BQC19 is a province-wide biobank established in Quebec, Canada, in March 2020 to foster collaborations on COVID-19 research by collecting, storing, and sharing of samples and data of people affected by COVID-19 (see bqc19.ca) (25). Participants were recruited upon contact with the healthcare system in the emergency room or during acute care hospitalization between March 2020 and August 2021. Participants were eligible to be enrolled to BQC19 if they had undergone PCR-testing for SARS-CoV-2. Participants with a positive PCR result were enrolled in the COVID+ group, and those with a negative test served as controls. Upon informed consent confirmation, medical charts containing comorbidity data were extracted, questionnaire were administered to participants and blood was collected at several timepoints. Clinical outcomes were assessed for all participants.

We analyzed proteomics and clinical data from 903 adults diagnosed with COVID-19 in the acute phase of infection, and 295 controls. All controls had negative COVID-19 test results. Controls were hospitalized for various reasons including COVID-19 negative respiratory infections, cardiovascular events, including embolic diagnoses, digestive symptoms or uncontrolled diabetes. For each patient, we considered the first available measurements of ACBP, i.e., the nearest in time to symptom onset. None among the participants was vaccinated against COVID-19 at the time of sample collection.

All participants gave written, oral, or substituted consent to participate in BQC-19. The Management framework of BQC-19 (including provision for oral or substituted consent) was approved by the center de recherche du centre hospitalier de l’université de Montréal (CR-CHUM) Research ethics board, and the present project was approved by the centre universitaire de santé McGill (CUSM) research ethics board.

The WHO Working Group on the Clinical Characterization and Management of COVID-19 criteria were used to categorize participants into mild, moderate or severe infections (26). Mild disease referred to participants requiring ambulatory care (emergency room visits without hospitalization), including people with asymptomatic presentation but detectable SARS-CoV-2 RT-PCR, or symptomatic but requiring little or no assistance. Moderate diseases encompass hospitalized patients who did not require oxygen therapy or received oxygen by mask or nasal cannula. Severe disease refers to hospitalized patients requiring oxygen delivered by positive airways pressure, intubation or mechanical ventilation, dialysis or extracorporeal membrane oxygenation. Participants who passed away during their hospitalization were included in a separate group when indicated.

Whole blood was obtained through venipuncture using acid-citrate-dextrose vacutainer tubes, and plasma was separated by centrifugation at 750g, 10 min at room temperature. Isolated plasma was aliquoted and stored at −80°C until analysis.

Blood samples from a total of 1198 BQC19 participants were included for the plasma proteomics analysis. Proteomic profiles were assessed at SomaLogic using the SomaScan v4.0 proteomic platform. This platform provides measurements on 4701 unique human circulating proteins using 4987 Slow Off-Rate Modified Aptamers (SOMAmer reagents) and quantifies protein levels in the form of relative fluorescence units (RFUs) (27). Experimental process and data normalization including hybridization control normalization, intraplate median signal normalization, and plate scaling and calibration were performed as described (27–29).

In a subset of samples (n=33), plasma was obtained for ACBP levels quantification using a commercial kit following the supplier recommendations (Abnova, ELISA Kit cat. #KA6327, Taiwan).

SARS-CoV2 Spike-specific IgG levels were quantified by means of an ELISA and a cell-based ELISA (CBE) method as previously described (30, 31, 60). For the CBE, HOS cells were transfected to express SARS-CoV-2 spike proteins. Transfected cells were washed and incubated with diluted plasma (1:250). After washing, anti-human IgG, IgM and IgA antibody coupled to horseradish peroxidase (HRP) was added. After washing again, substrate was added, and light emission was quantified using a luminometer.

GraphPad Prism 9.0 (GraphPad, CA, USA) was used to perform group comparisons and correlation analyses. Spearman’s rank correlation test was used to identify associations between 2 continuous variables. Mann-Whitney’s test and student t-test were used to compare levels of continuous variables between two independent groups, as appropriate. Kruskal-Wallis’ test, also known as one-way ANOVA test, was used to compare levels of continuous variables in more than 2 independent study groups. Distribution comparisons between COVID-19 patients and controls were performed using Chi-square tests. P-values <0.001 were considered significant for samples with n>250, and <0.05 for samples less than 250. A minimum value of 0.2 was considered for r values in significant associations assessed with the Spearman’s test. Logistic regression univariable models were used to generate Receiver operating characteristic (ROC) curves. SPSS (IBM, Chicago, IL, USA) was used for multivariable analyses.

The BQC19 was approved by the research ethics boards (REB) of the Jewish General Hospital and the Centre Hospitalier de l’Université de Montréal (CHUM) in Montréal, QC, Canada. Informed consent was obtained from all participants. Data analyses for this project were approved by the research ethics board of the McGill University Health Centre (MUHC) from 2021 to 2024 (REB approval # 2021-7241).

A total of 1198 participants were included in the analysis. Among these participants, 903 were COVID-19 positive as determined by RT-qPCR tests (median age 59.5, range 18-99), encompassing 46% females and 54% males. In addition, 295 RT-qPCR COVID-19 negative hospitalized participants were included as controls (median age 49, range 20-89), encompassing 47% females and 53% males. Using WHO criteria (26), of the COVID-19 positive cohort (n=903), 259 were classified as mild, 384 as moderate, 236 as severe, and 24 as fatal. Slight differences in the frequency of participants with diabetes and cancer were observed between the COVID-19 and control groups. Details are included in Table 1 .

Approximately 5000 plasma proteins were measured by SomaScan proteomics in COVID-19 infected and hospitalized control adult participants. Plasma ACBP levels were similar between COVID-19-positive participants and hospitalized controls ( Figure 1A ). However, in the COVID-19 positive group, plasma ACBP levels increased with disease severity, and the highest median ACBP levels were detected in the fatal group (p<0.0001) ( Figure 1B ). Severe COVID-19 groups also had significantly higher plasma ACBP levels compared to moderate and mild COVID-19 groups (p<0.0001 for both comparisons). Both severe and fatal groups had higher plasma ACBP levels than hospitalized controls (p<0.001 and 0.0043).

We were able to quantify plasma ACBP by ELISA and found a strong correlation between ACBP levels detected by SomaScan proteomics and ELISA (r=0.61, p<0.001) in 33 samples ( Supplementary Figure 1 ). Also, we compared plasma ACBP levels assessed by ELISA in 33 COVID-19 patients, including 25 severe cases and 12 healthy controls (demographics detailed in Supplementary Table 1 ). We found higher levels of ACBP in severe COVID-19 than healthy controls (p=0.04) ( Supplementary Figure 1 ).

Plasma ACBP concentrations were associated with age (r=0.29, p<0.001), with levels of GDF-15, a biomarker of aging (r=0.29, p<0.0001) ( Table 2 ) (28), as well as with ferritin levels, which is an established marker of COVID-19 severity (r=0.3, p<0.001) ( Table 2 ) (7). Multivariable analysis revealed that age, sex or the presence of comorbidities did not influence the association between ACBP levels and disease severity ( Table 3 ). We evaluated the predictive ability of plasma ACBP to discriminate mild from severe COVID-19 states (including fatal cases) using receiver operating characteristic (ROC) curves ( Figure 1C ). The area under the curve (AUC) was estimated for ACBP levels and their predicted values by fitting regression models. Plasma ACBP concentration levels predicted COVID-19 disease severity with an AUC=0.79 ± 0.026 (p<0.0001).

All other 5000 plasma proteins were compared between patients with mild and severe/fatal disease presentation using multiple Mann-Whitney’s tests. ACBP was found as one of the 712 proteins best associated with severity as shown by being one of the proteins with the most significant q and p values (-log₁₀ 118.2; p<0.00001). Of note, the Q-values were less significant for IL-6 or CRP (-log₁₀ 35.81 and 108.3, respectively, p<0.00001 for both), which are two common biomarkers of COVID-19 severity (4, 32) ( Supplementary Table 2 ).

Two waves of COVID-19 cases were observed during the study period, the first between March 2020 and August 2020 and the second between September 2020 and May 2021 in Quebec, Canada (https://www.inspq.qc.ca/en/node/34836) (33). The first was dominated by B1.1 ancestral subtypes of SARS-CoV-2 while the second one was predominantly due to the α variant of SARS-CoV-2 (33). Interestingly, plasma ACBP levels were more elevated in the plasma of COVID-19 patients collected during the second wave as compared to the first one (8591 vs. 5999 RFU/50µL, p<0.0001) ( Figure 1D ).

COVID-19 patients were older during the first wave compared to the second one (66.9 vs. 57.4, p<0.001) ( Supplementary Figure 2 ). This appears important because normal aging (in apparently healthy individuals) is associated with an increase in ACBP levels (15, 34). However, multivariable analyses showed that the variation in plasma ACBP concentrations was independent of age, comorbidities and sex in the two collection periods.

We then assessed whether comorbidities were associated with further elevation of ACBP levels in COVID-19 patients. The most common comorbidities observed in COVID-19 infected patients (n=903) were cardiovascular disease (CVD) (n=483, 53.4%), diabetes (244, 27.0%), cancer (16, 1.8%), obesity (89, 9.9%), chronic obstructive pulmonary disease (COPD) (84, 9.3%) and infection by human immunodeficiency virus (HIV) (7; 0.77%) ( Table 1 ).

Plasma ACBP concentrations correlated with age, a known co-factor of COVID-19 severity, in both COVID-19-positive patients and COVID-19-negative controls (r=0.29, p<0.001 and r=0.28, p<0.001 respectively) ( Figure 2A ). SARS-CoV-2-infected patients with comorbidities had higher plasma ACBP levels compared to those without known comorbidities ( Figure 2B ). Thus, COVID-19 patients with diabetes, obesity, CVD, or COPD had significantly higher plasma ACBP levels (p<0.001 for all comparisons) than patients without any of these comorbidities ( Figure 2C ). Although ACBP levels are elevated in people with HIV (PWH) (24), we observed similar ACBP levels in PWH with COVID-19 and HIV-uninfected COVID-19 patients. As a caveat, only 7 PWH participants were included in the present study.

In COVID-negative hospitalized controls, patients with cardiovascular diseases, diabetes or cancer had higher levels of ACBP compared to those without these condition ( Supplementary Figure 3 ).

We then assessed whether inflammation would be associated with plasma ACBP levels. Plasma ACBP concentrations correlated with the elevated neutrophil/lymphocyte ratio (r=0.29, p<0.0001), which is an established marker of inflammation and severity (35, 36). In contrast, neither CD4 counts nor CD8 counts, which were measured in a subset of participants, were associated with circulating ACBP levels ( Table 4 ).

Of note, circulating ACBP levels were strongly associated with markers of innate inflammation, such as C-reactive protein (CRP) and interleukin (IL)-6, but not with IL-1β or IL-8 ( Table 4 ). However, plasma ACBP levels inversely correlated with several markers of lymphoid immune responses including TNF-α and IFN-γ. Moreover, ACBP anticorrelated with IL-21, which is participates to the crosstalk between CD4 and CD8 T cells and promotes antibody maturation and secretion (37) ( Table 4 ).

In conclusion, ACBP levels correlate with surrogate markers of inflammation and anticorrelate with markers of T cell-mediated immune responses.

To assess whether circulating ACBP levels were associated with anti-SARS-CoV-2 immune response, we compared ACBP levels with two measures of SARS-CoV-2 specific antibodies in plasma. Participants were enrolled early during their infection, at the time of initial healthcare contact. Despite this early time point, most participants (61.6%) already possessed circulating antibodies against the spike protein (see details in Supplementary Figure 4 ).

When all seropositive and seronegative participants were considered, plasma ACBP concentrations correlated positively with the titers of spike-specific IgA and IgM ( Table 5 ; Supplementary Figure 5 ), as well as with circulating Spike-RBD antibody levels detected by several methods such as ELISA and antibody binding to Spike expressing cells.

Interestingly, among SARS-CoV-2-seropositive patients, only the circulating levels of anti-RBD total IgG were associated with plasma ACBP levels ( Table 6 ). All correlations were statistically significant when they were computed for the relationship between plasma ACBP and IgG, IgA, IgM, as well as for total antibodies capable of binding to spike expressing cells.

Extracellular ACBP is a well-known inhibitor of autophagy (38, 39). Autophagy is a tightly regulated pathway involving multiple factors from the autophagy protein (ATG) family. We compared circulating levels of the autophagy inhibitor ACBP, with autophagy involved proteins detected by SomaScan proteomics such as the ATG proteins (ATG3, ATG4B, ATG5, ATG7) and other established effectors of autophagy such as Beclin-1 (BECN1), galectins 3 and 9 (40), as well as with the lipidated form of microtubule-associated proteins 1A/1B light chain 3B (LC3II) ( Table 7 ). Plasma ACBP concentrations positively correlated with ATG 3, ATG5 and LC3II, but negatively correlated with ATG4B, BECN1 and galectin 8 in participants with COVID-19.

Altogether, these results suggest a link between the COVID-19 severity-related elevation of ACBP and autophagy in circulating leukocytes.