Factors Associated with Systemic Bleeding in Bothrops Envenomation in a Tertiary Hospital in the Brazilian Amazon

Bothrops snakebites usually present systemic bleeding, and the clinical–epidemiological and laboratorial factors associated with the development of this manifestation are not well established. In this study, we assessed the prevalence of Bothrops snakebites with systemic bleeding reported at the Fundação de Medicina Tropical Dr. Heitor Vieira Dourado, in Manaus, Amazonas State, Brazil, and the clinical–epidemiological and laboratorial factors associated with systemic bleeding. This is an observational, cross-sectional study carried out between August, 2013 and July, 2016. Patients who developed systemic bleeding on admission or during hospitalization were considered cases, and those with non-systemic bleeding were included in the control group. Systemic bleeding was observed in 63 (15.3%) of the 442 Bothrops snakebites evaluated. Bothrops snakebites mostly occurred in males (78.2%), in rural areas (89.0%) and in the age group of 11 to 30 years old (40.4%). It took most of the patients (59.8%) less than 3 h to receive medical assistance. Unclottable blood (AOR = 3.11 (95% CI = 1.53 to 6.31; p = 0.002)) and thrombocytopenia (AOR = 4.52 (95% CI = 2.03 to 10.09; p < 0.001)) on admission were independently associated with systemic bleeding during hospitalization. These hemostatic disorders on admission increase the chances of systemic bleeding during hospitalization. Prospective studies are needed to clarify the pathophysiology of systemic bleeding in Bothrops snakebites in the Amazon region.


Introduction
Snakebite envenomation is a potentially life-threatening injury and a major public health problem in rural areas of tropical and sub-tropical countries of Africa, the Middle-East, Asia, Oceania, and Latin in our study ( Figure 1). The majority of Bothrops envenomations were in males (78.2%) and in the age group of 11 to 30 years old (40.4%). These snakebites mostly occurred in Manaus (57.4%), Amazonas State capital, in rural areas (89.0%), and only 39.8% were related to work. The majority of patients (59.8%) took less than 3 h to reach medical assistance. The most affected anatomical site was the foot (72.1%). The Bothrops bite was confirmed by snake identification in 32.8% of the included patients' records. All Bothrops snakes were identified as being Bothrops atrox (Table 1).
Toxins 2018, 10, x FOR PEER REVIEW 3 of 16 in our study ( Figure 1). The majority of Bothrops envenomations were in males (78.2%) and in the age group of 11 to 30 years old (40.4%). These snakebites mostly occurred in Manaus (57.4%), Amazonas State capital, in rural areas (89.0%), and only 39.8% were related to work. The majority of patients (59.8%) took less than 3 h to reach medical assistance. The most affected anatomical site was the foot (72.1%). The Bothrops bite was confirmed by snake identification in 32.8% of the included patients' records. All Bothrops snakes were identified as being Bothrops atrox (Table 1).    Clinical characterization of Bothrops snakebite patients on admission showed that the most frequent local manifestations were pain (93.9%) followed by swelling (93.4%). Among the systemic alterations related to hemostasis, unclottable blood (58.2%), bleeding (12.6%), and thrombocytopenia (8.7%) were observed. Other common systemic manifestations were headache (11.4%), nausea (10.7%), and vomiting (9.7%). Most cases were moderate (54.2%), followed by mild (27.7%) cases. Comorbidities such as arterial hypertension (9.3%) and diabetes (2.6%) were also observed. There were no recorded deaths due to a snakebite in this study period (Table 2).

Hemostasis Parameters
Systemic bleeding was observed during hospitalization in 15.3% of the Bothrops snakebite patients. Conjunctival bleeding was the most frequent form of systemic bleeding (6.7%), followed by gingival bleeding (6.3%) ( Table 3). The majority of the Bothrops snakebite patients with systemic bleeding during hospitalization showed a platelet count on admission that was higher than 150,000/µL, irrespective of the type of systemic bleeding, except for one patient who presented petechiae (<50,000/µL) ( Table 4).
Median platelet count on admission of Bothrops snakebite patients with (217,000/µL) and without (239,500/µL) systemic bleeding during hospitalization showed significant statistical differences (p = 0.035) ( Figure 2). Negative (inverse) correlation was found between the variables of number of platelets and mean platelet volume on admission (r = −0.463; p < 0.001) for absolute values during testing ( Figure 3).         The area under the Receiver operating characteristics (ROC) curve showed that platelet counts obtained on admission presented a poor ability for discriminating Bothrops snakebite patients with systemic bleeding during hospitalization (AUROC = 0.6133; 95% CI = 0.531 to 0.696; p = 0.007) ( Figure 4). The area under the Receiver operating characteristics (ROC) curve showed that platelet counts obtained on admission presented a poor ability for discriminating Bothrops snakebite patients with systemic bleeding during hospitalization (AUROC = 0.6133; 95% CI =0.531 to 0.696; p = 0.007) ( Figure  4). Systemic bleeding and unclottable blood presented an improvement within 24 h in most patients after administration of the specific antivenom, while the frequency of thrombocytopenia increased in the first 48 h after antivenom administration, though it did decrease on discharge, and the frequency of abnormal mean platelet volume showed a decrease until discharge. ( Figure 5). No bleeding episode was recorded in patients who returned for clinical reassessment after being discharged. Systemic bleeding and unclottable blood presented an improvement within 24 h in most patients after administration of the specific antivenom, while the frequency of thrombocytopenia increased in the first 48 h after antivenom administration, though it did decrease on discharge, and the frequency of abnormal mean platelet volume showed a decrease until discharge ( Figure 5). No bleeding episode was recorded in patients who returned for clinical reassessment after being discharged.

Factors Associated with Systemic Bleeding
A total of 63 Bothrops snakebite patients with systemic bleeding during hospitalization were included as cases and 349 Bothrops snakebite patients with non-systemic bleeding were included as controls. Unclottable blood (AOR = 3.11 (95% CI = 1.53 to 6.31; p = 0.002)) and thrombocytopenia (AOR = 4.52 (95% CI = 2.03 to 10.09; p < 0.001)) on admission were independently associated with the risk of developing systemic bleeding after a Bothrops snakebite during hospitalization (Table 5).

Factors Associated with Systemic Bleeding
A total of 63 Bothrops snakebite patients with systemic bleeding during hospitalization were included as cases and 349 Bothrops snakebite patients with non-systemic bleeding were included as controls. Unclottable blood (AOR = 3.11 (95% CI = 1.53 to 6.31; p = 0.002)) and thrombocytopenia (AOR = 4.52 (95% CI = 2.03 to 10.09; p < 0.001)) on admission were independently associated with the risk of developing systemic bleeding after a Bothrops snakebite during hospitalization (Table 5).

Discussion
Bothrops snake envenomation usually results in systemic effects such as coagulation disorders and systemic bleeding [7,22]. Although systemic bleeding is a common clinical manifestation in Bothrops snakebites, the clinical-epidemiological and laboratorial factors associated with the development of this effect are not well known in relation to the envenomations that occur in the Amazon region. In this study, systemic bleeding was present in 15.3% of the Bothrops snakebite patients during hospitalization, the most common being conjunctival and gingival, in consonance with previous studies from the Brazilian Amazon [7,23,24]. However, this rate was lower than that observed in envenomations by other species of Bothrops in Central and South America [15,21,25], which suggests that other mechanisms may be involved in the development of systemic bleeding.
The majority of Bothrops snakebites occurred in males, between 11 and 30 years old, living in rural areas and were mainly unrelated to work. Most of our patients took less than 3 h to reach medical assistance. These characteristics were also observed in other studies regarding the Brazilian Amazon and other regions of Brazil [4,21,23], as well as in other countries [16,26]. However, there was no significant association between systemic bleeding and epidemiological aspects of Bothrops snakebites in this study, although age and time taken to reach medical assistance were independently associated with severity and mortality in other studies of snakebites conducted in the Brazilian Amazon [4].
In this study, thrombocytopenia was an infrequent event. This observation was similar to that found in another study carried out in Belém, Pará State, Brazil, where B. atrox also causes most of the snakebites [7]. Contrary the this, B. jararaca snakebites showed a higher frequency of thrombocytopenia [15,21,22,27]. Most of the cases of systemic bleeding observed in our study showed a normal platelet count on admission, irrespective of the type of systemic bleeding, except for one case of petechiae. However, although thrombocytopenia is an uncommon finding, in this study thrombocytopenia on admission was independently associated with the development of systemic bleeding during hospitalization.
Indeed, platelets participate in reactions of hemostasis such as in adhesion to the cut end of a blood vessel, spreading of adherent platelets on the exposed subendothelial surface, secretion of stored platelet constituents, and formation of platelet aggregates [28,29]. In addition, platelets interact with proteins of the coagulation through surface receptors and phospholipids [30,31]. Recently, it has been observed that platelets also have roles in immune and inflammatory processes [32,33]. In snakebites, thrombocytopenia probably results from a multifactorial etiology as venom-induced platelet aggregation [34], sequestration to the areas of damage near the site of the bite [35], and venom-induced oxidative stress leading to a drop-in platelet count by apoptosis [36]. Components isolated from the B. atrox venom such as thrombocytin [34], batroxostatin [37], botrocetin [38], and batroxrhagin [39] may be acting on platelets leading to thrombocytopenia or changes in platelet function.
On the other hand, a negative correlation was found between the number of platelets and mean platelet volume on patient admission, which suggests a consequence of peripheral platelet destruction in which the mean platelet volume tends to increase. Studies show large platelets are more reactive than ordinary size platelets, as measured by aggregation and total release of granular content [40]. However, the ROC curve showed that only platelet counts on admission are not a good predictor of systemic bleeding during hospitalization in this specific envenomation, indicating that other parameters are involved in the development of systemic bleeding during hospitalization.
In this study, unclottable blood on admission was observed in more than half of Bothrops snakebite patients. The coagulation disorders observed in B. atrox snakebites patients can be explained by the presence of components of the venom with thrombin-like activity, which directly hydrolyze fibrinogen in fibrin [41][42][43] and pro-coagulants, which activate coagulation factors II and X [44,45], which results in the intravascular thrombin generation. Other coagulation factors activated by components isolated from B. atrox venom are the factors XIII [34] and V [46]. In addition, B. atrox venom components that present fibrin(geno)lytic activity can also contribute to coagulopathy [47]. Unclottable blood on admission was independently associated with the development of systemic bleeding during hospitalization in this study.
We observed a significative association between thrombocytopenia and unclottable blood on admission in Bothrops snakebites patients who developed systemic bleeding during hospitalization. Nonetheless, the poor results found in the discriminative analysis (ROC curve) prompts us to believe that these parameters alone are not the only ones involved in the systemic bleeding phenomena shown here. This corroborates with findings regarding B. jararaca snakebites patients, which suggest that systemic bleeding may occur in patients with coagulable blood and thrombocytopenia and that coagulation disorders are not therefore the primary cause of systemic bleeding [27]. Similar reports have been described with B. atrox snakebites in Colombia [20]. P-III SVMP is the markedly predominant toxin of the B. atrox venom, which comes from the snake responsible for the majority of snakebites in the Brazilian Amazon [12]. Interestingly, a PIII-SVMP called Batroxrhagin, the major component of B. atrox venom, is highly similar to Jarharagin, and is able to inhibit collagen-induced platelet-aggregation, hydrolyze fibrin, and is highly hemorrhagic [39]. P-III hemorrhagic SVMPs induce hemorrhage by their accumulation at the basement membrane by binding to collagens [48,49]. P-I SVMPs, such as batroxase [47] and atroxlysin-I [50], that induce bleeding through hydrolysis of extracellular matrix components have also been isolated from B. atrox venom. In addition, it has been observed in Bothrops snakebites that haemorrhagins present in snake venom cause local haemorrhage and systemic bleeding by the direct action on components of the basement membrane of capillaries [15,51]. Serum haemorraghin levels were significantly higher in patients with clinical signs of systemic bleeding than those without in envenomation by B. jararaca snakes. Haemorraghin levels were also correlated inversely with platelet count in these patients [15]. In our study, high serum haemorraghin levels probably also contributed to the development of systemic bleeding in Bothrops snakebites in the Brazilian Amazon. However, these levels were not determined. Likewise, changes in platelet function were not evaluated.
Another aspect to be considered in this study was antivenom therapy of these Bothrops snakebites. Bothrops antigen used in Brazil to immunize horses in order to produce the antivenom consists of a mixture of venoms of B. jararaca (50%), B. alternatus (12.5%), B. moojeni (12.5%), B. neuwiedi (12.5%), and B. jararacussu (12.5%) (manufacturer's data). B. atrox venom is not included in the immunization pool used to produce Bothrops antivenom. Nevertheless, the results suggest that venom-induced systemic bleeding and unclottable blood were able to be ceased in most patients after 24 h of antivenom administration. These results were also found in other studies on Bothrops snakebites in the Brazilian Amazon [7,19] and other regions of Brazil [22]. On the other hand, it has been noted that Bothrops antivenom is less effective for neutralizing factor X activation activity of B atrox venom when compared to prothrombin activation activity [52]. In contrast, the frequency of thrombocytopenia increased in the first 48 h after antivenom administration, though decreased on discharge, while the frequency of abnormal mean platelet volume showed a decrease on discharge. Studies propose that venom-induced oxidative stress could lead to thrombocytopenia by apoptosis [36]. In Bothrops jararaca and B. jararacussu snakebites, it was observed that Bothrops envenomation promotes persistent and pronounced oxidative stress in the blood of the victims up to 1 month after hospitalization [53], which could explain the thrombocytopenia observed in our study up to 48 h after antivenom administration.
In conclusion, systemic bleeding observed in Bothrops snakebite patients in this part of the Amazon was an important systemic effect of the envenomation. The prevalence of systemic bleeding in this study was similar to that of others from the same region, but lower than that of other Bothrops spp envenomations. Thrombocytopenia and unclottable blood detected on admission were independently associated with the risk of developing systemic bleeding during hospitalization, which suggests that these hemostatic disorders increase the chances of systemic bleeding. Moreover, the effect of other parameters such as some toxins components of the venom can be involved in the development of systemic bleeding. Prospective studies are needed to elucidate the pathophysiology of systemic bleeding in Bothrops snakebites in the Brazilian Amazon and identify the different factors involved in its development as well as on the therapeutic response of hemostatic disorders.

Study Design and Data Source
This was an observational, cross-sectional study designed to assess the prevalence of and factors associated with systemic bleeding in Bothrops snakebite patients. Clinical-epidemiological and laboratorial data were obtained from patients' records with clinical-epidemiological diagnosis of Bothrops envenomation who were attended to at the Fundação de Medicina Tropical Dr. Heitor Vieira Dourado (FMT-HVD), a specialist hospital for snakebites in Manaus, capital of the Amazonas State, Brazil, between August, 2013 and July, 2016. Eligible cases were those with clinical signs of Bothrops envenomation. Patients that underwent previous antivenom therapy in other health service centers were not included in this study. According to the Brazilian Ministry of Health Guidelines for Snakebite Diagnosis and Treatment [54], envenomation by Bothrops species is classified as follows: (a) Mild, characterized by pain and mild or absent local edema, mild or absent systemic bleeding, with or without change in coagulation time; (b) moderate: Characterized by pain and evident edema involving three or more segments of the affected limb, accompanied or not by systemic bleeding and coagulopathy; (c) severe: Characterized by edema involving the entire affected limb, usually accompanied by severe pain. Systemic manifestations such as hypotension, shock, oliguria/anuria, or severe bleeding are defined as a severe case, regardless of the local effect. This study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of the FMT-HVD, Manaus, Brazil (approval number: 1.433.431, approval date: 2 March 2016). Term of commitment of use of data was signed by the researchers.

Clinical-Epidemiological and Laboratorial Parameters
The clinical-epidemiological variables analyzed were gender, age (in years), geographical location of the occurrence of the snakebite, type of area where the snakebite occurred (rural or urban), work-related bite, occupation, time taken to reach medical assistance (in hours), anatomical region of the bite, pre-hospital treatment (use of topical or oral medicines, use of tourniquet and other procedures), Bothrops bite confirmation by snake, local and systemic manifestations, clinical severity of envenomation (mild, moderate, or severe according to the Brazilian Health Ministry guidelines), presence of comorbidities, time spent in hospital (in days), and outcome (discharge or death). Analyzed laboratory variables were clotting time (in minutes), hemoglobin (mg/dL), hematocrit (%), platelet count (number/µL), and mean platelet volume (fL). The snakes were identified by a trained biologist at the FMT-HVD. All variables were checked by two independent researchers before analysis and further investigated for possible association with systemic bleeding as a dependent variable.
Snakebite-induced coagulopathy was evaluated by a modification of the Lee-White clotting time (LWCT) [55] which is recommended by the Brazilian Ministry of Health. With the use a plastic unlubricated syringe, 1 mL of venous blood was collected and placed into a new glass tube (13 × 75 mm) without any anticoagulants at 37 • C. Using a stopwatch, the timing started as soon as the blood was drawn into the tube. The glass tube was left undisturbed for 5 min and then checked for clots every following minute by gently tilting the tube. Unclottable blood was defined when the blood was not clotted until 30 min. The sensitivity of the LWCT performed in routine clinical settings for the diagnosis of hypofibrinogenemia is 78% [56]. Thrombocytopenia was defined by a platelet count below 150,000/µL. Low hemoglobin was defined when values were lower than 12.5 g/dL. Low hematocrit was defined when values were below 36%. Low and high mean platelet volume was defined when values were below 7.4 fL and above 10.4 fL, respectively. In order to identify clinical-epidemiological and laboratorial variables collected on admission associated with systemic bleeding, patients evolving to systemic bleeding on admission or during hospitalization were classified as a case, and one with non-systemic bleeding was included as a control. All patients were asked to return to the hospital for clinical reassessment in case of complications at any time, or on day 7 after discharge.

Statistical Analysis
A database and descriptive analyses were produced using Microsoft Excel Office 365 software which was fed information by two independent individuals. Statistical analyses were made using the software STATA statistical package version 13 (Stata Corp, College Station, TX, USA) and Graphpad Prism version 5 (Graphpad Software, Inc., San Diego, CA, USA). The Mann-Whitney test was used for comparison of medians. Differences in results were considered statistically significant when p < 0.05. The Pearson correlation coefficient was calculated to measure the degree of association between the platelet count and mean platelet volume on admission. The performance of platelet counts on admission to predict systemic bleeding during hospitalization in Bothrops snakebite patients was assessed using ROC curve.
Proportions of systemic bleeding during hospitalization were compared by Chi-square test (corrected by Fisher' test if necessary); differences were considered statistically significant for p < 0.05. The crude Odds Ratio (OR) with its respective 95% Confidence Interval (95% CI) was determined considering systemic bleeding during hospitalization as the dependent variable. Logistic regression was used for the multivariable analyses and the adjusted Odds Ratios with 95% CI were also calculated. All variables associated with the outcome at a significance level of p < 0.20 in the univariate analysis were included in the multivariable analysis. Statistical significance was considered if p < 0.05 in the Hosmer-Lemeshow goodness-of-fit test.