Chronic Parenchymal Lung Changes after COVID-19 Infection

Background: Persistent parenchymal lung changes are an important long-term sequela of COVID-19. There are limited data on the disease characteristics and trajectories. This study aims to evaluate persistent COVID-19-related parenchymal lung changes after 10 weeks of acute viral pneumonia and to identify its risk factors. Methods: This was a retrospective case-control observational study involving 38 COVID-19 conrmed cases using nasopharyngeal swab reverse-transcriptase-polymerase-chain-reaction (RT-PCR) at King Abdullah Medical City (KAMC) hospital, Makkah. Patients were recruited from the Post-COVID interstitial lung disease (ILD) clinic. Referral to this clinic was based on the pulmonology consultants’ assessment of hospitalized patients suspected of developing COVID-19-related ILD changes during hospitalization. Measurements and Main Results: Nineteen patients with persistent parenchymal lung changes after 10 weeks of the acute illness (group-1) were compared with 19 control patients referred for assessment of post-COVID-19 ILD and had accelerated clinical and/or radiological features (group-2). Group-1 was found to have more severe clinical and radiological disease, with higher peak value of inammatory biomarkers. Two risk factors were identied, NLR >3.13 at admission increases the odds ratio (OR) of chronic parenchymal changes by 6.42 and 13.09 in the univariate and multivariate analyses, respectively. Invasive mechanical ventilation had a more profound effect with ORs of 5.92 and 44.5 in the univariate and multivariate analyses, respectively. Conclusion: Herein, persistent parenchymal lung changes were observed in several patients 10 weeks after acute COVID-19 infection. We found that only receiving invasive mechanical ventilation and NLR >3.13 at admission were strong risk factors for persistent parenchymal lung changes.


Background
The novel coronavirus disease was rst detected in December 2019 in Wuhan, China. In February 2020, the World Health Organization (WHO) announced that the coronavirus disease was caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), and it was named COVID-19. This shortly led to a global pandemic. By May 2021, more than 150 million con rmed cases and approximately 3 million deaths were reported worldwide [1]. Since the pandemic started, there has been an over ow of studies targeting disease pathophysiology, treatment, and prevention, and some have been published. However, many aspects and the nature of the disease, including long-term pulmonary sequelae, are not well understood. Early reports described persistent symptoms after acute COVID-19 infection [2] [3]. These symptoms and organ dysfunction were not limited to the lungs but included psychological, cardiovascular, neurological, hematological, and other system disorders [3], [4] . Persistent parenchymal lung changes, in particular, have been described in several observational studies. [5]- [7]. Nevertheless, there are limited data on the disease characteristics and trajectories. This study aimed to elaborate on the clinical and radiological features of parenchymal pulmonary sequelae and factors that may contribute to the development of brosis post-acute COVID-19 pneumonia.

I. Study design and patient selection
This retrospective case-control observational study involved 38 COVID-19 con rmed cases using nasopharyngeal swab RT-PCR at King Abdullah Medical City (KAMC), Makkah-Saudi Arabia. Patients were recruited from the Post-COVID-19 interstitial lung disease (ILD) clinic. Referral to this clinic was based on a pulmonologist's assessment of hospitalized patients suspected of developing COVID-19related ILD changes during hospitalization.
At the Post-COVID ILD clinic, detailed medical history and physical examination were performed.
Laboratory and physiological assessments were requested for all patients; however, follow-up chest CT was requested for patients with signi cant residual disease, such as patients with severe symptoms who still required oxygen at home in order to maintain oxygen saturation for more than two months after discharge and those in whom the prednisolone dose could not be tapered down. According to evidence of persistent COVID-19-related parenchymal lung changes, the 38 selected patients were divided into two groups. Those patients who had undergone follow-up chest CT, at least 10 weeks after the rst positive RT-PCR swab, and had residual parenchymal lung diseases were included in the case group "Group 1".
The control group included patients with earlier clinical and/or radiological resolution.
Inclusion criteria included all adults (age >12 years) and RT-PCR-con rmed COVID-19 infection with radiological evidence of pneumonia. We excluded patients diagnosed with brotic or other structural lung diseases prior to having COVID-19 infection. Patients who were lost to follow-up after discharge or judged to need a follow-up chest CT and missed their radiology appointments were also excluded.
Baseline patient characteristics, including age, sex, smoking history, and obesity, comorbidities, and other clinical variables related to acute COVID-19 disease, such as in ammatory markers, treatment intervention, and intensive care unit (ICU) admission, were collected from electronic the patients' records. II. Chest CT image protocol and interpretation CT scans were acquired without ECG gating on a 64-slice multidetector CT (Siemens SOMATOM Sensation 64) with a 64 × 0.625 mm collimation and a spiral pitch factor of 1.3. Scans were obtained in the craniocaudal direction, supine position, and during end-inspiration without the administration of an intravenous contrast agent with a standard dose scanning protocol. For patients who had a clinical suspicion of pulmonary embolism (PE), additional CT pulmonary angiography was conducted. Axial reconstructions were performed with a slice thickness of 1.5 mm.
Two radiologists, each with 6 years of experience, reviewed all chest CT images. The images were reviewed independently, and any discrepancies were resolved by consensus. There was no major disagreement between the radiologists' interpretations.
For each of the two patient groups, the CT examinations were evaluated for the following characteristics: (1) CT severity scores (CT-SS) for initial and follow-up CT as described by Li et al. The CT severity scores were based on a visual quantitative evaluation of the percentage of involvement in each lobe as well as the overall lung. The severity scores were classi ed as none (0%), minimal (1-25%), mild (26-50%), moderate (51-75%), or severe (76-100%), with corresponding scores of 0, 1, 2, 3, or 4, respectively. The CT-SS was reached by summing the ve lobe scores (ranging from 0 to 20).
In the "case group," the CT performed during acute illness was termed "the initial CT," and CT performed 10 weeks after the rst RT-PCR was termed "the follow-up CT." However, follow-up CT images were not obtained in most patients in the "control group" as they showed signi cant clinical improvement; thus, the rst CT performed during acute illness was termed as "the initial CT," and the latest CT subsequently performed (not necessarily after 10 weeks of the acute illness) was termed "the follow-up CT." III.

Results
The patients' demographic and clinical characteristics are presented in Table 1. Patients with persistent ILD were predominantly males (73.68%); however, this difference was not statistically signi cant between the two groups. Age, BMI, smoking history, and comorbidity were not statistically signi cant. 0.14 At admission, patients with a neutrophil to lymphocyte ratio of above 3.13 at admission were more prevalent in cases the case group (78.95%) than in the control group (36.84 %), and this was statistically signi cant (p = 0.01). Patients in group 1 had a higher peak value of in ammatory biomarkers, including ESR, CRP, PCT, LDH, and ferritin compared with those in the control group. However, these were not statistically signi cant; the same applies to the D-dimer peak admission value.
Overall, 84.3 % of cases had severe to critical disease in the control group compared with 100% in the case group (p = 0.10). About two-third of the patients in both the groups were admitted to the ICU: 52.63 % in the control group and 68.42 % in case group. The number of days in the ICU, pharmacological treatment, and oxygen therapy other than mechanical ventilation were also comparable. Mechanical ventilation was strongly associated with persistent parenchymal lung changes; other hospital interventions and clinic-related clinical data are shown in Table 2. The initial CT-SS was higher in the cases group than in the control group (13.06 and 9.58, respectively) (p = 0.043). However, after adjusting for age in the univariate and multivariate analysis, they were no longer signi cant with p-values of 0.05 and 0.08, respectively. Ground-glass opacities were detected in all cases, followed by parenchymal bands as the second most prevalent abnormality in nearly 80% of cases. None of the cases had a honeycomb appearance on chest CT imaging. The radiological severity score and other persistent parenchymal abnormalities on follow-up chest CT are shown in Table 3 and illustrated in Fig. 1. Table 3 Initial CT-SS as risk factor for post-COVID-19 brosis CT severity score (SD) 9.84 (5.75) Table 4 shows the univariate and multivariate analyses of both groups for the possible risk factors of post COVID-19 brosis. Our analysis showed that a neutrophil-to-lymphocyte ratio (NLR) of > 3.13 and receiving invasive mechanical ventilation increased the odds ratio (OR) of chronic parenchymal changes. The OR were 6.42 and 13.09, respectively, in the univariate analysis and 5.92 and 44.5, respectively, in the multivariate analysis. Male sex, obesity, and comorbidity had a modest increased OR but were statistically insigni cant. Other in ammatory markers and treatments are presented in Table 4.

Discussion
Persistent parenchymal lung changes beyond acute COVID-19 infection are one of the most important post-COVID-19 sequelae. Disease characteristics and trajectories have not been well studied. In this study, we followed-up with 38 patients, 10 weeks after COVID-19 infection, and compared a group of patients with persistent parenchymal lung changes with a group of patients who had clinical resolution of symptoms. We found that ground-glass opacity (GGO) was the predominant lung change after 10 weeks.
Clinical characteristics that independently predict a protracted course for pulmonary parenchymal changes included a NLR of > 3.13 at admission and invasive mechanical ventilation.
In a 12-month follow-up study of 311 patients with SARS, 21.5% of patients had lung brosis 65 days after discharge [11]. Das et al. reported that 13 out of 36 patients with Middle East respiratory syndrome (MERS) had persistent parenchymal changes 32-230 days after being discharged [12]. Regarding COVID-19, persistent symptoms, impaired diffusion capacity of the lung for carbon monoxide (DLCO), and persistent radiological ndings beyond acute illness have been reported [13]. Trinkmann et al. reported that 113 out of 246 patients remained symptomatic after a mean follow-up period of 68 days; dyspnea was the most common symptom (32%) [14]. In Wuhan, a study showed impairment of DLCO after 90 days of discharge in 54% of patients [15]. In Norway, a multi-center prospective study reported that onefourth of the patients had persistent CT ndings on follow-up 3 months after discharge from acute COVID − 19 hospitalization, and ICU admission was reported to be a risk factor. [16]. From the cases referred by the pulmonology consultants to follow-up of post-COVID-19 ILD, 19 cases con rmed using chest CT showed prolonged parenchymal abnormalities after at least 10 weeks of the acute infection. These cases were matched with 19 cases that showed early clinical and radiological recovery.
The pathogenesis beyond the development of lung brosis in COVID-19 survivors is not clear. It is likely that SARS-CoV-2 binds and interacts with angiotensin-converting enzyme (ACE)-2, which increases transforming growth factor beta (TGF β 1) and connective tissue growth factor (CTGF) levels, which may result in the development of brosis through the activation of brosis-related genes [17]. Surfactant abnormality and alveolar type-2 (AT2) cell injury result from the interaction between environmental factors, such as viruses and genetic factors, causing alveolar collapse, and repeated injury from ventilation could explain the progression to ventilator-induced lung injury (VILI) and lung brosis [18].
Evidence has shown that the most persistent ILD post-COVID-19 is reported in severe and critical cases [16], [19]- [21]. This raises the suspicion that ventilation-induced lung injury (VILI) and brosis following mechanical ventilation in ARDS has a major impact on the pathogenesis of post-COVID-19 ILD [18], [22].
Our study showed that receiving invasive mechanical ventilation had the highest impact on the likelihood of developing prolonged parenchymal changes. It increases the odds ratio (OR) of chronic parenchymal changes by 13.09 and 44.5 in the univariate and multivariate analyses, respectively.
Regarding disease severity, admission to the ICU and length of stay in the intensive care were not found to impact parenchymal lung sequelae; none of the medications administered for acute illness had an impact. However, CT-SS on initial chest CT was found to be higher in cases than in the control group, re ecting more radiologically severe disease in Group-1. After adjusting for age, the OR was found to be 1.16 (0.99-1.34), P = 0.05 and 1.45 (0.98-1.34), P = 0.08 in the univariate and the multivariate analysis, respectively.
A NLR of ≥ 3.13 was found to be an independent risk factor for severe and critical COVID-19 infection [23]. In this study, NLR was demonstrated to be a strong predictor of persistent parenchymal lung changes regardless of COVID-19 disease severity. There may be a role for increased neutrophils or decreased lymphocytes in disease pathogenesis and this requires further investigation.
In patients with idiopathic pulmonary brosis (IPF), a study has shown that a red cell distribution width (RDW) more than 14.1 was shown to be a negative prognostic factor as it correlates with lower forced vital capacity (FVC) and DLCO compared with IPF patients with normal RDW [24]. Increased RDW can be used as an indirect marker of hypoxemia. We did not nd a difference in the percentage of patients who had RDW values above 14.1 between the groups. The absolute RDW value was also not different. We selected the value measured at the date of hospital discharge to allow time for acute illness-related hypoxemia to impact the RDW value.
Most of the reported lung brosis following MERS-CoV were observed in patients that were admitted for longer days in the ICU, elderly patients, and those with higher LDH levels [12]. The Swiss COVID-19 lung study evaluated pulmonary functions and radiological ndings at four months of discharge and found that impaired DLCO was the strongest factor associated with previous severe to critical disease [19]. In this study, days of ICU admission, older age, and LDH level did not correlate with post-COVID-19 persistent parenchymal lung changes, in contrast to what was observed in MERS-CoV [12]. Moreover, we found that prolonged viral shedding beyond 14 days had no direct impact on the development of persistent parenchymal lung changes after acute COVID-19 infection.

Limitations
It is important to acknowledge the limitations of our study, which includes its retrospective design and the limited number of patients. We did not perform chest CT for all control cases; however, clinical resolution of symptoms mirrors radiological changes resolution based on previous studies [21].

Conclusion
In this study, persistent parenchymal lung changes were observed in a number of patients 10 weeks after acute COVID-19 infection. We found that only receiving invasive mechanical ventilation and a NLR of > 3.13 at admission, are strong risk factors for persistent parenchymal lung changes. We also showed that neither the clinical nor the initial radiological severity of the acute illness predict the outcome of patients with COVID-19, and that no medication received during the acute illness altered the disease course. None of the medications received during the acute phase of the illness altered the disease course.