Associations of D-Dimer on Admission and Clinical Features of COVID-19 Patients: A Systematic Review, Meta-Analysis, and Meta-Regression

Abstract

Background: Dynamic D-dimer level is a key biomarker for the severity and mortality of COVID-19 (coronavirus disease 2019). How aberrant fibrinolysis influences the clinical progression of COVID-19 presents a clinicopathological dilemma challenging intensivists.

Methods: We performed meta-analysis and meta regression to analyze the associations of plasma D-dimer with 106 clinical variables to identify a panoramic view of the derangements of fibrinolysis in 14,862 patients of 42 studies. There were no limitations of age, gender, race, and country. Raw data of each group were extracted separately by two investigators. Individual data of case series, median and interquartile range, and ranges of median or mean were converted to SDM (standard deviation of mean).

Findings: The weighted mean difference of D-dimer was 0.97 µg/mL (95% CI 0.65, 1.29) between mild and severe groups, as shown by meta-analysis. Publication bias was significant. Meta-regression identified 58 of 106 clinical variables were associated with plasma D-dimer levels. Of these, 11 readouts were negatively related to the level of plasma D-dimer. Further, age and gender were confounding factors. There were 22 variables independently correlated with the D-dimer level, including respiratory rate, dyspnea plasma K⁺, glucose, SpO2, BUN (blood urea nitrogen), bilirubin, ALT (alanine aminotransferase), AST (aspartate aminotransferase), systolic blood pressure, and CK (creatine kinase).

Interpretation: These findings support elevated D-dimer as an independent predictor for both mortality and complications. The identified D-dimer-associated clinical variables draw a landscape integrating the aggregate effects of systemically suppressive and pulmonary hyperactive derangements of fibrinolysis, and the D-dimer-associated clinical biomarkers, and conceptually parameters could be combined for risk stratification, potentially for tracking thrombolytic therapy or alternative interventions.

Keywords: COVID-19; D-dimer; comorbidity; fibrinogenolysis; fibrinolysis; meta-regression.

Figure 1

PRISMA 2009 flow diagram for searching included studies.

Figure 2

Random-effects meta-analysis of D-dimer level. (A) Forest plot. We selected studies that had subgroups, which could be divided into mild (including normal) and severe groups. We pooled weighted mean differences (WMD, black diamond, and gray square) and 95% CI (horizontal lines through the diamonds) of D-dimer from eligible 23 studies. Studies with only one group or moderate group were excluded. The red diamond represents the overall WMD. (B) Egger’s publication bias plot. N=89, P<0.001. (C) Filled funnel plot. P<0.001. Circle, raw data; square, pseudo data needed for symmetric distribution.

Figure 3

Meta-regressions of plasma D-dimer level on admission with some clinical variables. Dashed vertical red lines indicate the normal range.

Figure 4

Correlation of D-dimer elevations and aberrant fibrin deposition with organ dysfunction. (A) D-dimer-associated clinical parameters sorted by the system and clinical relevance. (B) Outcomes correlate with D-dimer levels based on our hypothesis.

Figure 5

Schematic mechanisms for the dynamic D-dimer level in COVID-19 patients. (A) Regulation of the fibrinolysis system. The half lifetime for crucial components is given in followed brackets. (B) Clinicopathological mechanisms for elevated plasma D-dimer in COVID-19 patients.