Coagulation Status in Dogs Naturally Infected with Angiostrongylus vasorum

Angiostrongylus vasorum infection has been associated with coagulopathies including hyperfibrinolysis. We compared coagulation status including thromboelastometry (ROTEM) parameters in dogs naturally infected with A. vasorum versus healthy dogs to determine clinicopathological parameters associated with bleeding, hypocoagulopathy, and hyperfibrinolysis. Clinical signs, white blood cell count, platelet count, hematocrit, plasmatic coagulation tests (PT, aPTT, fibrinogen concentration), D-dimer, and ROTEM S parameters (Ex-tem, In-tem, Fib-tem, Ap-tem) were analysed and compared between bleeding, nonbleeding, and control dogs and between hypo- and normocoagulable animals. Clinical signs of bleeding were present in 6/9 (67%) hypocoagulable and 1/9 (11%) normocoagulable dogs. PT, fibrinogen concentration, and several ROTEM parameters were significantly different between hypocoagulable and normocoagulabe A. vasorum infected dogs. Hyperfibrinolysis was identified in 44% of infected dogs and was significantly more common in bleeding and hypocoagulable dogs. Hyperfibrinolysis was significantly associated with low MCFFib-tem but not with low fibrinogen concentration or increased D-dimers. CFTEx-tem > 248 swas 100% sensitive and 89% specific to predict hyperfibrinolysis. Hyperfibrinolysis, hypocoagulability and bleeding are common in A. vasorum infected dogs. Only Ex-tem and Fib-tem parameters and potentially PT were associated with bleeding or hypocoagulability. Ex-tem analysis enables detection of bleeding, hypocoagulability and hyperfibrinolysis within minutes.


Introduction
The canine heart-and lungworm Angiostrongylus vasorum (Baillet 1866) causes serious infections in dogs and is associated with a high mortality rate if not diagnosed and treated in due time [1]. The disease often causes a broad spectrum of clinical signs, including unspecific clinical signs. Bleeding diathesis, present in about 60% of clinical cases, can lead to rapid life-threatening complications [2] and may occur within the central nervous system, the lung, body cavities, mucous membranes, or in the integument [3,4]. The mechanisms and pathogenesis of these coagulopathies are poorly understood and several unresolved hypotheses such as chronic disseminated intravascular coagulation (DIC), FV and FVIII deficiency, immune-mediated thrombocytopenia, anticoagulants secreted by the parasite or activation of the fibrinolytic system of the host have been formulated [2,[4][5][6][7][8][9]. A recent study evaluating serum proteome of experimentally infected dogs did not support an underlying DIC but rather consumption of some key factors of the coagulation and complement cascade [10]. Interestingly, a previous retrospective study identified hyperfibrinolysis (HFL) and associated hypofibrinogenemia in 67% of dogs with an A. vasorum infection and simultaneous clinical signs of bleeding [6]. Hyperfibrinolysis, which can be diagnosed with rotational thromboelastometry (ROTEM) [11], but not with plasmatic coagulation tests, may explain the bleeding diathesis seen in some dogs with A. vasorum infection [6].
Eventually, the identification of dogs at risk for bleeding diathesis remains a challenge. In a previous study, 73% of dogs with A. vasorum infection were hypocoagulable based on thromboelastography (TEG, which is comparable to ROTEM), with TEG parameters being significantly different between bleeding and nonbleeding dogs [2]. These authors found no association between other coagulation parameters (fibrinogen concentration, antithrombin activity, D-dimers, von Willebrand factor) and hypocoagulable parameters on TEG.
To our knowledge, clinical signs of bleeding, plasmatic coagulation times, true prevalence of HFL and its association with bleeding diathesis had not been prospectively evaluated in dogs with A. vasorum infection.
The present study aimed to describe the coagulation status in dogs naturally infected with A. vasorum, to establish the incidence of HFL and to determine clinicopathological parameters indicating the presence of HFL by comparing coagulation parameters of infected dogs to a healthy control group. An additional objective was to evaluate clinicopathological differences between hypocoagulable and normocoagulable A. vasorum infected dogs.
We hypothesized that dogs naturally infected with A. vasorum and with clinical signs of bleeding would show changes in coagulation parameters, and that ROTEM parameters would be superior predictors of HFL than plasmatic coagulation tests or clinical signs at presentation.

Results
Angiostrongylus vasorum positive dogs included five Beagles, three mixed breed dogs and one each of the following breeds: Belgian Shepherd, Bernese Mountain Dog, Chihuahua, Dalmatian, Doberman, German Shepherd, French Bulldog, Golden Retriever, Hovawart, and Schnauzer. The control group included three mixed breed dogs and one Beagle, one Belgian Shepherd, one German Pointer, one Golden Retriever, and one Nova Scotia Duck Tolling Retriever. Breed (p = 0.483), sex (p = 0.230), age (p = 0.154), and weight (p = 0.441) were not significantly different between dogs with A. vasorum infection and control dogs.
Of the 18 dogs with A. vasorum infection, 7 presented with clinical or magnetic resonance imaging (MRI) signs of bleeding (39%). Bleeding occurred in the brain (n = 3), episcleral (n = 2), lip (n = 2), lung (n = 1), and nose (n = 1). Three of seven dogs showed more than one bleeding location. All three dogs with signs of central nervous system (CNS) bleeding on MRI presented with episcleral bleeding.
Differences between control, infected bleeding and infected nonbleeding dogs are summarized in Tables 1 and 2. None of the evaluated clinical parameters at presentation was significantly different between groups ( Table 1). The group of dogs with bleeding diathesis was thrombocytopenic, had a prolonged prothrombin time (PT) and hypocoagulable Ex-tem and Fib-tem parameters compared to both nonbleeding and control dogs.  Differences between hypocoagulable and normocoagulable A. vasorum infected dogs are summarized in Tables 3 and 4. None of the clinical signs occurred in different frequency between hypo-and normocoagulable dogs. Platelet count (but not the presence of thrombocytopenia), PT, and several ROTEM parameters were significantly different between hypocoagulable and normocoagulabe A. vasorum infected dogs. Clinical signs and the localization of clinical manifestations were not significantly different between hypo-and normocoagulable dogs.  Overall incidence of HFL defined by maximum lysis above the reference interval in any of the ROTEM tracings was 44% in dogs with A. vasorum infection (8/18 dogs) and was significantly higher in bleeding dogs (p = 0.011; Table 2) and hypocoagulable dogs (p = 0.015, Table 4). Four of 7 (57%) bleeding dogs showed increased Ex-tem lysis compared to none of the nonbleeding dogs (p = 0.011). Hyperfibrinolysis was significantly associated with low MCF Fib-tem (p = 0.001) but not with low fibrinogen concentration (p = 0.088). AUROC results for prediction of HFL are shown in Table 5. CFT Ex-tem > 248 s was 100% sensitive and 89% specific and CT Extem > 103 s was 75% sensitive and 99% specific. Hyperfibrinolysis was not associated with high D-dimers (p = 0.559). All dogs were discharged from the hospital. Four dogs had been treated with fresh frozen plasma (20-30 mL/kg BW) and additionally received tranexamic acid (20-60 mg/kg). These four dogs were bleeding, hypocoagulable, and lacked clot formation in the Fib-tem tracing. An additional bleeding dog was treated with tranexamic acid. Plasma transfusion was significantly associated with low fibrinogen Clauss (p = 0.045), low MCF Fib-tem (p = 0.011), CT Ex-tem prolongation (p = 0.011), and bleeding (p = 0.011) but not with hypocoagulability (p = 0.082).
Ap-tem tracings showed both weaker and stronger MCFs than Ex-tem profiles. MCF Ap-tem was not stronger (indicating HFL) in any of the 5 dogs with HFL diagnosed in Ex-tem or In-tem but was > 15% stronger in 4/15 dogs with no HFL and ML's within the reference interval.

Discussion
The present prospective study identified bleeding diathesis in 39% of dogs naturally infected with A. vasorum. Among those, 86% were diagnosed hypocoagulable by thromboelastometry and plasmatic coagulation tests. Hyperfibrinolysis was present in 44% of all infected dogs and in 56% of those dogs with bleeding diathesis.
The true incidence of A. vasorum associated bleeding diathesis remains unknown, as this study was based mostly on referred cases. However, our incidence of 39% is in the range of previously reported incidences between 35% and 57% [6,12].
The main clinical signs of dogs naturally infected with A. vasorum are coughing and tachypnea/dyspnea together with unspecific signs such as lethargy, anorexia, and vomiting/diarrhea [2,4,[12][13][14][15]. Haemorrhage and neurological signs are less frequently observed [4,12]. None of the clinically evaluated parameters in the current study were significantly associated with the presence of haemorrhage, as previously observed [2].
ROTEM analysis identified several parameters that were significantly different between dogs with bleeding diathesis compared to infected dogs without bleeding and healthy control dogs. Prolonged CT Ex-TEM , CFT Ex-TEM , decreased MCF Ex-tem and MCF Fib-tem , and Ex-TEM HFL were significantly associated with the presence of bleeding. These findings indicate that both clot formation and clot strength are affected in dogs with A. vasorum induced coagulopathy.
In contrast to a previous report [2], our study population of dogs with bleeding diathesis showed a significantly lower median platelet count compared to both nonbleeding dogs and healthy control dogs. The platelet count of bleeding dogs was below the reference interval but not in a range that would explain spontaneous bleeding. Several studies and case reports demonstrated thrombocytopenia and prolonged clotting times [8,[16][17][18], but experimental studies indicate that such findings are inconsistent over time [7,9,13]. Thrombocytopenia was significantly associated with the presence of bleeding in our study population and 83% of dogs with bleeding were thrombocytopenic (n = 5/6), while this was observed in only 10% (n = 1/10) of nonbleeding infected dogs. However, no reasonable platelet count cut-off value to identify dogs with bleeding diathesis could be identified and whether the decreased platelet count is the cause or the result of bleeding is unknown. Together with the identified mild thrombocytopenia, we interpret this finding as evidence that thrombocytopenia is a consequence rather than the cause of bleeding, indicating that mechanisms other than thrombocytopenia are causing bleeding diathesis.
The PT, aPTT, and fibrinogen concentration were not significantly associated with bleeding, indicating that viscoelastic testing is superior to classical plasmatic coagulation times for the identification of bleeding in A. vasorum infected dogs.
While overt clinical signs of bleeding will guide the clinician to investigate the coagulation status, internal bleeding in the brain or lung may be fatal if hypocoagulability is not detected and treated [17,19]. The recognition of dogs at risk for bleeding diathesis remains a challenge [6]. Therefore, the identification of valid prediction parameters to detect hypocoagulability are of relevant clinical interest.
We therefore analysed clinical and laboratory parameters to detect hypocoagulability in A. vasorum infected dogs. Of interest, only two-thirds of hypocoagulable dogs showed clinical signs of bleeding and more than one-quarter of nonbleeding dogs were hypocoagulable and hence at risk to develop bleeding during hospitalization. Cough, pulmonary changes consistent with radiographic signs of A. vasorum infection [20] on radiographs and presence of respiratory signs were not associated with hypocoagulability, suggesting that respiratory signs and the typical interstitial-alveolar pattern seen on thoracic radiographs are not caused by haemorrhage; consequently, thoracic radiographs will not aid in the identification of hypocoagulable dogs. This finding is supported by a study evaluating radiographic, CT and necropsy findings in experimentally infected dogs and linking them to granulomatous inflammation rather than haemorrhage [20].
Of note, all dogs with neurological signs were hypocoagulable. Such A. vasorum infected dogs commonly do not present with classical pulmonary signs and internal haemorrhage may not be apparent [17]. However, the three dogs with neurological signs all presented with episcleral bleeding, a finding that should be investigated in a larger study population. Until then, it is important to not only include A. vasorum infection as a differential diagnosis to central nervous clinical signs but to further investigate the coagulation status for hypocoagulability if the dogs are infected.
In contrast, none of the dogs with an incidental diagnosis of A. vasorum infection was hypocoagulable. We suspect that these dogs were either not infected long enough to develop coagulopathies or with too few worms to show hypocoagulability or clinical signs of bleeding. Experimentally inoculated dogs with higher worm burdens showed earlier and more severe respiratory signs and more pronounced coagulopathies [13]. The magnitude of parasite burden was unknown in our study population; therefore, a direct comparison cannot be made. In addition, individual differences between dogs cannot be excluded: the study of dog serum proteome in experimentally infected dogs revealed that the lectin pathway and the coagulation cascades were particularly affected; these may be additionally associated with individual differences that altogether may explain the onset or absence of hypocoagulability [10].
In the present work, not all hypocoagulable dogs exhibited identifiable signs of bleeding, highlighting the need for laboratory analysis of coagulation status to identify dogs at risk of bleeding. Hypocoagulable dogs had significantly longer PT and abnormal Ex-tem, In-tem, and Fib-tem parameters but no significant thrombocytopenia or prolonged aPTT. Previously, no associations could be identified between coagulation parameters (fibrinogen concentration, antithrombin activity, D-dimers, von Willebrand factor) and hypocoagulable thromboelastography parameters [2], indicating that viscoelastic tests may be more appropriate to identify hypocoagulable dogs.
We recently retrospectively identified HFL and associated hypofibrinogenemia in A. vasorum infected dogs [6], two findings that are supported by the data presented here. Hyperfibrinolysis was identified in 44% of dogs naturally infected with A. vasorum and 5/9 (56%) of bleeding dogs showed increased Ex-tem or In-tem lysis compared to 1/9 (11%) nonbleeding dogs. Both bleeding and hypocoagulable dogs had low to no fibrin clot formation compared to nonbleeding dogs. Hypofibrinogenemia was significantly more common in hypocoagulable dogs. Bleeding dogs displaying decreased fibrinogen concentration have also been described previously [2]. As suggested earlier [5,6], the cause of decreased fibrin clot formation in the Fib-tem tracing may be explained by HFL and not by low fibrinogen concentrations. Hyperfibrinolysis was significantly associated with MCF Fib-TEM but not with fibrinogen concentration measured by Clauss. Identification and treatment of HFL therefore seems to be of clinical relevance. Neither platelet count nor plasmatic coagulation times are able to predict or diagnose HFL. ROTEM analysis; however, was able to predict HFL by means of prolonged CT and CFT within minutes.
The initial study protocol envisaged Ap-tem profiles to confirm HFL identified in Ex-tem profiles. Our study, showing inconsistent Ap-tem results, supports earlier evidence that Ap-tem assays are not helpful in the diagnosis of HFL in dogs [6,11].
Elevated D-dimer concentrations in dogs with A. vasorum infection have also been described [2,17,21]. Although expected to increase with fibrinolysis, D-dimer levels were not different between hypo-and normocoagulable dogs in our study population and were not associated with HFL. A reason for this finding may be that the presence of Ddimers requires lysis of crosslinked fibrin while A. vasorum associated lysis of fibrinogen or fibrin strands may occur prior to crosslinking to a stable fibrin clot [6]. Fibrinogen, being an acute phase protein, is expected to be present at high levels in dogs with A. vasorum infections. Another possible explanation for the lack of stable fibrinogen clot is FXIII deficiency or malfunction. FXIII is needed for cross-linking of fibrin and congenital or acquired deficiency leads to unexpected bleeding in people [22,23]. In line with this, a recent time-course proteomics study performed on sera of A. vasorum infected dogs identified FXIII-B, a subunit of FXIII, as downregulated among other candidates of the coagulation cascade and complement pathway compared to before inoculation with the parasite [10].
While significantly more hypocoagulable dogs were treated with tranexamic acid, hypocoagulability was not significantly associated with administration of a plasma transfusion. Significantly more plasma transfusions were given to bleeding dogs, however, a plasma transfusion was not deemed necessary in 3/7 dogs with bleeding, based on coagulation parameters. Together with the wide variety of identified ROTEM abnormalities in hypocoagulable dogs, this indicates that coagulation testing the complete set of coagulation parameters is crucial to identify not only dogs at risk for bleeding but also to treat the coagulation disorder with specific therapeutic measures.
In former studies up to 30% of A. vasorum infected dogs died [2,12], while all dogs in this study were discharged from the hospital. It is possible that early identification of coagulation problems and early therapeutic intervention has led to the excellent survival rate in our study population. As all patients were treated based on ROTEM results as well as at the discretion of the clinician in charge, the effect of treatment on survival cannot be investigated further.
This study included only a small group of A. vasorum infected dogs and only part of them showed bleeding diathesis. Unfortunately, bleeding dogs were often presented during emergency hours when plasmatic coagulation testing was not available. We however showed that results of plasmatic coagulation testing should be interpreted with caution. Additionally, when ROTEM tests were chosen by the clinician in charge during emergency service hours, in several patients In-tem analysis was not performed. In-tem parameters between bleeding and nonbleeding A. vasorum infected dogs could therefore not be compared. Bleeding was furthermore based on visible clinical signs of bleeding and MRI, with the possibility of having falsely assigned an internally bleeding dog to the nonbleeding group.

Materials and Methods
Dogs presenting to the Small Animal Clinic, Vetsuisse Faculty, University of Zurich between March 2016 and July 2019 and being diagnosed with A. vasorum infection by either serological ELISA antigen or rapid-assay (AngioDetect TM ) [24,25] and/or copromicroscopic identification of first stage larvae by the Baermann funnel method were eligible to enter this and a concurrent study. Healthy dogs were used as a control group. Control dogs had a normal complete blood cell and chemistry panel, plasmatic coagulation times including fibrinogen measured by Clauss and were negative for A. vasorum infection. The project was approved by the ethics committee on animal research of the canton of Zurich (ZH001/16,27384) and informed owner consent was obtained for both healthy and infected dogs, in which additional blood tests were performed for the study.
Dogs were excluded if they were pre-treated with antifibrinolytics, blood products or NSAIDS within five days of presentation and if ROTEM analysis was not performed within 30 min after blood sampling.
Signalment, sex, weight, presenting complaint, clinical signs including the presence of bleeding were determined at presentation. The infected dogs were assigned to the bleeding or the nonbleeding group based on the presence of clinical signs of bleeding or MRI signs compatible with bleeding in dogs undergoing MRI exam. In addition, clinical signs were categorized as mucocutaneous bleeding, lung or CNS localization or incidental diagnosis without clinical signs (localization of clinical manifestation). Diagnostic imaging findings and survival to discharge were extracted later from patient records.
Approximately 8-12 mL of venous blood were drawn at presentation or immediately after diagnosis of A. vasorum infection using a 20 G hypodermic needle. Blood samples were transferred to tubes in the following order: two 3.2% citrate and one EDTA tube (all Sarstedt AG, Sevelen, Switzerland). One citrate and the EDTA tube were sent to the in-house laboratory for plasmatic coagulation tests (PT, aPTT, TT, fibrinogen Clauss , and D-dimers) (Start 4 STAGO CH SA, Glattbrugg, Switzerland) and complete blood cell status (Sysmex-XT 2000iV, Sysmex Cooperation, Kobe, Japan). Another citrate tube was placed within the warming plate of the ROTEM device for 10-20 min.
Ex-tem S, In-tem S, Fib-tem S, and Ap-tem S analysis (TEM International GmbH, Munich) were performed by the primary investigators according to an institutional protocol based on the manufacturer's instruction and international guidelines [26,27]. Briefly, 300 µL citrated whole blood were incubated with the lyophilized reagent before placement in the cup and start of measurement. An automated pipetting system was used. Tracings were run for 60 min. All tracings were checked for artefacts by two of the authors (N.E.S, A.P.N.K). Clotting time (CT), clot formation time (CFT), alpha angle (a), maximum clot firmness (MCF), maximum clot elasticity (MCE) and maximum lysis (ML) were extracted from the ROTEM database. A green line in the fib-tem tracing was defined as an MCF Fib-tem of 0 mm. If MCF did not reach 20 mm, CFT was defined as 3600 s. Hypofibrinogenemia was defined as fibrinogen measured by Clauss < 1.2 g/L (in-house reference interval) or MCF Fib-tem < 2 mm.
Based on ROTEM profiles, dogs were categorized as hypocoagulable if ≥2 of the following parameters were hypocoagulable compared to reference intervals [28]: CT Ex-tem or PT, CT In-tem or aPTT, CFT Ex-tem / In-tem or α Ex-tem / In-tem , MCF Ex-tem / In-tem / Fib-tem or MCE Ex-tem / In-tem / Fib-tem . Hyperfibrinolysis was defined as ML > reference interval in any of the tracings. Dogs were treated at the discretion of the clinician in charge, generally based on a standardized protocol including 50 mg/kg BW fenbendazole daily for three weeks and 1 mg/kg BW prednisolone daily for seven days was administered; strict exercise constraint was advised. Fresh frozen plasma (20 mL/kg BW) was transfused for hypofibrinogenemia and tranexamic acid was prescribed to treat HFL (20 mg/kg BW, repeated until HFL resolved) as needed.

Statistical Analysis
Data were entered into a spreadsheet and analyses were performed using the statistical software program IBM SPSS (Version 23.0. IBM Corp, Armonk, NY, USA). Continuous data is presented as median (min-max). Clinical signs and coagulation parameters were compared between dogs with and without bleeding diathesis and control dogs using Kruskal-Wallis tests followed by post hoc analysis with Dunn-Bonferroni for continuous variables and chi 2 test for categorical variables. In a second step dogs with A. vasorum infection were grouped as hypo-or normocoagulable and clinical signs and coagulation parameters were compared across groups using Mann-Whitney U-test or Fisher's exact test. The accuracy of standard coagulation tests and ROTEM parameters to determine dogs with HFL was evaluated using receiver operating characteristics (ROC) and the area under the ROC curve (AUROC). Youden's J statistic was used to select the optimum cut-off point of the ROC curves. A value of p < 0.05 was considered significant.

Conclusions
In conclusion, HFL, hypocoagulability and bleeding are commonly occurring in A. vasorum infected dogs. None of the evaluated clinical parameters are significantly associated with bleeding or hypocoagulability, and plasmatic coagulation tests did not allow clear discrimination between bleeding and nonbleeding A. vasorum infected dogs. In contrast, most ROTEM parameters were significantly different between groups, allowing the differentiation between bleeding and nonbleeding A. vasorum infected dogs. Ex-tem analysis was able to detect bleeding, hypocoagulability, and HFL within a few minutes after blood draw and may therefore be recommended. Informed Consent Statement: Informed consent was obtained from all dogs involved in the study in which additional blood tests were performed for the study.

Data Availability Statement:
The data presented in this study is available on request from the corresponding author.