Research ArticleCORONAVIRUS

Increased complement activation is a distinctive feature of severe SARS-CoV-2 infection

See allHide authors and affiliations

Science Immunology  13 May 2021:
Vol. 6, Issue 59, eabh2259
DOI: 10.1126/sciimmunol.abh2259
  • Fig. 1 Markers of complement activation are higher in COVID-19 compared with non–COVID-19 respiratory failure.

    Plasma for determination of circulating markers of complement activation was obtained in patients with COVID-19 (n = 134) and influenza (n = 54) at Barnes-Jewish Hospital (BJH)/WUSM. (A) CONSORT flow diagram showing patient enrollment, allocation, and outcomes in the COVID-19 cohort. The CONSORT diagram for the influenza and non-COVID acute respiratory failure cohorts are in fig. S1. Violin plots of differences in sC5b-9 between (B) the influenza (EDFLU, n = 54) and COVID-19 (n = 124) cohorts, (C) the non-COVID acute respiratory failure (IPS, n = 22) and COVID-19 cohorts (n = 124), and (D) restricting both the cohorts from (C) to those who died (non-COVID, IPS, n = 8 versus COVID-19, n = 30). Statistical significance is determined using Mann-Whitney U test. ER, emergency room.

  • Fig. 2 Complement activation is associated with worse outcomes in COVID-19 in two independent cohorts.

    Markers of complement activation were quantified in the plasma at WUSM and Yale University School of Medicine (Yale). Violin plots of sC5b-9 levels in the WUSM COVID-19 cohort in (A) patients requiring ICU admission (n = 72) versus those who did not (n = 62), (B) patients requiring IMV (n = 29) versus those who did not (n = 105), and (C) patients who died (n = 30) versus those who survived (n = 104). (D) A linear regression line shows the relationship between plasma levels of sC5b-9 and C5a. The spline chart demonstrates the mean with 95% confidence intervals (CIs). Coefficient of determination (R2) represents the goodness of fit. The degree of correlation is assessed using Spearman’s rank correlation coefficient test (ρ = 0.4909; 95% CI, 0.2321 to 0.6848; N = 48). In the Yale cross-sectional cohort, concurrently measured C5a levels are used to compare (E) patients requiring ICU admission (n = 40) versus those who did not (n = 9), and (F) patients requiring IMV (n = 26) versus those who did not (n = 23). Statistical significance is determined using Mann-Whitney U test.

  • Fig. 3 AP activation is associated with worse outcomes in COVID-19.

    Comparisons in the levels of components involved in the AP in plasma of patients requiring ICU admission (n = 26) versus those who did not (n = 22), in the WUSM COVID-19 cohort, are presented using violin plots: (A) iC3b:C3 ratio, (B) Factor B, and (D) Ba. (C) A linear regression line shows the relationship between plasma levels of sC5b-9 and Factor B. The spline chart demonstrates the mean with 95% CI. R2 represents the goodness of fit. The degree of correlation is assessed using Spearman’s rank correlation coefficient test (ρ = 0.4768; 95% CI, 0.2146 to 0.6749; n = 48). (E) Plasma Ba levels are compared in patients who survived [1301.0 (966.0 to 2250.0), n = 29] versus those who did not [3266 (2368 to 6236), n = 19], as are the plasma levels of Factor D (F). Statistical significance is determined using Mann-Whitney U test.

  • Fig. 4 Complement activation is associated with markers of endothelial injury and a prothrombotic state in patients with COVID-19.

    A linear regression line shows the relationship between plasma levels of Factor D and (A) Ang2, (B) thrombomodulin, and (C) vWF:Ag in the Yale cross-sectional cohort. The spline chart demonstrates the mean with 95% CI. R2 represents the goodness of fit. The degree of correlation is assessed using Spearman’s rank correlation coefficient test between Factor D and (i) Ang2 (ρ = 0.5095; 95% CI, 0.2585 to 0.6960; n = 49), (ii) thrombomodulin (ρ = 0.6050; 95% CI, 0.3829 to 0.7609; n = 49), and (iii) vWF:Ag (ρ = 0.3367; 95% CI, 0.04612 to 0.5747; n = 47). Violin plots are used for comparing the levels of (D) Ang2, (E) thrombomodulin, and (F) vWF:Ag in plasma of patients requiring IMV versus those who did not. Statistical significance is determined using Mann-Whitney U test.

  • Table 1 Demographic characteristics of the cohorts.

    COVID-19, participants with SARS-CoV-2 infection; CS, cross-sectional; IPS, Immunity in Pneumonia and Sepsis; LT, longitudinal.

    COVID (n = 134)EDFLU (n = 54)IPS (n = 22)Yale, LT cohort (n = 23)Yale, CS cohort
    (n = 49)
    Demographics
    Age in years, mean ± SD63 ± 1653 ± 1754 ± 1765 ± 1263 ± 17
    Gender
    Female41% (55)50% (27)50% (11)39% (9)33% (16)
    Male59% (79)50% (27)50% (11)61% (14)67% (33)
    Ethnicity
    Black or African-
    American
    79.1% (106)54.5% (12)26% (6)24.5% (12)
    White19.4% (26)40.9% (9)52% (12)51% (25)
    Other1.5% (2)4.5% (1)21% (5)*24.5% (12)*
    Clinical characteristics
    Hospital admission92.5% (124)96.3% (52)100.0% (22)100% (23)100% (49)
    ICU admission53.7% (72)24.1% (13)100.0% (22)61% (14)82% (40)
    IMV21.6% (29)11.1% (6)100.0% (22)9% (2)53% (26)
    In-hospital mortality22.4% (30)3.7% (2)36.4% (8)30% (7)24.5 (12)
    Comorbidities
    Smoking history46.3% (62)63.6% (14)65% (15)8% (4)
    Chronic lung disease19.4% (26)22.2% (12)13.6% (3)35% (8)10% (5)
    End-stage renal disease
    (ESRD)
    6% (8)7.4% (4)4.5% (1)4% (1)4% (2)
    Diabetes mellitus, type 252.2% (70)35.2% (19)18.2% (4)39% (9)27% (13)

    *Hispanic/Latino, Asian, and unknown.

    †Current or former smoker. Smoking history was unavailable for the EDFLU cohort.

    ‡Based on chronic dialysis requirement.

    • Table 2 Complement analytes in the WUSM COVID-19 cohort.

      Statistical tests for comparison were done using the Mann-Whitney U test. Values are represented as medians (interquartile range).

      Non-ICU
      (n = 62)
      ICU (n = 72)P
      sC5b-9 (ng/ml)559.5
      (343.3–813.0)
      715.4
      (448.5–1084.0)
      0.0335
      C5a (pg/ml)*635.0
      (471.9–892.6)
      918.4
      (666.7–1081)
      0.034
      iC3b:C3 ratio*0.56 (0.51–0.65)0.70 (0.57–1.39)0.002
      Factor B, ng/ml*21,606
      (17,834–26,853)
      25,840
      (20,544–32,832)
      0.033
      AP hemolytic
      activity, %
      85.0
      (76.0–102.5)
      81.0 (74.5–90.0)0.2519
      Ba, ng/ml*1191
      (901.3–1981)
      3112
      (2022–6612)
      <0.0001
      Factor D, ng/ml*4640.3
      (3659–9887.3)
      6622.5
      (4308–10,854.1)
      0.166

      *C5a, iC3b:C3 ratio, Factor B, Ba, and Factor D were measured in 48 patients, among whom 26 needed an ICU admission and 22 did not.

      †AP hemolytic activity was performed in 38 patients (21 ICU and 17 non-ICU) on the basis of the availability of samples.

      • Table 3 Complement analytes in patients with COVID-19 in the Yale School of Medicine longitudinal cohort.

        Statistical tests for comparison were done using the Mann-Whitney U test. All samples were drawn within 24 hours of hospital admission. Values are represented as medians (interquartile range).

        Non-ICU
        (n = 14)
        ICU (n = 9)P
        C5a, pg/ml43.2 (43.2–43.2)77.6
        (43.2–285.6)
        0.0016
        Factor D, ng/ml1442
        (1234–1803)
        1825
        (1541–2576)
        0.07
      • Table 4 Markers of complement activation, endothelial injury, and coagulation in the Yale School of Medicine cross-sectional cohort.

        Statistical tests for comparison were done using the Mann-Whitney U test. Values are represented as medians (interquartile range).

        Non-IMV (n = 23)IMV (n = 26)P
        C5a, pg/ml263.8 (225.5–848)475.6 (317.9–1353.0)0.017
        Factor D, ng/ml4605 (3721–6187)6437 (3445–9674)0.09
        Ang2, ng/ml4077 (2149–7633)11,470 (6711–15,103)<0.0001
        Thrombomodulin, ng/ml2.9 (1.9–4.5)5.0 (3.0–8.1)0.0068
        vWF:Ag, %375.0 (266.0–559.0)*558.5 (409.8–685.3)0.0063

        *Samples for measuring vWF:Ag were available in 21 of 23 patients who did not need IMV.

        Supplementary Materials

        Stay Connected to Science Immunology

        Navigate This Article