ALBERT

Description: Abstract 3326 Immune thrombocytopenia (ITP) is a relatively prevalent disease in dogs with significant morbidity and mortality. Canine ITP is clinically analogous to human ITP, with heterogeneity in bleeding manifestations in individuals with similar platelet counts. With a view to ultimately investigate this bleeding heterogeneity, we set out to develop a canine model of ITP. There are currently no existing large animal models of ITP. An induced canine ITP model would be representative of ITP without the confounding co-morbidities seen in clinical cases. Since spontaneous ITP occurs in both dogs and humans, the dog is an ideal translational model. We hypothesized that 2F9, a murine IgG2a monoclonal antibody to the canine platelet glycoprotein GPIIb (a common target of autoantibodies in ITP), would induce predictable dose-dependent thrombocytopenia (TCP) in healthy dogs. 2F9 had not been previously administered in vivo. We produced highly purified 2F9 and αYFA antibodies from the 2F9 hybridoma (gift of David Wilcox, Blood Research Institute, Wisconsin) and an isotype control murine anti-yellow fever antibody (αYFA) hybridoma. A dose titration (2 dogs) and a dose repeatability study (3 dogs) were performed in healthy adult research dogs by repeated intravenous infusion (≤ 6 doses) of 2F9 antibody until a target nadir of 5–30 × 103 platelets/μl was reached. Platelet counts were performed hourly until the platelet count reached the desired nadir range (t=0 hrs), after which complete blood counts were performed at 2, 4, 6, 8, 12, 24 hours, then q 24 hours for 10 days. The following were evaluated throughout the study: physical examination, buccal mucosal bleeding time (BMBT, baseline and t=0 only), serum cytokines and chemokines (INFγ, Interleukin (IL) 2, 6, 7, 8, 10, 15, 18, KC, IP-10, MCP-1, GM-CSF, TNFα; Milliplex CCYTOMAG-90K), fibrinogen, and D-dimers. Specificity of the 2F9 effect was confirmed by IV infusion of the isotype control (αYFA) to 3 dogs at the highest cumulative effective dose of 2F9 (167 μg/kg); all parameters were measured as above (t=0 hrs was one hour after αYFA dosing). Within 2 hours of a median cumulative 2F9 administration of 63 μg/kg (range 50.0–166.6 μg/kg), all dogs developed profound TCP (range 11–28 × 103/μl). Compared to the control group, platelet nadir was significantly lower (median (range): 6 (4–11) × 103/μl vs. 200 (179–209) × 103/μl; p= 0.036) and change in platelet count from baseline to nadir was significantly greater in the 2F9-treated group (median (range): 238 (179–325) × 103/μl vs. 4 (0–10) × 103/μl; p=0.036) (Fig 1); p-values were calculated using the exact Wilcoxon rank-sum test. Platelet nadir was in our target range and platelet count remained 〈 40 × 103/μl in all 2F9-treated dogs for 24 hours. Dosing was predictable: in each dog, after an initial dose of 50 μg/kg 2F9, the second dose needed to reach the target nadir could be accurately calculated from the initial platelet decrease. 2F9-treated dogs developed a range of clinical bleeding from none to petechiae, ecchymoses, melena, and hematuria. At t=0 hrs, BMBT increased 3–8 fold in treated dogs, compared to 〈 2 fold in control dogs. Dogs had no changes in vital signs or demeanor and did not require any transfusion support. The model does not appear pro-thrombotic as fibrinogen and D-dimers were similar over time in 2F9-treated vs. control dogs. 2F9 infusion also generated negligible systemic inflammation, as assessed by white blood cell count and serum cytokine measurement. Unexpectedly, however, serum IL8 tracked faithfully with platelet count, demonstrating that platelets are a major source of serum IL8 in dogs (Fig 2). Although α granules are known to contain IL8, platelets have not been previously described as a significant serum IL8 source. Since IL8 is an important neutrophil chemokine, our finding may illuminate a novel mechanism of platelet-neutrophil cross-talk. In summary, we have developed a novel large animal ITP model that is highly representative of the spontaneous disease. Like naturally-occurring ITP, dogs demonstrate bleeding heterogeneity despite similar platelet counts (data not shown). We expect our model to lead to further insights into bleeding mechanisms in ITP. Ultimately, understanding what factors predispose certain patients to bleed will allow us to exploit these factors therapeutically as novel ITP treatments. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 2193 In both canine and human patients with Immune Thrombocytopenia (ITP), bleeding risk is challenging to predict, and potentially leads to over-treatment of patients at low risk. Conversely, recent studies have highlighted the risk of thrombosis in ITP during platelet recovery. Given these clinical observations, we hypothesized that in ITP, changes in platelet response to agonists may occur in addition to changes in platelet numbers. In response to dual agonist activation (thrombin and convulxin), a subpopulation of platelets in both humans and dogs develops enhanced procoagulant activity. This subpopulation is termed coated platelets, and differences in individuals' potential to form coated platelets have been correlated with both hemorrhagic and thrombotic outcomes. In this exploratory study, we serially evaluated ex vivo platelet responsiveness to both thrombin and dual agonists (termed coated platelet potential) in a novel canine model of ITP. Dogs (n=4) were infused with a murine monoclonal anti-GPIIb antibody (2F9) in order to model ITP and generate predictable severe thrombocytopenia. Control dogs (n=3) were infused with a control antibody. Platelet count, thrombin responsiveness, and coated platelet potential were measured at baseline, time zero, 6 hours, 24 hours, and every 24hrs thereafter until the platelet count was ≥ baseline for at least two consecutive measures (recovery). Time zero was defined as the time when platelet count first fell to ≤ 30,000/μl following 2F9 infusion, or 1 hour following control antibody infusion. For platelet thrombin responsiveness, a monoclonal antibody to P-selectin was used to determine platelet P-selectin surface expression by flow cytometry after stimulation with graded doses of thrombin. The ED50 Thrombin was defined as the concentration of thrombin required for half-maximal P-selectin expression. Coated platelet potential was defined as the percent of platelets activated to the highly procoagulant state after dual stimulation with thrombin and convulxin, as determined by binding of biotinylated fibrinogen by platelets by flow cytometry. All dogs in the treated group developed severe thrombocytopenia (median=6×103, range=4–11×103 platelets/uL); no dogs in the control group developed thrombocytopenia. All treated dogs had platelet recovery by 240 hours (median=132 hours, range 120–240hours). Of interest, at 6 hours, ED50 Thrombin in the treated group increased nearly twofold (fig 1A) (ratio of median ED50 Thrombin treated/baseline=1.6, range 1.3–2.3), which correlated with a decline in coated platelet potential by nearly half of baseline (fig 1B) (median 52.4% of baseline, range 19.6–61.5%); minimal change from baseline was observed in controls. In both groups, ED50 Thrombin was lower at recovery than baseline (fig 1A) (treated median ED50 Thrombin=71.5% of baseline; control median ED50 Thrombin=67% of baseline). A trend of rising coated platelet potential was also noted as platelets recovered in the treated group. In conclusion, in this exploratory study of a canine model of ITP, we observed dynamic changes in platelet responsiveness. During severe thrombocytopenia, we observed a rise in ED50, indicating a decline in response to thrombin, which correlated with a fall in coated platelet potential. We speculate that this early fall in platelet thrombin response and coated platelet potential could contribute to hemorrhage risk in ITP. As a complement to this finding, in the treated group, there was a rise in coated platelet potential as platelets rebounded and coated platelet potential was slightly greater than baseline at recovery. This is consistent with others' observation that younger platelets are more likely to have coated platelet potential. We also observed a decline in ED50 Thrombin at recovery, not only in the treated dogs, but also control dogs. Thus, at recovery, the decline in ED50 Thrombin was independent of treatment group. However, this may be an artifact of our small sample size. Our observed increase in coated platelet potential during platelet recovery could potentially contribute to the thrombotic tendency of some ITP patients. Future studies are planned to explore the relationship of hemorrhagic and thrombotic risk with platelet thrombin responsiveness and coated platelet potential in this model of ITP and clinical studies of canine and human ITP. Disclosures: No relevant conflicts of interest to declare.

Description: Pet dogs experience similar environmental influences to their owners, and thus provide valuable spontaneous models for investigation of novel therapies for human disease. DNase and treatments targeting neutrophil extracellular trap (NET) proteins show promise in murine thrombosis models. Dogs at risk of thrombosis could provide evidence for the efficacy of these treatments in a second species, but the pro-thrombotic effects of NETs on canine plasma must be confirmed. We isolated neutrophils from healthy canine blood by density gradient and stimulated NET release using phorbol 12-myristate 13-acetate (PMA). To release NETs into suspension, culture media was replaced by the restriction enzyme Alu1. Supernatants were centrifuged to remove any intact cells. By gel electrophoresis, NET fragments were predominantly 600-1000 base pairs which is sufficient to simultaneously bind plasmin and fibrin (Arterioscler Thromb Vasc Biol., 2015, 35: 2544-53). To distinguish the effects of DNA and NET proteins, a subset of NET fragments were further digested with DNase. Supernatants were identically prepared from non-stimulated neutrophils. DNA concentration was 106ng/ml (sd 49, n=5 dogs) for non-stimulated supernatants; 401ng/ml (sd 94) for NET supernatants and reduced to 113ng/ml (sd 18) by DNase. The effect of NETs on clot formation and lysis were measured spectrophotometrically. To bring fibrinogen into the range typically found in dogs suffering from pro-thrombotic disease, 4.2mg/ml of plasminogen-depleted fibrinogen was mixed with pooled platelet poor plasma from healthy dogs. Plasma was incubated with non-stimulated supernatants; NETs +/- DNase or controls for effects of restriction enzymes, PMA and culture media. We added tissue factor, 5mM calcium chloride and 100µg/ml tissue plasminogen activator and measured absorption at 405nm q. 10 seconds for 25 minutes. NETs increased maximum formation velocity (MFV) (192mOD/min sd. 12, p=0.0011), measured by a line of best fit for the upswing of the curve, and delayed time to 50% clot lysis (448s sd. 43) compared with non-stimulated supernatants (165mOD/min sd. 12; 395s sd 38; n=5). Using a repeated measures ANOVA followed by paired t-tests with Tukey's correction, the effect on MFV (p=0.0089) but not time to 50% clot lysis (p=0.08) remained significant after DNase digestion. NETs did not have detectable effects on area under the curve (AUC) or peak (pANOVA 〉0.2) (Fig. 1). Reagent controls had no detectable effects on any parameter (pANOVA≥0.717; 5-6 technical replicates). Reduction in fibrinolysis by DNA replicates its effect on human plasma (Arterioscler Thromb Vasc Biol., 2015, 35: 2544-53). The lack of effect of DNA on canine clot formation was surprising given the ability of DNA to increase human plasma thrombin generation (Arterioscler Thromb Vasc Biol. 2014, 34:1977-84). To determine if this was a species difference or a concentration dependent effect, DNA from canine leukocytes was added to plasma at a final concentration of 200ng/ml (approximately equal to the final concentration in our NET studies) or 1000ng/ml. At 1000ng/ml but not 200ng/ml, DNA increased MFV (p=0.038). To determine if NET proteins were increasing MFV by a direct effect on fibrinogen or thrombin, we next studied effects of NETs on thrombin induced clotting of plasminogen-depleted fibrinogen. NETs with or without DNase digestion had no detectable effects on peak, AUC, MFV or fibrin fiber diameter estimated by turbidity. This suggests NET proteins increase plasma clot formation rate through interactions with coagulation components other than thrombin or fibrinogen. Based on human and murine studies showing NET proteins inhibit endogenous anticoagulants, we selected histones and elastase as candidate proteins for further investigation. Purified histones from calf thymus added to plasma at approximately the concentration present in NET supernatants (150ng/ml) had no effect on MFV. Pre-incubation of elastase with canine plasma abolished elastase activity measured by N-Succinyl-Ala-Ala-Ala-p-nitroanilide. Therefore, it is unlikely that the ability of canine NET proteins to increase MFV is mediated by elastase or histones alone. In conclusion, NETs increase clot formation rate and delay lysis in canine platelet poor plasma. Since effects were mediated by both proteins and DNA, dogs are a promising model for trials of both DNase and therapies targeting NET proteins. Disclosures No relevant conflicts of interest to declare.

Description: Canine immune thrombocytopenia (ITP), unlike mouse models of induced anti-platelet immunity, occurs spontaneously in non-inbred populations. Canine ITP demonstrates the same spectrum of bleeding severity and variable response to front-line therapy as the human disease counterpart. Pet dogs and their owners live in close contact in a shared environment with the same potential exposures to environmental triggers of an immune response. Moreover, the breed structure and large litter size of purebred dogs facilitate familial studies to uncover genetic risk factors of autoimmunity that are impossible to perform in human populations. We performed a multi-institution cohort study of severe thrombocytopenia (defined as platelet count 〈 50,000/uL) in dogs to further develop the canine ITP model for comparative studies. At enrollment, bleeding severity was scored with a standardized bleeding assessment tool "DOGiBAT" based on the ITP bleeding scale, and samples were collected for routine clinicopathologic testing, immune profiling (cytokine analysis), and flow cytometric detection of platelet-bound antibodies (PSAIg). In the 61 dog sample set, immune-mediated platelet destruction was the most common cause of thrombocytopenia: 57% of cases had no inciting disease and were classified as primary immune (PI), 28% of cases were classified as secondary immune (SI) in association with lymphoid or myeloproliferative disease, drug exposure, or infectious disease, and the remaining 15% of cases had non-immune thrombocytopenia (NI) attributed to bone marrow aplasia or consumptive coagulopathy. The PI cases had a median age of 8 years, were 54% female, presented for petechiae and mucosal bleeding with high prevalence of PSAIg (76%) and had a 94% survival rate. These features are similar to the demographics, clinical findings, and low mortality of ITP in people. Among the significant differences between disease categories, PI dogs had higher DOGiBAT scores than NI dogs denoting more severe bleeding and both PI and SI cases had marked decreases in platelet CD61 expression compared to NI dogs (Figure 1, individual and group median values). Such aberrant platelet surface antigen expression has been described in childhood ITP. The surviving PI cases had a median duration of hospitalization of 4 days, with significant correlation between duration of hospital stay and admission DOGiBAT score (r2 = 0.18; p = 0.02). Serum cytokines and chemokines (Interferon (IFN) gamma, Interleukin (IL)-2, 6, 7, 8, 10, 15, 18, Canine orthologue of Chemokine C-X-C motif ligand 1 (KC), IFN-gamma-inducible protein-10, Monocyte chemoattractant protein-1, Granulocyte-macrophage colony-stimulating factor, and Tumor necrosis factor alpha) were measured with a multiplex assay (Milliplex CCYTOMAG-90K) in PI dogs and compared to those of healthy control dogs. Dogs with PI had significantly lower serum IL-8 compared to healthy control dogs (Figure 2). A negative correlation between platelet count and serum IL-8 has previously been reported in human ITP. We also performed SNP genotyping for a genome-wide association study of ITP in purebred dogs. Initial analyses of 25 PI cases and 14 controls revealed over-representation of a dog leukocyte antigen (DLA) allele variant in PI dogs that was recently found to be associated with canine autoimmune disease in an independent study. Furthermore, we found a significant association between PI dogs and a novel locus on canine chromosome 1 (Figure 3, Manhattan plot). Together, these preliminary phenotypic and genotypic studies reveal that canine ITP represents a new translational model to discover genetic loci and pathways associated with autoimmunity, biomarkers of primary and secondary immune targeted platelet destruction, and for proof of concept treatment trials utilizing novel therapeutic approaches. Disclosures No relevant conflicts of interest to declare.