and 44 days after onset of COVID-19 symptoms also showed that CT patterns typically observed in COVID-19, such as GGOs and consolidations, were not related to specific histologic patterns.³⁷  In addition, tissue specimens obtained from radiologically unaffected lung parenchyma demonstrated morphologic evidence of vascular damage and thrombosis. Another study of 70 lung lobes from 14 autopsy cases in which CT imaging studies were performed at a mean interval of 3.7 days before death (range, 0–17 days), and with a mean time of 10.4 days between onset of symptoms and death, showed that GGOs correlated with capillary dilatation and congestion (97%; 57 of 59 lobes), interstitial edema (90%; 53 of 59) acute DAD (78%; 46 of 59), microthrombosis (37%; 22 of 59), organizing DAD (36%; 21 of 59) and rarely acute bronchopneumonia (13%; 8 of 59). Consolidations on CT correlated with capillary dilatation and congestion (94%; 31 of 33), interstitial edema (88%; 29 of 33), and acute DAD (64%; 21 of 33) most often, and in some instances with microthrombosis (42%; 14 of 33) or organizing DAD (24%; 8 of 33).³⁸  Lastly, the number of cases available for comparison between histologic and imaging findings was small in our study and therefore underpowered for that particular analysis.

The morphologic and imaging findings in lungs in our cohort of patients who had at least 28 days of COVID-19 corroborate clinical studies that show that lung function and imaging findings did not return to baseline in a proportion of patients with COVID-19. Indeed, a study by Huang et al³⁹  that followed patients for up to 1 year after COVID-19 diagnosis demonstrated that lung diffusion less than 80% of predicted was present in 23% (13 of 56) of patients in severity scale 3 (admitted to hospital but not requiring supplemental oxygen), 31% (36 of 117) in scale 4 (admitted and requiring supplemental oxygen), and 54% (38 of 70) in scale 5 or 6 (admitted and requiring noninvasive mechanical ventilation or invasive mechanical ventilation and ECMO, respectively) groups. Similarly, total lung capacity was less than 80% in 5% (3 of 76), 7% (8 of 17), and 29% (20 of 70) of patients in each group, respectively.³⁹  CT scans of these patients 12 months after COVID-19 also showed GGOs in 39% (11 of 28), 27% (14 of 52), and 76% (29 of 38) of patients, respectively, consolidations in 0% (0 of 28), 0% (0 of 52), and 3% (1 of 38), and reticulations in 0% (0 of 28), 2% (1 of 52), and 8% (3 of 38), respectively.³⁹ 

Our study has several limitations. First, given that many cases were from consultation files, the presence or absence of clinical symptoms and/or an ALI shortly after contracting COVID-19 was unknown. Therefore, the incidence of ILD following ALI in patients with COVID-19 could not be calculated based on our data. In a few cases it was also difficult to differentiate an underlying fibrotic ILD from COVID-19–associated fibrotic ILD, as an underlying ILD was not known at the time of infection. Similarly, in some patients, contributions to ALI from an adverse drug reaction, a ventilator- or ECMO-related complication, and/or underlying diseases such as connective tissue disorders cannot be ruled out. Although our study population is relatively large for a study that includes lung specimens from patients with COVID-19 for at least 28 days, statistical significance may not have been reached for some potential associations because of low numbers of patients in certain subgroups.

In conclusion, our data show that ALI can be seen for many months after COVID-19 onset. The ALI can resolve or instead evolve into fibrotic ILD. Furthermore, testing of lung tissue for SARS-CoV-2 RNA is of value even in patients who were infected several months prior to specimen procurement, at least when morphologic features of ALI are present. Continued refinement of imaging features in COVID-19 patients through comparison with pathology specimens and clinical correlation should be sought. Finally, larger studies are necessary to more thoroughly elucidate new-onset, postinfectious fibrotic lung disease in patients with COVID-19 so that clinical management of affected patients continues to improve.