TYPE Brief Research Report PUBLISHED 08 August 2022 DOI 10.3389/fmed.2022.973918 Krebs von den Lungen-6 OPEN ACCESS glycoprotein circulating levels are not useful as prognostic EDITED BY Félix Gutiérrez, Miguel Hernández University of Elche, Spain REVIEWED BY marker in COVID-19 Victor Asensi, University of Oviedo, Spain Francesc Vidal, pneumonia: A large prospective University of Rovira i Virgili, Spain *CORRESPONDENCE cohort study Ivan Castellví icastellvi@santpau.cat SPECIALTY SECTION Ivan Castellví1*, Diego Castillo2 , Hèctor Corominas1 , This article was submitted to Anaís Mariscal3 , Sandra Orozco2 , Natividad Benito4 , Infectious Diseases - Surveillance, Prevention and Treatment, Virginia Pomar4 , Andrés Baucells3 , Isabel Mur4 , a section of the journal Frontiers in Medicine David de la Rosa-Carrillo2 , David Lobo3 , Ana Milena Millan3 , RECEIVED 20 June 2022 Nerea Hernández de Sosa5 , David Filella5 , Laia Matas5 , ACCEPTED 19 July 2022 Laura Martínez-Martínez3 , Cándido Juarez3 , PUBLISHED 08 August 2022 CITATION Jordi Casademont5 and Pere Domingo4 Castellví I, Castillo D, Corominas H, 1 Department of Rheumatology and Systemic Autoimmune Diseases, Hospital de la Santa Creu i Sant Mariscal A, Orozco S, Benito N, Pau, Barcelona, Spain, 2 Department of Pneumology, Hospital Universitari de la Santa Creu i Sant Pomar V, Baucells A, Mur I, de la Pau, Barcelona, Spain, 3 Department of Immunology, Hospital Universitari de la Santa Creu i Sant Rosa-Carrillo D, Lobo D, Millan AM, Pau, Barcelona, Spain, 4 Division of Infectious Diseases, Hospital Universitari de la Santa Creu i Sant Hernández de Sosa N, Filella D, Pau, Barcelona, Spain, 5 Department of Internal Medicine, Hospital de la Santa Creu i Sant Pau, Matas L, Martínez-Martínez L, Barcelona, Spain Juarez C, Casademont J and Domingo P (2022) Krebs von den Lungen-6 glycoprotein circulating levels are not useful as prognostic Coronavirus disease 2019 (COVID-19) has rapidly expanded worldwide. marker in COVID-19 pneumonia: A Currently, there are no biomarkers to predict respiratory worsening in patients large prospective cohort study. Front. Med. 9:973918. with mild to moderate COVID-19 pneumonia. Small studies explored the use doi: 10.3389/fmed.2022.973918 of Krebs von de Lungen-6 circulating serum levels (sKL-6) as a prognostic COPYRIGHT biomarker of the worsening of COVID-19 pneumonia. We aimed at a large © 2022 Castellví, Castillo, Corominas, study to determine the prognostic value of sKL-6 in predicting evolving trends Mariscal, Orozco, Benito, Pomar, Baucells, Mur, de la Rosa-Carrillo, in COVID-19. We prospectively analyzed the characteristics of 836 patients Lobo, Millan, Hernández de Sosa, with COVID-19 with mild lung disease on admission. sKL-6 was obtained in Filella, Matas, Martínez-Martínez, Juarez, Casademont and Domingo. all patients at least at baseline and compared among patients with or without This is an open-access article respiratory worsening. The receiver operating characteristic curve was used to distributed under the terms of the find the optimal cutoff level. A total of 159 (19%) patients developed respiratory Creative Commons Attribution License (CC BY). The use, distribution or worsening during hospitalization. Baseline sKL-6 levels were not higher in reproduction in other forums is patients who had respiratory worsening (median {IQR} 315.5 {209–469} vs. 306 permitted, provided the original author(s) and the copyright owner(s) {214–423} U/ml p = 0.38). The last sKL-6 and the change between baseline are credited and that the original and last sKL-6 were higher in the respiratory worsening group (p = 0.02 and p publication in this journal is cited, in < 0.0001, respectively). The best sKL-6 cutoff point for respiratory worsening accordance with accepted academic practice. No use, distribution or was 497 U/ml (area under the curve 0.52; 23% sensitivity and 85% specificity). reproduction is permitted which does sKL-6 was not found to be an independent predictor of respiratory worsening. not comply with these terms. Frontiers in Medicine 01 frontiersin.org Castellví et al. 10.3389/fmed.2022.973918 A conditional inference tree (CTREE) was not useful to discriminate patients at risk of worsening. We found that sKL-6 had a low sensibility to predict respiratory worsening in patients with mild-moderate COVID-19 pneumonia and may not be of use to assess the risk of present respiratory worsening in inpatients with COVID-19 pneumonia. KEYWORDS COVID-19, Krebs von den Lungen-6 (KL-6), pneumonia, biomarker, predictor Introduction Methods Since severe acute respiratory syndrome coronavirus 2 Study population (SARS-CoV-2) was identified as the cause of coronavirus disease 2019 (COVID-19) in Wuhan, China at the end of A total of 876 patients, admitted consecutively to a tertiary December 2019 (1), it has expanded rapidly, causing a global university hospital in the COVID-19 emergency context, were pandemic with more than 400 million confirmed cases and included in a prospective, observational, single-center study. a death toll of more than five million people, according All patients had a SARS-CoV-2 infection, laboratory-confirmed to the World Health Organization (1) (https://covid19.who. by a real-time reverse transcription-polymerase chain reaction int). (PCR) test using nasopharyngeal or oropharyngeal swabs, and The clinical manifestations of COVID-19 are variable. had mild to moderate COVID-19 pneumonia without baseline Most affected patients show mild respiratory symptoms needs for intensive care unit (ICU) admittance, high oxygen without special healthcare requirements (2, 3). Nevertheless, flux supplementation, or noninvasive ventilation. The clinical some patients require hospital admission due to respiratory status of our admitted patients was assessed using the WHO failure caused by severe pneumonia (4). Several therapies, eight-category ordinal scale (29): (1) not hospitalized and no including corticoids, Janus kinase inhibitors, interleukin-6 limitations of activities; (2) not hospitalized, with limitation of receptor antagonists, and antivirals (remdesivir, molnupiravir, activities, home oxygen requirement, or both; (3) hospitalized, and nirmatrelvir), improve the course of the disease (5), not requiring supplemental oxygen, and no longer requiring but only vaccines have demonstrated the strength required ongoing medical care (used if hospitalization was extended for to modify this scenario (6). Despite this, SARS-CoV-2 infection-control or other nonmedical reasons); (4) hospitalized infection can still lead to adult respiratory distress syndrome and not requiring supplemental oxygen but requiring ongoing (ARDS), especially in immunocompromised patients (7). medical care (related to COVID-19 or to other medical Different poor prognosis factors have been identified, such conditions); (5) hospitalized and requiring any supplemental as age, male sex, chronic comorbidities, and laboratory oxygen; (6) hospitalized, requiring noninvasive ventilation, or parameters (8–14). use of high-flow oxygen supply; (7) hospitalized, receiving The mucin-like glycoprotein Krebs von den Lungen-6 (KL- invasive mechanical ventilation, or extracorporeal membrane 6) is primarily expressed on the surface of type II epithelial oxygenation (ECMO); and (8) death. Patients with a clinical alveolar cells when the cell is damaged or regenerating (13). status corresponding to categories 4 or 5 at admission were KL-6 has been proposed as a potentially useful biomarker included in this study. in Interstitial Lung Diseases (ILDs) (15–20). sKL-6 has also All patients were admitted between 14 March 2020 been described as a biomarker for pulmonary damage in and 7 February 2021. Most patients (55.3%) were recruited ARDS since higher levels correlate well with mean and during the first and second COVID-19 waves and received peak airway pressure (21). The usefulness of the sKL-6 hydroxychloroquine 400 mg/12 h (day 1), hydroxychloroquine level as a prognostic biomarker in COVID-19 has been 200 mg/12 h (days 2–5), and azithromycin 500 mg/day (days suggested in previous cohort studies but their interpretation 1–3), according to our standard of care at that time point. is limited due to the sample size (from 5 to 364 cases) (22– In patients with cytokine release syndrome (CRS), defined 28). as respiratory function worsening and high or progressively Our study aimed to assess the potential of sKL- increasing D-dimer (>1,500 ng/ml) or high IL-6 levels (>40 6 as a biomarker of lung disease progression in a pg/ml), a single dose of intravenous tocilizumab (600 mg for large cohort of COVID-19 cases and to determine its patients ≥75 kg; 400 mg for those < 75 kg) was prescribed prognostic value. following Spanish Agency for Medicinal Products and Medical Frontiers in Medicine 02 frontiersin.org Castellví et al. 10.3389/fmed.2022.973918 Devices (AEMPS) recommendations (https://www.aemps. not meet the assumptions of normality as evaluated using the gob.es/la-aemps/ultima-informacion-de-la-aemps-acerca- Kolmogorov-Smirnov test. Nominal and ordinal categorical del-covid%e2%80%9119/tratamientos-disponibles-para-el- data were summarized in terms of the number of subjects manejo-de-la-infeccion-respiratoria-por-SARS-CoV-2/?lang= providing data at the relevant time point (n), frequency counts, en). Furthermore, all patients received a prophylactic low dose and percentages presented to one decimal place. Differences of molecular weight heparin unless otherwise contraindicated. between groups for nominal data were identified using the Oral consent was obtained from patients due to the pandemic Fisher-exact test. The p-value between groups for ordinal emergency situation, in line with our observational study design. data was estimated using the exact Mantel-Haenszel test. The The ethics committee of the Hospital Universitari de la Santa sensitivity, specificity, positive predictive value (PPV), negative Creu i Sant Pau approved this study (IIBSP-COV-2020-35). predictive value (NPV), accuracy, and error rate of all KL-6 values for respiratory worsening were calculated by the area under the receiver operating characteristic (ROC) curve (AUC). Data collection The optimal cutoff KL-6 value was obtained using the Youden’s index. Independent predictive factors of respiratory worsening At least one serum sample for KL-6 was obtained from were obtained using the multivariate logistic regression model. each patient during hospital admission. In patients with long- The final model selects predictive factors as those independent standing admission, more KL-6 samples were collected based variables with a p < 0.05 on the Wald test. The odds ratio on the criteria of the attending physician. Respiratory function (OR) and the 95% confidence interval (95% CI) were calculated worsening was considered if patients changed to a worse with the exponential model coefficients. The predictive accuracy category of the WHO eight-category ordinal scale during of the logistic regression model was evaluated using the hospitalization. The PaO2 /FiO2 (PAFI) ratio was estimated classification table, where observed values for the dependent according to the method developed by Brown et al. (30, 31). outcome and the predicted values (at a defined cutoff value, Demographic data, preexisting chronic medical conditions e.g., p = 0.50) were cross-classified, showing for each predicted (including presence or absence of lung diseases and/or value the % of correct events, % correct nonevents, sensitivity, connective tissue diseases), previous use of potential toxic lung specificity, PPV, and NPV. Time to worsening from admission drugs, clinical symptoms at hospitalization, length of symptoms, and from the first symptom was defined as the time from length of hospital stay, and clinical and laboratory (complete admission (or first symptom) until the date of worsening. blood count, renal and liver function, C-reactive protein, D- Patients without worsening were censored at the hospital dimer, and ferritin) outcomes during hospitalization, including discharge date. The median days of time free of worsening, the time to worsening since KL-6 serum extraction, of the need with its 95%CI, were estimated using the Kaplan-Meier (KM) for invasive mechanical ventilation, ICU admission, length of survival analysis and using KL-6 optimal cutoff point, and the hospital stay, and in-hospital death, were collected from each statistical significance between groups was performed using the patient. We analyzed all patients with at least one sKL-6 and log-rank test. Prognostic factors of time (days) to worsening performed an additional analysis in patients with serial sKL- were obtained using the multivariate cox regression model, with 6 levels. all potential prognostic factors, including KL-6 optimal cutoff, as independent variables. Thereafter, independent variables reaching a p < 0.05 were included in the final adjusted model. KL-6 assay The adjusted median days of time free of worsening, with its 95%CI, were estimated using the KL-6 optimal cutoff point. sKL-6 levels were determined in serum samples using the The p-value and the 95% confidence interval for the hazard chemiluminescence reagent “Lumipulse G KL-6” (Fujirebio ratio (HR) were estimated based on the Wald test. Additionally, Europe NV, Gent, Belgium) following the manufacturer’s unbiased conditional inference trees (CTREE) were fitted to instructions and expressed in U/ml. analyze the associations between worsening and all covariates at baseline (Initial visit 0), including demographic data, previous pathologies, clinical signs, and analytic data. CTREE is a Statistical analysis conditional recursive partitioning algorithm that solves both the overfitting problem and the variable selection bias present in Continuous or interval data are summarized in terms of other recursive partitioning algorithms (32). All the variables are the number of observations, mean and standard deviation (SD), potential candidates to be included in the model. The variable median, minimum, maximum, and interquartile range (IQR). selection process is automated, and no assumptions regarding Differences in continuous variables between groups (worsening the underlying structure and distribution are needed. As a result, vs. not worsening) were performed using a Student’s t-test the tree shows the variables in a hierarchical structure of the or Mann-Whitney U test (nonparametric) when the data did model that has a relative real weight for taking decisions. Frontiers in Medicine 03 frontiersin.org Castellví et al. 10.3389/fmed.2022.973918 A p < 0.05 was considered statistically significant. Baseline sKL-6 circulating levels and R All report outputs were produced using SAS version 9.4 respiratory function worsening (TS1M5) in a secure and validated environment. Copyright (c) 2016 by SAS Institute Inc., Cary, NC, USA. There was no difference in baseline sKL-6 levels in patients with or without respiratory function worsening (median {IQR} 315.5 {209–469} vs. 306 {214–423} U/ml, p = 0.39) and last sKL- 6 values increased significantly in both groups (p < 0.0001). Results Nevertheless, delta (1) sKL-6 (median 95 {29–259} vs. 36 {5– 113} U/ml, p < 0.0001) and the last sKL-6 levels were higher in Of 876 patients, 836 patients (487 men) met the inclusion patients that worsened (median {IQR} 414 {292–683.5} vs. 364 criteria. Their median age was 56 (IQR, 46–64) years, and {250–530} U/ml, p = 0.027). 273 (66.1%) were older than 50 years of age. Of 836 patients, The best cutoff baseline sKL-6 level discriminating patients 461 patients (55.1%) had at least one comorbidity, with with or without impending respiratory function worsening was hypertension (31.9%) and dyslipidemia (24.9%) being the 497 U/ml [AUC 0.52; 23% sensitivity, 85% specificity; 26% PPV; most frequent. Also, 267 patients (31.4%) had a history of and 82% NPV; (Figure 1)]. Patients with sKL-6≥ 497 U/ml tobacco use, and 152 (15.8%) had previous lung diseases. had a 32% higher risk of respiratory function worsening since The use of drugs with potential lung toxicity (chemotherapy, admission (p = 0.15; Figure 2). The worsening likelihood did not immunosuppressants, and angiotensin-converting enzyme increase when symptoms onset was considered. inhibitors) was documented in 17 (2%) patients. The most frequent presenting symptom was fever (n = 645, 77.2%), followed by cough (n = 544, 65.1%). All patients had Peak sKL-6 levels and respiratory ill-defined consolidations on chest X-rays, but only 400 function worsening (47.8%) reported dyspnea, while 50 (6.0%) had oxygen saturation by pulse oximeter (pSatO2) < 93% on their first Considering that sKL-6 increased in all patients and the high clinical exam. Patient baseline characteristics are summarized number of serial samples, we calculated the best cutoff point in Table 1. in the peak (max) sKL-6 level to improve the AUC. The best The median time between symptoms and hospitalization cutoff point of the peak sKL-6 level to discriminate patients was was 7 (IQR, 5-10) days, and the median hospital stay 408 U/ml (AUC 0.56; 46% sensitivity; 65% specificity; 24% PPV; was 6 (IQR, 4-9) days. All patients were admitted 84% NPV). between the first and third COVID-19 waves in Spain, A peak sKL-6 level > 408 U/ml was not associated with and national treatment protocols changed during a higher likelihood of respiratory function worsening. Patients that period. In total, 405 (48.4%) and 422 (50.5%) with a peak sKL-6≥ 408 U/ml had a 30% higher risk of patients received hydroxychloroquine and azithromycin, respiratory function worsening from admission (p = 0.098), but respectively. Also, 169 patients (20.4%) received a trend in the likelihood of increased worsening was found when tocilizumab, and corticosteroids were prescribed in 232 symptom onset was considered (p = 0.067). patients (27.8%). A total of 159 patients (19%) developed respiratory function worsening during their hospital stay. Also, 17 patients (2.0%) Use of sKL-6 in CTREE algorithm died, and 13 patients were in the worsening respiratory group. The four deaths in the nonworsening respiratory group were Figure 3 shows the CTREE algorithm according to the related to previous cancer complications (50%) and aspirative prevalence of worsening. The most important variables pneumonia (50%). associated with impending worsening were mentioned in decreasing order of importance: PAFI (p < 0.001), C-reactive protein (CRP) (p < 0.001), days from disease onset (p = 0.004), age (p = 0.005), and platelet count (p = 0.029). sKL-6 and peak Serum KL-6 concentrations sKL-6 level were not identified as prognostic factors for disease worsening. A total of 1,952 samples of sKL-6 from 836 patients were collected. All patients provided at least one sKL-6 sample. Notably, 419, 288, 207, 153, and 49 patients had two, three, four, Discussion five, or six sKL-6 samples, respectively. Median baseline sKL-6 was 307 (IQR, 211-430; range was 15.8–6,295 U/ml), and the last We analyzed the value of sKL-6 circulating levels as sKL-6 values were 368 (IQR, 254-542; range 1.5–2,202 U/ml). predictors of respiratory worsening in a large population Frontiers in Medicine 04 frontiersin.org Castellví et al. 10.3389/fmed.2022.973918 TABLE 1 Patient characteristics at hospital admission. Respiratory worsening during Yes No Total p hospitalization n (%) 159 (19) 677 (81) 836 (100) Female n (%) 61 (38.4) 288 (42.5) 171 (45.6) ns Age (median [IQR]) 59 [53–67] 56 [44–63] 56 [46–64] 0.0001 Comorbidities n (%) 100 (62.9) 361 (53.3) 461 (55.1) 0.034 Hypertension 60 (37.7) 207 (30.6) 267 (31.9) 0.09 Dyslipidemia 47 (29.6) 178 (26.3) 225 (26.9) ns Diabetes 20 (12.6) 99 (14.6) 119 (14.3) ns Lung disease 33 (20.8) 99 (14.6) 132 (15.8) 0.069 SAD 10 (6.3) 660 (97.5) 809 (96.8) 0.023 Lung Toxics Exposure n (%) Tobacco 43 (27) 168 (24.8) 211 (25.2) ns Drugs 8 (5) 9 (1.3) 17 (2) 0.0074 Symptoms n (%) Fever (≥37.3◦ C) 129 (81.1) 516 (76.2) 645 (77.2) ns Dyspnea 92 (57.9) 308 (45.5) 400 (47.8) 0.0061 Cough 96 (60.4) 448 (66.2) 544 (65.1) ns O2 Sat <93% n (%) 21 (13.2) 29 (4.3) 50 (6) 0.0001 % O2 Sat (median [IQR]) 95 [94–96] 96 [95–97] 96 [95–97] <0.0001 P/F Ratio (median [IQR])mmHg 304 [271–338] 329[295–375] 326 [290–367] <0.0001 Laboratory parameters Leucocytes, 109 /L 7.4 ± 4.3 6.7 ± 3.6 6.8 ± 3.7 0.0408 Lymphocytes,109 /L 1.1 ± 0.7 1.2 ± 1.7 1.2 ± 1.4 0.0078 Platelets, 109 /L 199 ± 91.5 214 ± 90.3 211.2 ± 90.7 0.0068 CRP, mg/L 113.5 ± 79.5 80.8 ± 72.1 87 ± 74.6 <0.0001 LDH, U/L 357.8 ± 136.5 339.3 ± 662.4 342.8 ± 600.3 <0.0001 ALT, U/L 42.5 ± 29.1 43.8 ± 48.1 43.6 ± 45.1 ns AST, U/L 43.2 ± 23.6 41.7 ± 42.5 42 ± 39.7 0.0014 GGT, U/L 90.1 ± 108.4 80.8 ± 91.1 82.8 ± 94.9 ns D–dimer, mg/L 1,673 ± 6,661.6 1,335.7 ± 7,030 1,398.2 ± 6,959.7 ns Ferritin, ug/L 1,174.5 ± 1,341.6 815.2 ± 908 882.1 ± 1,009.8 0.041 sKL−6 (median[IQR]) 315.5 [209–469] 306[214–423] 307 [211–430] ns IQR, interquartile range; SAD, systemic autoimmune disease; O2 Sat, oxygen saturation; P/F ratio, arterial oxygen partial pressure to fractional inspired oxygen ratio; CRP, C- reactive protein; LDH, lactate dehydrogenase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyltransferase; sKL-6, serum Krebs von den Lungen-6. of hospitalized patients with mild COVID-19. We did KL-6 and sKL-6 have recently emerged as potential not find differences between baseline sKL-6 levels in biomarkers in ILDs, based on their specificity translating patients with or without respiratory impairment during damage and regeneration in type II pneumocytes (33). It is their hospital stay, and we did not find baseline or peak well known that the viral infection of alveolar epithelial cells by sKL-6 to be a reliable biomarker of respiratory function SARS-CoV-2 is a crucial step in the development of the disease worsening in patients with mild COVID-19. However, we (34). We hypothesized that injured type II pneumocytes express did observe that the last sKL-6 values increased significantly excess KL-6, which spills over into systemic circulation due to in both groups and that this increase was higher in patients the increased permeability of the alveolar-capillary membrane. that worsened. A correlation between lung injury and sKL-6 levels has been Considering the dramatic worldwide impact of the COVID- demonstrated in ILDs where pneumocyte and alveolar-capillary 19 pandemic on healthcare systems, it is imperative to find a membrane damage is observed (15–20, 35–39). Similar findings useful tool to discriminate which patients will worsen before have been reported in 28 patients with ARDS who had higher a flare occurs. This is especially crucial in a period of the median sKL-6 levels than nine ventilated controls of matched pandemic, given the pressures of resource allocation. illness severity and 10 healthy donors (21). Frontiers in Medicine 05 frontiersin.org Castellví et al. 10.3389/fmed.2022.973918 FIGURE 1 Receiver operating characteristic (ROC) curve of Krebs von de Lungen-6 circulating serum levels (sKL-6) and respiratory worsening (n 835). Studies that analyze the effect of sKL-6 in patients with sKL-6 determinations (26). Interestingly, they found a statistical COVID-19 are scarce, and most have limitations in their association between baseline, peak, and 1 sKL-6 levels with interpretation of findings due to several factors. A recent meta- severe illness and survival. We also found differences in peak analysis of seven studies that analyzed sKL-6 levels in severe and 1 sKL-6, but we did not find different baseline sKL-6 levels and nonsevere patients supported the potential role of sKL-6 between groups, though we did not analyze survival in our circulating levels as predictors of severe COVID-19 (40). But cohort. The ROC curve did not allow us to identify a reliable those meta-analyses were based on small retrospective studies sKL-6 threshold to predict the worsening of COVID-19. There with high heterogenicity. Our prospective observational study are several differences between the Yamaya study and ours. includes more patients than all the studies analyzed in that meta- Unlike us, Yamaya et al. analyzed data retrospectively, and their analysis together, and this factor may be critical for a correct hospital did not admit patients in critical condition. This fact interpretation of the findings. may have biased their survival findings and ROC results. In our experience, neither the best cutoff point of baseline d’Alessandro et al. analyzed the peripheral natural killer cells sKL-6 nor peak sKL-6 during hospitalization shows an and sKL-6 as potential prognostic biomarkers of COVID-19 acceptable AUC that would allow us to consider KL-6 as a severity in 22 patients (24). They found higher sKL-6 levels in potential biomarker for predicting respiratory deterioration in patients with severe COVID-19 compared with patients with patients with mild to moderate COVID-19. A few studies have mild infection or a healthy group. The best cutoff point of looked at the sensitivity, specificity, and predictive values of sKL- 406.5 U/ml in sKL-6 was identified using the ROC analysis 6 in COVID-19. Yamaya et al. found a sensitivity of 83.3% and a (83% sensitivity and 89% specificity) in their study. This study specificity of 90.5 (AUC 0.89) to predict respiratory worsening has some similarities with ours, notably a similar cutoff point. with a sKL-6 cutoff point of 684 U/ml in 356 patients with Nevertheless, they did not detail the timeline of patients’ Frontiers in Medicine 06 frontiersin.org Castellví et al. 10.3389/fmed.2022.973918 FIGURE 2 Free survival of respiratory worsening since admission (baseline sKL-6). worsening, and baseline sKL-6 may therefore have been taken unbiased conditional inference trees to analyze sKL-6 circulating in patients who had already worsened. We collected baseline levels to predict respiratory worsening. CTREE solves both the samples in nonseverely ill patients before respiratory function overfitting problem and the variable selection bias present in worsened to assess the usefulness of sKL-6 as a predictor of other recursive partitioning algorithms (32). To date, only a evolving trends in patients. Finally, another retrospective study small study with 24 patients has explored a combination model compared different biomarkers in patients with 33 nonsevere with IL-6, sKL-6, and CRP values to detect patients with severe and 21 severe COVID-19 and found higher sKL-6 levels in the COVID-19 with promising results (43). Our more exact model severe group with a peak level 1 week after diagnosis (41). With does not support the use of sKL-6 in algorithm models. an optimal cutoff value of 371 U/ml, they found a sensitivity and There are a number of factors that may explain the failure specificity of 85.7 and 96.6%, respectively (AUC 0.84) to assess of sKL-6 to predict respiratory function worsening. Different severity in patients with COVID-19. In our prospective study subtypes of COVID-19 pneumonia have been proposed (44), with more patients (159 who worsened and 677 who did not), we and the kind of lung damage and subsequent sKL-6 production observed opposite results (sensitivity, 46%, specificity 65%, PPV can be different in those subtypes (23, 45). An example of this 24%, and NPV 84%; AUC 0.56), though with a similar threshold is provided by patients with COVID-19 pneumonia type H peak of sKL-6 (408 U/ml). More studies have analyzed the (high elastance, high right-to-left shunt, high lung weight, and sensitivity and specificity of sKL-6 in COVID-19 (24, 25, 28, 42) high lung recruitability). sKL-6 could be useful as a potential but their main goals were different than ours since they analyzed biomarker of pulmonary worsening in comparison with type the levels of this biomarker to predict irreversible ILD or death. L (low elastance, low ventilation-to-perfusion ratio, low lung We did not explore sKL-6 levels and their usefulness to predict weight, and low lung recruitability) COVID-19 pneumonia (23). outcomes after COVID-19 pneumonia. Unfortunately, we have no tools to know which pneumonia Our study did not demonstrate the utility of sKL-6 in subtype is existent in our patients. a composite algorithm to predict respiratory impairment in Our study has some limitations and weaknesses. First, we COVID-19. Our CTREE, which includes sKL-6 optimal values, do not have a control group of patients with balanced risk did not find sufficient discriminatory power to predict which factors, and we therefore only analyzed sKL-6 in hospitalized patients will worsen. This is the first study that includes patients with mild COVID-19 pneumonia on admission. The Frontiers in Medicine 07 frontiersin.org Castellví et al. 10.3389/fmed.2022.973918 FIGURE 3 Unbiased conditional inference trees (CTREE) with baseline sKL-6. absence of a control group does not allow us to determine in patients with COVID-19 pneumonia and the effect of sKL- whether sKL-6 could play a role in detecting patients before 6 to predict lung fibrosis after hospitalization are warranted to clinical pneumonia is present. During the outbreak, we had sKL- elucidate its role as a potential biomarker for further clinical and 6 reactive limitations in Europe, which meant that we focused therapeutic decisions. our study fieldwork on a significant sample of patients affected with mild to moderate pneumonia. Second, some patients may present a nonpreviously diagnosed ILD or have other diseases Data availability statement associated with abnormal sKL-6 levels not related to COVID- 19 pneumonia. Another limitation was the fact that we did The raw data supporting the conclusions of this article will not study the correlation between sKL-6 levels and imaging be made available by the authors, without undue reservation. tools (chest X-ray and computerized tomography) or other laboratory data. A major issue sometimes observed in patients with COVID-19 is that the cause of lung function worsening Ethics statement is related to thromboembolic phenomena (3% of patients with The studies involving human participants were reviewed and COVID-19 in our cohort), which may not directly affect KL-6 approved by Comité etico de investigación científica (CEIC), production and eventually spills over into circulation (46). Hospital de la Santa Creu i Sant Pau, Barcelona, Spain (IIBSP- Finally, we were unable to analyze the survival rates of our COV-2020-35). Written informed consent for participation was patients. These analyses were not possible due to the low rate of not required for this study in accordance with the national deaths found in our cohort. legislation and the institutional requirements. To summarize, our findings suggest that baseline sKL-6 levels do not predict respiratory worsening in patients with mild to moderate COVID-19 pneumonia. sKL-6 levels are not reliable Author contributions enough to be used as a prognostic screening tool on admission to stratify which patients will likely need more intensive respiratory Conceiving the research and manuscript drafting: IC, DC, care. Future controlled trials with serial determinations of sKL-6 HC, and PD. Patients recruitment: IC, SO, and DL. KL-6 Frontiers in Medicine 08 frontiersin.org Castellví et al. 10.3389/fmed.2022.973918 handling: AMa, AB, and LM. Data analysis: IC and PD. All secretaries, and all auxiliary personnel who fought and are still authors read and approved the final manuscript. fighting COVID-19 in our center. Funding Conflict of interest This study was funded by a COVID-PoC BioCAT grant The authors declare that the research was conducted in the from the Health Department of the Catalan Government, absence of any commercial or financial relationships that could Generalitat de Catalunya and supported by the Institut be construed as a potential conflict of interest. de Recerca de l’Hospital de la Santa Creu i Sant Pau, Barcelona, Spain. This study was partially supported either by the grant COV20/00070, Instituto de Salud Carlos III, Publisher’s note Madrid, Spain. All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated Acknowledgments organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or The authors would like to acknowledge the efforts of claim that may be made by its manufacturer, is not guaranteed all the physicians, fellows, nurses, technicians, investigators, or endorsed by the publisher. References 1. Zhou P, Yang XL, Wang XG, Hu B, Zhang L, Zhang W, et al. A pneumonia 12. Yao Y, Cao J, Wang Q, Shi Q, Liu K, Luo Z, et al. D-dimer as a biomarker outbreak associated with a new coronavirus of probable bat origin. Nature. (2020) for disease severity and mortality in COVID-19 patients: a case control study. J 579:270–3. doi: 10.1038/s41586-020-2951-z Intensive Care. (2020) 8:49. doi: 10.1186/s40560-020-00466-z 2. Epidemiology Working Group for Ncip Epidemic Response CCfDC, 13. Huang I, Pranata R, Lim MA, Oehadian A, Alisjahbana B. C-reactive Prevention. The epidemiological characteristics of an outbreak of 2019 novel protein, procalcitonin, D-dimer, and ferritin in severe coronavirus disease- coronavirus diseases (COVID-19) in China. Zhonghua Liu Xing Bing Xue Za Zhi. 2019: a meta-analysis. Ther Adv Respir Dis. (2020) 14:1753466620937175. (2020) 41:145–51. doi: 10.3760cma.j.issn.0254-6450.2020.02.003 doi: 10.1177/1753466620937175 3. Huang C, Wang Y, Li X, Ren L, Zhao J, Hu Y, et al. Clinical features of 14. Munoz-Rodriguez JR, Gomez-Romero FJ, Perez-Ortiz JM, Lopez-Juarez P, patients infected with 2019 novel coronavirus in Wuhan, China. Lancet. (2020) Santiago JL, Serrano-Oviedo L, et al. Characteristics and risk factors associated with 395:497–506. doi: 10.1016/S0140-6736(20)30183-5 mortality in a multicenter spanish cohort of patients with COVID-19 pneumonia. Arch Bronconeumol. (2021) 57:34–41. doi: 10.1016/j.arbres.2021.02.021 4. Golpe R, Blanco N, Castro-Anon O, Corredoira J, Garcia-Pais MJ. Perez-de-Llano LA, et al. Factors associated to hospital admission in a 15. Kuwana M, Shirai Y, Takeuchi T. Elevated serum Krebs von den Lungen- care protocol in COVID-19. Arch Bronconeumol (Engl Ed). (2020) 56:676–7. 6 in early disease predicts subsequent deterioration of pulmonary function in doi: 10.1016/j.arbr.2020.05.009 patients with systemic sclerosis and interstitial lung disease. J Rheumatol. (2016) 43:1825–31. doi: 10.3899/jrheum.160339 5. Drug treatments for covid-19: living systematic review and network meta- analysis. BMJ. (2021) 373:n967. doi: 10.1136/bmj.n967 16. Kobayashi N, Takezaki S, Kobayashi I, Iwata N, Mori M, Nagai K, et al. Clinical and laboratory features of fatal rapidly progressive interstitial lung 6. Butt AA, Yan P, Shaikh OS, Mayr FB, Omer SB. Rate and risk factors disease associated with juvenile dermatomyositis. Rheumatology (Oxford). (2015) for severe/critical disease among fully vaccinated persons with breakthrough 54:784–91. doi: 10.1093/rheumatology/keu385 SARS-CoV-2 infection in a high-risk national population. Clin Infect Dis. (2021) 10:ciab1023. doi: 10.1093/cid/ciab1023 17. Yokoyama A, Kondo K, Nakajima M, Matsushima T, Takahashi T, Nishimura M, et al. Prognostic value of circulating KL-6 in idiopathic pulmonary fibrosis. 7. Estella A, Canton ML, Munoz L, Higueras IR, Recuerda Nunez M, Tejero Respirology. (2006) 11:164–8. doi: 10.1111/j.1440-1843.2006.00834.x Aranguren J, et al. Vaccinated patients admitted in ICU with severe pneumonia due to SARS-CoV-2: a multicenter pilot study. J Pers Med. (2021) 11:1086. 18. Yokoyama A, Kohno N, Hamada H, Sakatani M, Ueda E, Kondo doi: 10.3390/jpm11111086 K, et al. Circulating KL-6 predicts the outcome of rapidly progressive idiopathic pulmonary fibrosis. Am J Respir Crit Care Med. (1998) 158:1680–4. 8. Tian W, Jiang W, Yao J, Nicholson CJ Li RH, Sigurslid HH, et al. Predictors doi: 10.1164/ajrccm.158.5.9803115 of mortality in hospitalized COVID-19 patients: a systematic review and meta- analysis. J Med Virol. (2020). doi: 10.1002/jmv.26050 19. Sokai A, Tanizawa K, Handa T, Kanatani K, Kubo T, Ikezoe K, et al. Importance of serial changes in biomarkers in idiopathic pulmonary fibrosis. ERJ 9. Guo W, Li M, Dong Y, Zhou H, Zhang Z, Tian C, et al. Diabetes is a risk factor Open Res. (2017) 3:00019-2016. doi: 10.1183/23120541.00019-2016 for the progression and prognosis of COVID-19. Diabetes Metab Res Rev. (2020) 36:e3319. doi: 10.1002/dmrr.3319 20. Hamai K, Iwamoto H, Ishikawa N, Horimasu Y, Masuda T, Miyamoto S, et al. Comparative study of circulating MMP-7, CCL18, KL-6, SP-A, and SP-D as disease 10. Wang D, Li R, Wang J, Jiang Q, Gao C, Yang J, et al. Correlation analysis markers of idiopathic pulmonary fibrosis. Dis Markers. (2016) 2016:4759040. between disease severity and clinical and biochemical characteristics of 143 cases doi: 10.1155/2016/4759040 of COVID-19 in Wuhan, China: a descriptive study. BMC Infect Dis. (2020) 20:519. doi: 10.1186/s12879-020-05242-w 21. Sato H, Callister ME, Mumby S, Quinlan GJ, Welsh KI. duBois RM, et al. KL-6 levels are elevated in plasma from patients with acute respiratory distress 11. Bao C, Tao X, Cui W, Yi B, Pan T, Young KH, et al. syndrome. Eur Respir J. (2004) 23:142–5. doi: 10.1183/09031936.03.00070303 SARS-CoV-2 induced thrombocytopenia as an important biomarker significantly correlated with abnormal coagulation function, increased 22. Xue M, Zheng P, Bian X, Huang Z, Huang H, Zeng Y, et al. Exploration intravascular blood clot risk and mortality in COVID-19 patients. and correlation analysis of changes in Krebs von den Lungen-6 levels in Exp Hematol Oncol. (2020) 9:16. doi: 10.1186/s40164-020-0 COVID-19 patients with different types in China. Biosci Trends. (2020). 0172-4 doi: 10.5582/bst.2020.03197 Frontiers in Medicine 09 frontiersin.org Castellví et al. 10.3389/fmed.2022.973918 23. Nakamura H, Miyagi K, Otsuki M, Higure Y, Nishiyama N, Kinjo T, et al. patients with interstitial lung disease. Am Rev Respir Dis. (1993) 148:637–42. Serum KL-6 can distinguish between different phenotypes of severe COVID-19. J doi: 10.1164/ajrccm/148.3.637 Med Virol. (2020). doi: 10.1002/jmv.26268 36. Elhai M, Avouac J, Allanore Y. Circulating lung biomarkers 24. d’Alessandro M, Cameli P, Refini RM, Bergantini L, Alonzi V, Lanzarone N, in idiopathic lung fibrosis and interstitial lung diseases associated et al. Serum KL-6 concentrations as a novel biomarker of severe COVID-19. J Med with connective tissue diseases: where do we stand? Semin Virol. (2020) 92:1902–14. doi: 10.1002/jmv.26087 Arthritis Rheum. (2020) 50:480–91. doi: 10.1016/j.semarthrit.2020. 01.006 25. Xue M, Zhang T, Chen H, Zeng Y, Lin R, Zhen Y, et al. Krebs Von den Lungen-6 as a predictive indicator for the risk of secondary pulmonary fibrosis 37. Yamakawa H, Hagiwara E, Kitamura H, Yamanaka Y, Ikeda S, Sekine and its reversibility in COVID-19 patients. Int J Biol Sci. (2021) 17:1565–73. A, et al. Serum KL-6 and surfactant protein-D as monitoring and predictive doi: 10.7150/ijbs.58825 markers of interstitial lung disease in patients with systemic sclerosis and mixed connective tissue disease. J Thorac Dis. (2017) 9:362–71. doi: 10.21037/jtd.201 26. Yamaya T, Hagiwara E, Baba T, Kitayama T, Murohashi K, Higa K, et al. 7.02.48 Serum Krebs von den Lungen-6 levels are associated with mortality and severity in patients with coronavirus disease 2019. Respir Investig. (2021) 59:596–601. 38. Ma H, Lu J, Song Y, Wang H, Yin S. The value of serum Krebs von doi: 10.1016/j.resinv.2021.04.002 den lungen-6 as a diagnostic marker in connective tissue disease associated with interstitial lung disease. BMC Pulm Med. (2020) 20:6. doi: 10.1186/s12890-019- 27. Deng K, Fan Q, Yang Y, Deng X, He R, Tan Y, et al. Prognostic roles of KL-6 in 1043-z disease severity and lung injury in COVID-19 patients: a longitudinal retrospective analysis. J Med Virol. (2021) 93:2505–12. doi: 10.1002/jmv.26793 39. Kim HC, Choi KH, Jacob J, Song JW. Prognostic role of blood KL-6 in rheumatoid arthritis-associated interstitial lung disease. PLoS ONE. (2020) 28. Chen H, Qin R, Huang Z, Luo W, Zheng P, Huang H, et al. Clinical relevance 15:e0229997. doi: 10.1371/journal.pone.0229997 of serum Krebs von den Lungen-6 levels in patients with coronavirus disease 2019. Cytokine. (2021) 148:155513. doi: 10.1016/j.cyto.2021.155513 40. Naderi N, Rahimzadeh M. Krebs von den Lungen-6 (KL-6) as a clinical marker for severe COVID-19: a systematic review and meta-analyses. Virology. 29. Beigel JH, Tomashek KM, Dodd LE. Remdesivir for the treatment (2022) 566:106–13. doi: 10.1016/j.virol.2021.11.006 of Covid-19 - preliminary report. Reply N Engl J Med. (2020) 383:994. doi: 10.1056/NEJMoa2007764 41. Awano N, Inomata M, Kuse N, Tone M, Takada K, Muto Y, et al. Serum KL-6 level is a useful biomarker for evaluating the severity of coronavirus disease 2019. 30. Brown SM, Grissom CK, Moss M, Rice TW, Schoenfeld D, Hou Respir Investig. (2020) 58:440–7. doi: 10.1016/j.resinv.2020.07.004 PC, et al. Nonlinear imputation of Pao2/Fio2 from Spo2/Fio2 among patients with acute respiratory distress syndrome. Chest. (2016) 150:307–13. 42. Scotto R, Pinchera B, Perna F, Atripaldi L, Giaccone A, Sequino D, et doi: 10.1016/j.chest.2016.01.003 al. Serum KL-6 could represent a reliable indicator of unfavourable outcome in patients with COVID-19 pneumonia. Int J Environ Res Public Health. (2021) 31. Brown SM, Duggal A, Hou PC, Tidswell M, Khan A, Exline M, et 18:18:2078. doi: 10.3390/ijerph18042078 al. Nonlinear imputation of PaO2/FIO2 from SpO2/FIO2 among mechanically ventilated patients in the ICU: a prospective, observational study. Crit Care Med. 43. Bergantini L, Bargagli E. d’Alessandro M, Refini RM, Cameli P, Galasso (2017) 45:1317–24. doi: 10.1097/CCM.0000000000002514 L, et al. Prognostic bioindicators in severe COVID-19 patients. Cytokine. (2021) 141:155455. doi: 10.1016/j.cyto.2021.155455 32. Strobl C, Boulesteix AL, Zeileis A, Hothorn T. Bias in random forest variable importance measures: illustrations, sources and a solution. BMC Bioinformatics. 44. Gattinoni L, Chiumello D, Caironi P, Busana M, Romitti F, Brazzi L, et al. (2007) 8:25. doi: 10.1186/1471-2105-8-25 COVID-19 pneumonia: different respiratory treatments for different phenotypes? Intensive Care Med. (2020) 46:1099–102. doi: 10.1007/s00134-020-06033-2 33. Ishikawa N, Hattori N, Yokoyama A, Kohno N. Utility of KL-6/MUC1 in the clinical management of interstitial lung diseases. Respir Investig. (2012) 50:3–13. 45. Japan EfC. Nationwide system to centralize decisions around ECMO use for doi: 10.1016/j.resinv.2012.02.001 severe COVID-19 pneumonia in Japan (Special Correspondence). J Intensive Care. (2020) 8:29. doi: 10.1186/s40560-020-00445-4 34. Domingo P, Mur I, Pomar V, Corominas H, Casademont J, de Benito N. The four horsemen of a viral Apocalypse: the pathogenesis of SARS-CoV-2 infection 46. Benito N, Filella D, Mateo J, Fortuna AM, Gutierrez-Alliende JE, (COVID-19). EBioMedicine. (2020) 58:102887. doi: 10.1016/j.ebiom.2020.102887 Hernandez N, et al. Pulmonary thrombosis or embolism in a large cohort 35. Kohno N, Awaya Y, Oyama T, Yamakido M, Akiyama M, Inoue Y, of hospitalized patients with Covid-19. Front Med (Lausanne). (2020) 7:557. et al. KL-6, a mucin-like glycoprotein, in bronchoalveolar lavage fluid from doi: 10.3389/fmed.2020.00557 Frontiers in Medicine 10 frontiersin.org