Introduction

Coronavirus disease 2019 (COVID-19) is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which is responsible for acute respiratory distress syndrome in a significant proportion of affected patients (27–31%) [1, 2].

The clinical spectrum of SARS-CoV-2 infection appears to be wide, encompassing asymptomatic infection, mild flu-like syndrome, and severe viral pneumonia with respiratory failure and superimposed bacterial infections which may be fatal [3,4,5].

Venous thromboembolism (VTE) includes deep vein thrombosis (DVT), pulmonary embolism (PE) and portal vein thrombosis. The association between VTE and COVID-19 was described for the first time by Zhou et al. [6], who identified thrombotic events in 2.9% of a cohort of COVID-19 patients. In previous Asian series, thromboembolic events have been reported in roughly one fourth of COVID-19 patients admitted to the intensive care unit, and these findings correlated with a poor prognosis [7]. A number of pathogenic mechanisms have been hypothesized for VTE in COVID-19 patients, including active inflammation, immobilization and intensive care treatments, but the limited evidence available in the literature does not allow to estimate the relative contribution of each of the above-mentioned factors [8].

Little is known regarding the risk of VTE in non-Asian countries, that are known to have a higher incidence of VTE in comparison to Asian countries [9]. Recent publications reported a high rate of VTE and arterial thrombotic events especially in critically ill patients (10–36% depending on the analyzed cohort), also in Occidental cohorts of COVID-19 subjects [10,11,12,13,14,15].

Starting from these premises, we here aimed to define VTE rates and types, not considering peripheral and central catheter-related thrombosis, among a cohort of COVID-19 patients during their hospital stay at the San Matteo Hospital Foundation (Pavia, Northern Italy).

Methods

This is a single-center, retrospective, observational study. We extracted data from medical records of all 259 consecutive patients with a diagnosis of COVID-19 admitted to the Departments of Internal Medicine, Infectious Disease, Intensive Care, and Respiratory Disease of the San Matteo Hospital Foundation (Pavia, Northern Italy), between March 19th and April 6th, 2020. The enrolment was limited to this time lapse as exhaustive data were available only for patients enrolled during this period. All patients had a confirmed diagnosis of COVID-19 by Real Time Polymerase Chain Reaction (RT-PCR) in respiratory samples (see Supplementary materials).

We reviewed records of all 259 COVID-19 patients for demographic information, co-morbidities, risk factors for VTE according to the Padua prediction score [16], laboratory tests and anticoagulation treatment at the time of hospital admission.

DVT and PE were diagnosed through ultrasound and CT-scan or CT-angiography, respectively. In particular, every patient with clinical suspect of VTE or worsening respiratory function, underwent four-points US or CT-scan. The regions tested included: the common femoral vein at the level of inguinal crease, the superficial femoral vein superior to the adductor canal, the popliteal vein in the popliteal fossa and the great saphenous vein. The remaining patients were tested for routine laboratory tests if not daily, at least every 2–3 days. Symptomatic patients were defined as follows: worsening PaO2/FiO2 despite optimized therapy, signs and symptoms of DVT (i.e., pain, swelling, warmth), increase of d-dimer. No peripheral and central catheter-related DVTs were included in the study.

We also examined the correlation between DVT and selected inflammatory markers, including C-reactive protein (CRP), interleukin (IL)-6, mean platelet volume (MPV), neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR).

The study was performed as a clinical audit using routinely collected clinical data and as such is exempt from the need to require written informed consent. The study was approved by the local ethics committee (San Matteo Hospital Foundation) on March 13th 2020.

The results of this study are reported according to the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) recommendations [17].

Statistical analyses

Given the observational nature of the study, sample size was not calculated a priori. Continuous variables were expressed as median and range and compared with the Mann–Whitney U test. Categorical variables were described as number (%), and proportions for categorical variables were compared using the Fisher exact test. Spearman correlation coefficient was calculated for relevant laboratory parameters. When variables were not available for some patients, these were excluded for percentage calculation. A two-sided α of < 0.05 was considered statistically significant. Statistical analyses were performed using MedCalc (Belgium, version 2020).

Results

A total of 259 cases of definite COVID-19 patients were included in the registry between March 19th and April 6th 2020. Demographic characteristics, medical history and laboratory parameters of all patients included in the study are reported in Table 1. The median age of the patients was 70 years (range 25–97), 32% of the total were female (median age 66, range 25–94), 68% were male with a median age of 74 (range 27–97).

Table 1 Demographic characteristics, medical history and laboratory results of the two whole cohort of COVID-19 patients enrolled in the study, and of the subcohorts of patients with and without VTE at the time of hospital admission

Twenty-five patients turned out to be complicated by VTE (9.6%), of which 21 had DVT, four had PE, and one among them also had portal vein thrombosis. All patients with VTE had a Padua score > 4. We did not identify any incidental diagnosis of VTE, because asymptomatic patients did not undergo diagnostic work-up. Of the 259 enrolled patients, 239 (92.3%) received thromboprophylaxis. Among them, 24 (10.4%) developed VTE.

Twenty-two out of the 25 VTE cohort patients underwent prophylactic therapy with low molecular weight heparin (LWMH) or calcic heparin. Three patients of the VTE-group had clinical evidence of lower DVT by the time of hospital admission, hence anticoagulant therapy was immediately started. The median time lapse between hospital admission and VTE onset was 5.0 days (range 0–5 days). The median time lapse between onset of COVID-19-related symptoms and VTE was 15.0 days (range 10–30 days). DVT was the most frequent event, distributed among upper extremities and lower limb, being upper DVT more prevalent. Of note, the prevalence of VTE in ICU patients was higher in respect to sub intensive/intermediate care patients departments (17% and 8%, respectively; p = 0.02). There was a significant correlation between platelet parameters and inflammation. In particular, platelet count directly correlated with either CRP (p < 0.0001, rs = 0.13392) or LDH (p = 0.0013, rs = 0.05438), and it inversely correlated with MPV (p < 0.0001, rs = − 0.75608). MPV significantly correlated in an inverse manner with either CRP (p < 0.0001, rs = − 0.01657) or LDH (p < 0.0001, rs = 0.16386). PLR significantly correlated directly with CRP (p < 0.0001, rs = 0.35769). d-dimer correlated directly with PLT (p = 0.006, rs = 0.03036), LDH (p = 0.007, rs = 0.33395) and inversely with MPV (p < 0.0001, rs = 0.5161) and lymphocyte (p < 0.0001, rs = − 0.0308). Of note, in the non-VTE group a correlation between PLR and CRP was not observed (p = 0.9, rs = 0.20984) (Table 2).

Table 2 Demographic, clinical and biochemical characteristics of COVID-19 patients at the time of hospital admission

Furthermore, 76% of VTE patients were male and the median age at admission was 62 (range 44–84), while in the non-VTE group patients were significantly older (median 70, range 25–97), and the male population was less represented. No differences were found between VTE and non-VTE cohorts in terms of anticoagulant prophylactic therapy.

There was no significant difference in d-dimer levels between the VTE and non-VTE patients (p = 0.09). LDH was significantly higher in VTE-group compared to the non-VTE group (median 590, range 184–4641 mU/ml). Serum levels of IL-6 were significantly increased in VTE compared to non-VTE patients (p = 0.04). All the other parameters did not significantly differ between VTE and non-VTE patients.

Supplementary Table 1 shows characteristics and differences between VTE patients admitted to intensive care unit (ICU) and VTE general ward patients.

Discussion

Our findings showed a high prevalence (9.6%) of VTE in a non-Asian cohort of COVID-19 patients, which is higher when compared to the prevalence of VTE in patients hospitalized for other causes (1.2–2.0%) [18, 19]. Surprisingly, this high rate of VTE was encountered despite prophylactic anticoagulation therapy in almost all patients. In fact, of all the patients who received thromboprophylaxis, 10.4% developed VTE, which is a higher rate compared to the standard of 0–2% of symptomatic VTE during enoxaparin prophylaxis. In the subcohort of ICU patients, VTE rate was 16.6%, which is rather comparable to the literature-reported rate for these critically ill patients (10–50%, if VTE is screened). Indeed, the VTE rate in our cohort is lower than that from some other recent reports [10, 12]. This might reflect the patients’ setting, since a minority of our patients was managed in the ICU and underwent chest CT-scan.

Surprisingly, 13 out of 25 VTE patients (52%) presented with upper extremities DVT. Of note, 8 of these were ICU patients (62%), while the remaining 5 (38%) were general ward patients and all of them received supportive ventilation through c-PAP. These findings support the notion that intensive care expose critically ill patients to a higher risk of developing DVT. In fact, critically ill patients are at high risk of VTE as they are susceptible to both general risks factors of VTE as well as those specific to ICU patients, such as sedation, immobilization, and use of vasopressors [20]. Furthermore, peripheral lines and peripherally inserted central catheters (PICCs) for intravenous therapy, could have contributed to the thrombotic process, particularly in an inflammatory state like the COVID-19 related one.

The relationship between viral pneumonia and predisposition to VTE has been well described [21, 22]. This phenomenon is the result of interaction between activated leukocytes and cellular adhesion molecules on the vein wall [23,24,25,26,27,28,29,30]. Inflammatory mechanisms up-regulate procoagulant factors, down-regulate natural anticoagulants and inhibit fibrinolytic activity [31]. In particular, as Marongiu et al. assert, the massive endothelial damage and activation of coagulation cascade is local [32], leading to pulmonary microvessels thrombosis [33]. Therefore, our study supports the notion that VTE correlates with an inflammatory burst and inflammatory indexes such as LDH, IL-6 and PLR could be useful in stratifying the risk for COVID-19 patients to develop VTE manifestations.

Unfortunately, there is still a gap of knowledge on the certain pathogenesis of VTE related to SARS-CoV-2 infection, so that we cannot exclude that a part of the PE reported were pulmonary thrombosis and not traditional isolated PE [34, 35].

Indexes such as MPV and PLR have been used as unfavorable predictors in other viral infections [36, 37]. The existence of a significant correlation between PLR and inflammatory indexes in the VTE subcohort seems to support the role of this marker as an unfavorable predictor in COVID-19 patients, in keeping with recent reports [38].

In contrast with Tang et al. [39], we did not find any significant difference in d-dimer between the two cohorts and this could be related to the numerical imbalance between VTE and non VTE patients.

Our study also showed that, unexpectedly, platelet count was only slightly reduced or normal in most of COVID-19 patients. That could be explained by the inflammatory state too, since thrombopoietin could increase in response to lung inflammation [40]. Hence, inflammatory profile, more than platelet count and coagulation profile, could help us to determine and estimate the severity of COVID-19. According to recent publications, we believe that enoxaparin and fondaparinux should be preferred instead of calcic heparin because of its anti-inflammatory properties [41]. Moreover, Mycroft-West et al. [42], showed that LMWH can bind to the SARS-CoV-2 surface protein (Spike) S1 receptor binding domain and block the replication of the virus, thus showing potential antiviral effects.

If early initiation of prophylactic anticoagulant therapy should be recommended in COVID-19 patients [8], how to optimize anticoagulant therapy is still a matter of debate [43]. It can be assumed that by implementing the dosage of LMWH traditionally considered as prophylactic, if renal function is permissive and in the absence of bleeding diathesis, most VTE events could be prevented, but further studies are necessary to confirm this therapeutic strategy.

Indeed, this study has some limitations, including the small sample size, the numerical imbalance between VTE and non-VTE cohorts, the lack of follow-up data, and the scarce use of chest CT-scan in the imaging diagnostic protocol of COVID-19 pneumonia at the time of hospital admission. Moreover, even if catheter-related DVTs were not included in the study, we cannot exclude that blood sampling could partially explain such a high incidence.

Nevertheless, our study confirmed the high rate of VTE in COVID-19 patients, especially in those admitted to the ICU. If anticoagulant therapy is then essential in the management of these patients, attention should be paid to the bleeding risk, due to the increased incidence of peptic ulcer disease in COVID-19 patients [44].

The relationship between VTE and inflammation might support the hypothesis that a cytokine burst plays a central role in thrombotic complication in COVID-19 patients. Supporting our speculation, recent works have discussed how hyperinflammation and detrimental immunothrombosis might play a central role in COVID-19 pathophysiology. The short median time lapse between hospital admission and VTE onset suggests that this phenomenon is not correlated with hospitalization, but rather with COVID-19 per se.

Since 100% of VTE patients had a Padua score greater than 4, we suggest performing upper and lower limb venous ultrasound in COVID-19 patients with a score > 4, as well as routinely in ICU patients to identify and treat DVT as soon as possible. We also suggest routine CT scans in patients with hypoxemia that is not fully explained by lung imaging, as well as in those with a profile risk similar to that described in the cohort of VTE patients.