Among 143 COVID-19 patients attending follow-up visits at the outpatient clinic, we included a total of 110 who agreed to participate in the current study. The most frequently affected pulmonary function parameter reported during the follow-up course was DLCO impairment. Among the selected biomarkers, suPAR showed the highest correlation with DLCO measurements, and a low cut-off value for suPAR exhibited the highest NPV value for DLCO impairment.
The FEV1/FVC ratio is typically used to assess airflow obstruction in the lungs: a normal FEV1/FVC ratio is usually greater than 0.70. A ratio of less than 0.70 is often used to define airflow limitation. In this study, the ratio was on average higher than 0.70, for both DLCO <80% and >80%. When a reduced DLCO is caused by chronic airflow limitation, such as in emphysema, it will be associated with a low FEV1/FVC, whereas a low DLCO caused by interstitial lung abnormalities will be associated with a normal or high FEV1/FVC ratio. This, combined with the low prevalence of COPD in the cohort, indicates that the likely cause of DLCO impairment was interstitial abnormalities due to COVID-19.
Of the 110 COVID-19 patients that were evaluated, we found a low correlation between baseline measured biomarkers and impaired pulmonary functional capacity assessed by DLCO measurement 4-5 months post discharge at the respiratory outpatient clinic. Wide confidence intervals complicate precise interpretation, making it challenging to determine correlation magnitudes, with some estimates significantly enhanced or reduced in effect. When predicting DLCO impairment using biomarker cut-off values, the low cut-off exhibited the highest sensitivity but low specificity, while the high cut-off had low sensitivity and high specificity. This could be attributed to the large number of patients with biomarker values falling within the middle category, which represents the difference between high and low cut-offs. This suggests that the cut-off values are effective at identifying either the most impaired individuals (low cut-off) or the most non-impaired individuals (high cut-off), but are less effective in identifying patients in the middle group. Both PPV and NPV values for most cut-offs hover around 50% suggesting that at least half of these allocations are misclassifications.
The integration of all included biomarkers yielded the most substantial improvement, resulting in an NPV of 0.93. This suggests that nearly all non-impaired patients are accurately classified. This could indeed assist physicians in reducing the number of patients requiring follow-up in the clinical setting. However, the specificity remained low at 0.41, indicating that less than half of all non-impaired patients are correctly identified.
A follow-up study by Fortini et al,2021, investigated symptoms and pulmonary alterations in COVID-19 survivors (n=105) from 3 to 6-months post-discharge (21). This study revealed persisting diffusing capacity alterations in the lungs, which is consistent with the findings of our own study. The DLCO impairment compromised gas diffusion between alveoli and pulmonary capillaries, possibly due to epithelial and endothelial pulmonal damage or a damaged intra -alveolar diffusion pathway. In clinical practice, reduced DLCO values are often observed in conditions involving extensive interstitial fibrosis or vascular obstruction. Low DLCO levels may indicate various conditions, including diffuse parenchymal pulmonal disease, reduced vasculature and alveolar tissue (as seen in chronic obstructive pulmonary disease, COPD), or pulmonary vascular diseases such as pulmonary embolism, chronic thromboembolic pulmonary hypertension, or idiopathic pulmonary arterial hypertension) (22,23). Histologic findings in the lungs have shown alveolar damage with hyaline membrane formation, alongside microthrombi in small pulmonary vessels (24). We therefore hypothesize that COVID-19 patients could experience a course characterized by a combination of interstitial pneumonia with oedema and micro/ macro-thrombosis (25). This finding further supports the notion that DLCO is a particularly sensitive tool for monitoring COVID-19 patients during follow-up.
The association between the biomarker suPAR and asthma, as well as suPAR and COPD patients, regarding readmission rate and mortality, had already been described before the COVID-19 pandemic (26,27). Asthma is not normally associated with a reduction in DLCO, and the low prevalence of COPD (6.4%) in the present study, makes it less plausible that the reduction in pulmonary function parameters, especially DLCO, is due to chronic pulmonary conditions, such as COPD. Therefore, COVID-19 could indeed be responsible for the clinical impairments observed in the study.
A double-blinded, randomized controlled phase 3 trial, involving COVID-19 patients with suPAR level above 6 ng/ml who were randomized to receive anakinra (IL-1 receptor antagonist) treatment or a placebo, showed protection towards COVID-19 progression among the patients treated with anakinra. Whether anakinra also mitigates long-term decline in DLCO and other pulmonary function parameters is currently unknown (28).
In the early phase of the first wave of the COVID-19 pandemic, the clinical outcomes such as the development of respiratory failure and the need for mechanical ventilation among COVID-19 patients with varying suPAR cut-off values were addressed (13,20). Our previous studies further explored whether COVID-19 patients developed respiratory failure requiring treatment with mechanical ventilation (13), and examined different disease trajectories, ranging from mild to very severe (20). Since long-term COVID-19 (post-Covid-19 or PASC) is a relatively new group of health problems, research is needed to elucidate this area and provide physicians with a better understanding of the molecular pathophysiology. Therefore, in the current study, we investigated three independent biomarkers, CRP, leucocyte cell count and suPAR, to determine if they could predict long-term pulmonary function impairment as part of long-COVID-19 among survivors. Such knowledge can enable researchers to predict long-term impairment in patients’ health and can aid in designing a more comprehensive follow-up course including potential treatment regimens (6–9). Early intervention could be expected to help patients in managing their impairments and sequelae. However, whether the treatment will lead to a complete curative outcome remains to be elucidated.
Our study has both strengths and limitations that need to be addressed. In terms of strengths, the study is prospective, including patients across a range of severity levels, and comprehensively evaluates various aspects of pulmonary function tests.
The lack of pre-hospitalization information regarding pulmonary function should also be mentioned. The absence of DLCO measurements for patients who died before the follow-up visit introduced an immortal time bias into the study population. All predictive estimates rely on patients surviving until the follow-up time, a factor that cannot be determined at baseline. At least, it is important to mention that suPAR measurements were not part of the standard laboratory work at the ED on initial admission. Due to the high workload during the COVID-19 pandemic, suPAR measurements were unfortunately only conducted for those patients we were able to reach out to.
It can be speculated that the measured biomarkers from the admission index may be influenced by differences in illness severity within the study population, particularly in terms of who would develop pulmonary functional impairment and experience impaired DLCO values at follow-up. Another study analyzed DLCO (% predicted) at follow-up and found no correlation with CRP levels measured upon admission (29).
Since there were no previous pulmonary function tests conducted among the hospitalized patients in this study, preexisting DLCO impairment might have already been present in some of the COVID-19 patients. Without baseline DLCO measurements before the onset of COVID-19, it becomes challenging to directly attribute the reduction in DLCO solely to COVID-19. Several factors could contribute to decreased DLCO in patients presenting with COVID-19, including pre-existing lung conditions, smoking history, age-related changes in lung function, and other concurrent illnesses or medications. However, despite the lack of pre-COVID-19 DLCO measurements, certain aspects of the clinical presentation and disease progression suggest a potential association between COVID-19 and reduced DLCO. These aspects include the temporal relationship, the absence of other known causes, and consistency with existing literature on the association between COVID-19 and decreased DLCO.
The results of the study have not been evaluated in an external sample to verify if a similar performance is reproducible or only applicable to the study sample.