Elevated von Willebrand Factor Antigen Is an Early Predictor of Mortality and Prolonged Length of Stay for Coronavirus Disease 2019 (COVID-19) Inpatients

The study was conducted in 2020 within a large integrated health care delivery organization spanning 15 medical centers across Southern California (see Supplemental Table 1 for a list of locations; see supplemental digital content at https://meridian.allenpress.com/aplm in the January 2022 table of contents). It was approved by the organization's Institutional Review Board as project 12546.

Residual frozen citrated plasma samples collected from COVID-19 inpatients for D-dimer measurement were thawed and tested for vWF:Ag (HemosIL vWF:Ag reagent kit, ACL TOP 750 coagulation analyzer, Instrumentation Laboratory, Bedford, Massachusetts). These were “samples of convenience,” the collection of which was predicated upon COVID-19 clinical protocols and perceived risk of thrombosis. Most COVID-19 inpatients in our system had D-dimer measured upon admission and whenever their condition appeared to worsen.

Automated review of the electronic medical record was used to extract hospitalization dates and laboratory results; imaging studies, pulmonary parameters, and intensive care unit admissions were retrieved by the authors. Patient outcomes were classified according to National Institutes of Health (NIH) COVID-19 Treatment Guideline Panel¹¹  recommendations, with minor modifications. The “severe” category was subdivided based on need for supplemental oxygen at discharge. Thrombotic events (pulmonary embolism, deep venous thrombosis) were added to those criteria which elevated a patient's outcome from “severe” to “critical.” Patients dying within 30 days of discharge were classified as COVID-19–related deaths.

Peak measurements for vWF:Ag and the other APRs were defined simply as the highest measurement in a given patient's record. Outcome severity trends against APR peak measurements were detected by Jonckheere-Terpstra testing. Binary logistic regression was used to determine risk of death and prolonged (>4 days) length of stay (LOS). Data were analyzed using SAS version 9.4 (SAS Institute, Cary, North Carolina). A P value less than .05 (P < .05) was considered significant.

From 127 patients admitted for COVID-19, 342 samples were collected for D-dimer measurement and were later assayed by us for vWF:Ag. Eight samples were excluded because of clots, hemolysis, icterus, or lipemia, and 1 was excluded because the patient did not have active COVID-19. The remaining 120 patients included 67 male (56%) and 53 female (44%) patients of ages 23 to 100 years (mean ± SD, 63 ± 16 years). They were ethnically diverse, reflecting the Southern California population: 65 Hispanic (54%), 29 white (24%), 13 Asian/Pacific Islander (11%), 10 Black (8%), and 4 other (3%) patients. Although 44 patients (37%) were sampled only once, 76 patients (63%) had between 2 and 9 samples collected during our enrollment period.

Outcomes were classified as Moderate (“M”; n = 17; 14%); Severe, discharged without supplemental oxygen (“S”; n = 20; 17%); Severe, discharged with oxygen (“O”; n = 41; 34%); Critical (“C”; n = 13; 11%); and Death (“D”; n = 29; 24%).

Similar to a previous report,⁸  we demonstrated a clear trend of increasing median peak vWF:Ag among an outcome category with worsening outcomes (Figure, A; P < .001). This trend was also evident for CRP (P < .001), D-dimer (P < .001), and ferritin (P < .001; Figure, B through D).

The vWF:Ag data exhibited a significant clustering of values above and below 300% (Figure, A). Patients with a peak vWF:Ag ≥300% had a 5-fold increase in risk of death (21 of 52 versus 8 of 68; odds ratio [OR], 5.08; P < .001) and a 30-fold risk of prolonged LOS (51 of 52 versus 43 of 68; OR, 29.65; P = .001) compared with those with a peak vWF:Ag <300% (Table 1).

There was also a clear division among peak D-dimer results at 3.8 FEU mg/L, corresponding to the 25th percentile of peak D-dimer results among patients who died (Figure, C). Higher peak D-dimer results were associated with a 15-fold increase in risk of death (OR, 14.73; P < .001) and a 5-fold increased risk of prolonged LOS (OR, 4.55; P = .02; Table 2).

To assess the prognostic utility of both tests early in admission we examined the initial paired measurement of vWF:Ag and D-dimer for each patient using the same cutoffs that we applied to peak measurements. Again, higher vWF:Ag was associated with a significantly increased risk of death (OR, 3.54; P = .004) and prolonged LOS (OR, 20.19; P = .004; Table 1). In contrast, higher early D-dimer was not significantly associated with either death (OR, 1.9; P = .29) or prolonged LOS (OR, 1.02; P = .98; Table 2).

Thrombotic risk in COVID-19 is well recognized, and preventative measures such as heparin are employed. Still, these prophylactic measures are counterbalanced by an increased risk of heparin-induced thrombocytopenia.^12–14  D-dimer is the established marker of thrombotic complications in COVID-19. Our data showed that both higher peak vWF:Ag and higher peak D-dimer were very predictive of significant adverse outcomes, such as death and prolonged LOS, among COVID-19 inpatients. The prediction of prolonged LOS was quite gratifying because it was not used in the derivation of the cutoffs that we used.

Guided by our understanding of the role of endothelial injury in thrombosis⁹  and the direct association of elevated vWF:Ag with hypercoagulability,¹⁵  we suspected that it may provide prognostic information before thrombosis had progressed and D-dimer had risen. Our results demonstrated that vWF:Ag was also a significant early predictor of death and prolonged LOS among the early paired samples, whereas concurrent D-dimer was not.

Thrombotic complications were prevalent among our critical and deceased patients. Among the few deceased patients who had autopsies, there was 1 patient with a pulmonary embolus and another with an acute intracerebral infarct. Among the “critical” patients, 5 had pulmonary emboli based upon imaging studies. We suspect that many more of the patients had microthrombi underlying respiratory or multisystem organ failure, as has been found in autopsy series.¹⁶  Our data suggest that vWF:Ag may prove a valuable early guide to the institution and escalation of anticoagulant treatment.

Although there have been numerous reports of promising new biomarkers correlating with illness severity in patients with COVID-19, most are only available in a research setting. In contrast, there are US Food and Drug Administration–approved and Conformité Européenne (CE)–marked vWF:Ag assays readily available in many clinical laboratories. Although vWF:Ag is typically offered at tertiary medical centers and specialized coagulation laboratories, it is a test menu option for many of the same instruments on which D-dimer and prothrombin time is measured in community hospital laboratories.

There are several limitations to this study. We found the outcomes classification criteria of the NIH COVID-19 Treatment Panel¹¹  difficult to apply. Another limitation was the use of a sample of convenience, which measured vWF:Ag and other APRs only at limited points across the hospitalization course. The number of available samples for any given patient was dependent upon both the clinical suspicion for thrombosis and the arbitrary overlap between their hospitalization and our study enrollment period. A study with regular, frequent measurements throughout the course of hospitalization could confirm precisely how soon vWF:Ag rises in the hospital course and how long before D-dimer. Further study may also resolve the difference between the 300% prognostic threshold that we proposed and a cutoff of 423% identified by other authors.⁸ 

Another significant limitation was heterogeneity in the use of anticoagulation in our patient population. At the time of patient sampling, there was no uniform consensus on when to use anticoagulation in patients with COVID-19. Most patients did receive at least “DVT prevention” levels of anticoagulation with low-dose enoxaparin, as is the common practice in any hospitalized patient without a contraindication. Many patients with elevated D-dimer levels received more intensive anticoagulation with enoxaparin, or unfractionated heparin if renal function was significantly impaired. To address this limitation of varied treatment patterns, we feel that a prospective study using vWF:Ag to guide anticoagulation escalation among COVID-19 patients according to a strict protocol is also warranted.

The authors wish to thank the leadership and staff of our SCPMG Regional Reference Special Coagulation Laboratory (Fraulein Chacon; May Wong Lee; Mohammed Salman; Gene Usher; JiYeon Kim, MD, MPH; and Steven McLaren, DO) for the vWF:Ag measurements, and Ms Stephanie Tovar for her project management.