Fatal pulmonary embolism following splenectomy in a patient with Evan’s syndrome: case report and review of the literature

Splenectomy is utilized in the diagnosis and treatment of a variety of hematologic disorders unresponsive to medical management in hopes of reducing the frequency of relapse and lowering the maintenance dose of steroids. Splenectomy, when performed for hematologic disorders, is associated with a complication rate as high as 40% [15]. Complications include perioperative bleeding, overwhelming infection with encapsulated organisms and a growing body of evidence is showing a heightened risk of peri-operative and post-operative venous and arterial vascular events, especially venous thromboembolism. Splenectomy is a known risk factor for VTE independent of the underlying indication (i.e.: malignancy) or comorbidities (i.e.: antiphospholipid syndrome) [16, 17]. It has an associated 5-fold increased risk of fatal PE [18]. In addition, warm autoimmune hemolytic anemia (w-AIHA) is associated with a heightened risk of VTE with an incidence of 15–33% [19‐22]. Risk appears to be higher in patients with a more severe hemolytic process highlighted by lower hemoglobin levels and marked hemolysis with higher reticulocyte counts, bilirubin and LDH levels, and lower haptoglobin levels. In these studies, splenectomy did not appear to increase the risk of VTE relative to the baseline risk associated with w-AIHA [22, 23]. The incidence of thromboembolic complications following splenectomy when specifically performed for hematological diseases has been reported to range from 10 to 30% which includes PSVT, PE, and DVT [17, 24]. Despite these reports, VTE risk following splenectomy in the background of w-AIHA, or related hemolytic processes, has not been reported in depth in the literature.

In the original case series of patients having undergone splenectomy for ES, AIHA, or ITP, there was no report of VTE [1]. In another case series of 12 patients with ES, one patient died of pulmonary embolism 6 months after diagnosis, however this patient did not undergo splenectomy. Three of the 12 underwent splenectomy, none of whom had reported post-procedural venous thromboembolic events [2]. In similar cases with AIHA and ITP, there were no cases of post-splenectomy VTE [25, 26]. However, evidence for the association of VTE with splenectomy unfolded over ensuing decades. In a case series of patients with hemolytic anemia, 13% of patients developed thromboembolic complications following splenectomy, and two patients died because of VTE [27]. In one large population based study looking at incidence of thromboembolism in post-splenectomy ITP the procedure was associated with a 4.3% higher cumulative risk of VTE [28]. A Dutch study reviewing 563 splenectomies, identified 45 cases performed for hemolytic anemia, of which, 9% (4 out of 45 patients) acquired portal vein thrombosis [29]. A retrospective chart review reported 3 out of the 6 patients who underwent laparoscopic splenectomy for AIHA developed thrombosis in the portal venous network. Each patient also had post-splenectomy thrombocytosis with platelet counts ranging from 500 K/mm³ to 896 K/mm³. One of these patients also developed a concomitant DVT. None of them had received prophylactic anticoagulation during the perioperative period [30]. Mohren et al. [31] highlight events in a case similar to ours in which a 24-year old female patient with AIHA who underwent splenectomy for continued anemia despite medical therapy. She received appropriate prophylactic dose anticoagulation during her hospital say. After discharge, she developed extensive pulmonary embolism, portal vein thrombosis, and mesenteric vein thrombosis. She subsequently died due to bowel ischemia. The platelet count reported at the time of death was 499,000/L.

Reit [29] and Winslow [32] have individually demonstrated that hemolytic anemia is an independent risk factor for PSVT-- independent of splenectomy and thrombocytosis. Systemic venous thrombosis has also been observed in patients with hemolytic syndromes who have undergone splenectomy. In a case series of 47 patients with AIHA (6 of whom had ES), venous thromboembolism was the most common cause of death. Each patient who had succumbed to VTE had undergone splenectomy during their treatment and was on corticosteroid therapy at the time of death. None of the cases of Evan’s Syndrome was associated with a thrombotic complication [33]. The risk for thrombosis extends beyond immune-mediated hemolytic processes. In a large case-series looking at Thalassemia intermedia patients, this hemolytic process was associated with an increased incidence of venous thromboembolism beyond the portal circulation (32% DVT, 16% PSVT and 13% PE) [34]. In addition, there have been numerous case reports of thrombosis after splenectomy in patients with hereditary stomatocytosis, a hemolytic syndrome characterized by anemia secondary to abnormalities in red cell cation permeability. In fact, post-operative thrombosis is so prevalent that splenectomy is no longer recommended in the treatment of hereditary stomatocytosis [35].

The burden of disease is further complicated by the risk for post-splenectomy reactive thrombocytosis (RT). This phenomenon is common especially in patients with splenomegaly, occurring in 75% of cases [36], and is characterized by a mild asymptomatic thrombocytosis reaching a peak at the end of the 2nd week and subsiding by 3 months post-splenectomy [26]. Typically, RT is not considered a risk factor for thromboembolic complications, as only 1.6% of patients had thrombotic event in only large case-series, with a majority occurring in setting of other significant risk factors including post-surgery and malignancy [37]. Therefore, reactive thrombocytosis following splenectomy carries a more modest risk. One study identified a 5% risk of PSVT [38]. Even in cases of extreme reactive thrombocytosis (platelet count >1,000,000/mm³), the risk of thrombotic complication is relatively modest (4–5%) [39]. In fact, increasing platelet count has inconsistently shown a correlation with risk of thrombosis [40]. What seems to be a greater driver of post-splenectomy thrombotic risk is the evidence of ongoing hemolysis. When splenectomy is effective at ameliorating hemolysis, thrombocytosis is usually milder and the risk of thrombosis is low. However, with ongoing hemolysis, thrombocytosis may persist, and is associated with a higher thrombotic potential [27] (Table 1).

Proposed mechanisms of thrombosis are vast and include local and systemic factors. These local and systemic factors contribute to all thrombosis, portal/splenic vein thrombosis, deep vein thrombosis, and pulmonary embolus, alike. Local factors and systemic factors that promote thrombosis are centered upon Virchow’s triad. Local factors center around hemostasis, while systemic influences include, but are not limited to, platelet abnormalities and the post-operative inflammatory state which leads to hypercoagulability, endothelium injury, and platelet activation.

Stasis of blood in the splenic vein remnant appears to be a primary causative factor for PSVT [29, 41]. Splenomegaly is a risk factor for stasis of blood. This has been attributed to the relationship of spleen mass to size of splenic vein stump. Increasing size of the splenic vein stump promotes venous stasis which enhances the risk for thrombosis [42]. Danno et al. [43] confirmed that a pre-operative splenic vein diameter greater than 8 mm was an independent predictor of PSVT. Similarly, the higher incidence of PSVT noted with removal of heavier spleens have been attributed to longer surgical time and prolonged increased intraabdominal pressure. Prolonged exposure to increased intraperitoneal pressures during laparoscopic procedures promotes venous stasis within the portal venous network by decreasing portal blood flow [44]. A recent prospective study identified a 55% higher incidence of PSVT noted with laparoscopic compared to open splenectomy [45]. This contradicts reports by Winslow [32] (5% in laparoscopic vs 9% in open), Chaffanjon [42] (6.7% with open), and Valeri [46] (8.3% with open), and was not seen in the animal study done by Lyass [47]. It is likely that these retrospective studies are biased based on patient selection and that the etiology of thrombosis is likely multifactorial.

The systemic mechanisms that may promote VTE in splenectomized patients are unclear, but a vast number of systemic influences have been proposed to promote thrombosis, including endothelial activating factors, increased platelet activation and/or released microparticles [35, 48]. Further pro-thrombotic mechanisms have been proposed, primarily centered on the continued intravascular hemolysis despite splenectomy. The spleen primarily functions in filtering and removing unwanted bloodstream products which can include infectious organisms, cellular debris, senescent cells, or immunologic-marked cells. In the absence of this inherent phagocytic property, ongoing hemolysis allows thrombogenic microparticles formed from erythrocyte fragmentation to remain in the bloodstream. Hemolysis in the absence of splenic sequestration also promotes hyperviscosity secondary to retained intracellular inclusions (i.e.: hemolytic byproducts). Persistent hemolysis also drives activation of vascular endothelium which increases platelet adherence. Lastly, ongoing hemolysis is a stimulus for thrombopoiesis. When thrombopoiesis exceeds the potential for platelet removal, thrombocytosis results. Independent of hemolysis, persistence of abnormal erythrocytes in circulation is also postulated to contribute to thrombosis [35]. Further, adhesive platelets’ pre-hemolytic thromboplastic activity and significant production of anti-plasmin as a result of splenectomy are additional proposed factors contributing to post-splenectomy thromboembolism [17].

Qualitative and quantitative platelet abnormalities also promote a thrombogenic potential. Gordon et al. [49] described post-splenectomy PSVT associated with these platelet abnormalities in the setting of myeloproliferative and hemolytic disorders. These platelet anomalies can originate in several fashions. First, splenectomy, by itself, can precipitate a rise is inflammatory cytokines which, in turn, mediates a hypercoagulable state. In addition, the absence of the spleen gives rise to high levels of markers of persistent platelet activation, namely platelet microparticles (PMP) and CD26P [41]. This promotes thrombosis independent of the platelet count. As discussed previously, there is mixed evidence in support of relationship between higher platelet counts and risk for thrombosis [40]. Despite this, it is most evident that a majority of thrombotic events occur with platelet counts >600,000/mm³ [50]. The exact mechanism(s) that drives clotting vs. bleeding associated with thrombocytosis has not been elucidated.

Reticulated platelets (RP) are the high ribonucleic acid (RNA) containing young platelets in circulation and are an index of thrombopoiesis. Reticulated platelet percentage (RP%) and absolute RP count have been shown in a study [51] to be significantly increased in the state of transient thrombocytosis, including the post-splenectomy setting. Moreover, higher RP% and absolute RP count had a positive predictive value of 38% and 88%, respectively, for developing symptomatic thrombosis. This study was unable to show a causal relation of secondary thrombocytosis as a prothrombotic state. A similar study by Ryningen et al. [52] confirmed the above finding and also showed a significantly elevated RP% in immune thrombocytopenic purpura. This was also proposed in the surgical literature [29, 32]. Considering the above studies, patients with secondary Evans syndrome associated with an underlying autoimmune phenomenon who undergo splenectomy are undoubtedly at a profound risk for developing thrombotic complications. RP% and absolute RP count if validated in prospective trials may help to guide therapy in patients who are at higher risk of developing thrombosis.

Lastly, high dose oral steroid treatment, which forms the first line treatment of ES, is increasingly being recognized as a prominent risk factor in producing a prothrombotic state [53, 54]. Stuijver et al. [55] demonstrated the risk of PE was highest in the first 30 days of glucocorticoid exposure and is associated with a 10-fold increased risk in patients receiving higher doses of steroids, prednisolone >30 mg, or its equivalent.

Due to lack of sufficient data in patients with Evans syndrome, most of the treatment strategies are extended from studies done in essential thrombocythemia (ET), ITP, and various chronic autoimmune hemolytic anemia syndromes. Corticosteroids are considered the first line therapy which our patient received. IVIG and Rituximab are found to be especially useful in patients with refractory ITP or chronic severe steroid dependency. In our patient, each of the above treatment modalities were implemented. Given the lack of response, splenectomy was rightfully chosen as the next best step. Knowing splenectomy and steroids both carry a heightened risk for thrombotic events, it is important to always consider both pharmacologic and mechanical DVT prophylaxis.

Low dose aspirin therapy, 81 mg, is standard of care in low to moderate risk ET, defined as presence of JAK2 mutation or cardiovascular disease but age < 60, platelet count less than 1,500,000/L, and no history of thrombosis. Aspirin use has been associated with improved outcomes in the setting of secondary causes of thrombocytosis [39]. Therefore, many are advocating for standard use of aspirin in those with RT. Aspirin is indicated as a standard therapy in the comorbid cardiovascular conditions regardless of RT but in the presence of RT especially in older patients, it may be protective. This lends evidence that aspirin use for RT in the presence of other thrombotic risk factors may prove to be beneficial in elderly patients but there is still no reliable approach to predict who would benefit from aspirin.

In higher risk ET, aspirin is used with a cytoreductive agent such as hydroxyurea or anagrelide. As per a large randomized study [56] performed in patients with ET who were at high risk for thrombosis, anagrelide plus aspirin had a lower rate of venous thrombosis but a higher rate of arterial thrombosis as compared to hydroxyurea plus aspirin. As the incidence of venous thrombosis in untreated patients with high risk ET is unknown, it was unclear if this rate was increased by hydroxyurea. It is well known that the incidence of arterial thrombosis is more than three times the venous thrombosis in ET. The same is unfortunately not well established in case of RT and it is difficult to extrapolate this data to our patient with RT. Granted, in our patient, hydroxyurea was used as a cytoreductive therapy and theoretically could have contributed to venous thrombosis. More data needs to be collected to better elucidate the risks of hydroxyurea vs anagrelide in regards to thrombosis.

Melphalan has been used in hemorrhagic thrombocythemia post-splenectomy as a cytoreductive agent and was proven to be effective in controlling the platelet proliferation thereby reducing the bleeding complications in some situations [57]. The role of Melphalan has not been investigated in depth over the ensuing decades and could be studied further in patients with RT to prevent thrombotic events and/or be used as a more effective agent compared to hydroxyurea or anagrelide.

Plateletpheresis has been shown to be a useful method to rapidly decrease the platelet counts especially in patients with acute thromboembolic events of neurologic or pulmonary origin [58]. Perioperative plateletpheresis was used during microsurgical free tissue transfer in a patient with post-splenectomy thrombocytosis [59]. One of the largest reviews reported more symptomatic events in patients with platelet counts greater than 1,500,000/mm³ and treatment benefit reached statistical significance in patients with ET but not in patients with post splenectomy reactive thrombocytosis. Another small case series [60] of five patients with underlying myeloproliferative neoplasm (MPN) and complications of thrombocytosis reported beneficial effects of plateletpheresis. Most patients required at least 2 to 3 procedures for control of platelet counts. This procedure does come with its own risks of bleeding, infection, arrhythmias, hypotension, and cost. There is lack of controlled trials to make any recommendation for or against this procedure especially in patients with RT. As discussed previously, the existing literature is inconsistent in its association between platelet count and thrombosis. Plateletpheresis should therefore not be used as a prophylactic measure. Its use in acute events is questionable and shouldn’t be used in the first line setting. However, its use can be considered in acute, symptomatic, uncontrolled events refractory to other cytoreductive agents.

A great unanswered question in regards to treatment is how best to prevent thrombosis. As discussed in previous sections the inherent risks factors for thrombosis following splenectomy are extensive. Our patient had several inherent risks for VTE including morbid obesity, prolonged hospitalization with bed rest, underlying hemolysis, sepsis, and major surgery. He also had central venous access for much of the hospital stay. The 2005 version of the Caprini Risk Assessment Model [61] is the most widely used and well-validated risk prediction for postsurgical patients and is referenced in the ACCP guidelines [3]. The updated version of this model stratifies risk into five categories- lowest, low, moderate, high, and highest. Our patient had a Caprini VTE Risk Assessment score of 8 which carries a 6.0% risk of VTE without anticoagulation. With this risk, the Caprini model advocates for out of hospital use of anticoagulation for at least 7–10 days [62]. In addition, the incidence of thromboembolic complications following splenectomy when specifically performed for hematological diseases, as previously mentioned, has been reported to range from 10 to 30% [17, 24]. With this added risk, Caprini would advocate for 30 days of anticoagulation.

Prophylactic anticoagulation is used outside the hospital in various settings including after orthopedic surgery [63], following major abdominal surgery in setting of cancer [64], following bariatric surgery [65], and after spinal cord injuries [66]. With the inherent risks associated with splenectomy for hematological diseases, and a high Caprini VTE risk score, our patient may have benefited from outpatient anticoagulation. Prophylactic subcutaneous heparin alone is known to be insufficient to prevent the development of PSVT [32]. Prior recommendations have included combination therapy with antiplatelet agents, warfarin, and heparin. Our patient’s BMI exceeded 40 kg/m². Caution must be exercised when offering prophylactic anticoagulation for patients with extreme BMI’s as high and low BMI’s are not well-represented in clinical trials. Use of Lovenox 40-60 mg twice daily for 30 days may have prevented our patient’s fatal outcome [67].

Unpredictable use of pharmacologic prophylaxis stems from fear of bleeding. Bleeding risk associated with RT is not well-described in the literature. It is important to note that the platelets in RT are normal and therefore it carries a theoretical lower risk for bleeding compared to ET, as in ET, platelets are abnormal. An acquired Von Willebrand factor deficiency has also been described in ET. This is associated with increased risk of bleeding [58]. Again, this is not reported in patients with RT. Because of the known risks of bleeding with ET, many of the same fears are carried over to RT. This has likely contributed to the inconsistent use of anticoagulation in these patients. In addition, associated pre-operative thrombocytopenia and severe anemia, as seen in our case, offers further reluctance to use anticoagulation. However, thrombocytopenia does not always correlate with risk of thrombosis. In fact, the rate of VTE in hematologic malignancies, where platelet counts are often profoundly low, mirrors the frequency seen in solid tumors [7]. This has prompted one group to suggest decreasing dose of LMWH by 50% when platelet counts are below 50,000/mm³, as opposed to eliminating pharmacologic prophylaxis altogether in that setting [68]. Hence, diligent use of anticoagulation is of utmost important in patients with thrombocytopenia where the decision for or against pharmacologic prophylaxis should be made on a case-by-case basis.

Proper use of prophylactic dose anticoagulation needs to become a greater priority in these patients. However, prophylactic decisions like these remains difficult without consensus guidelines. Expert consensus panels currently do not address the use of prophylactic anticoagulation in the settings of post-splenectomy for hematological disease and/or reactive thrombocytosis [8‐14]. Without guidelines or randomized trials directing evidence-based use of anticoagulation, we must rely on case series and case studies. We encourage further efforts to investigate optimal length of therapy. It is possible anticoagulation needs to be extended several weeks beyond the hospitalization as directed by the Caprini risk assessment. Another proposal is to use anticoagulation until there is no further evidence of hemolysis which would require close follow-up with dedicated laboratory surveillance.