Short- and long-term risks of splenectomy for benign haematological disorders: should we revisit the indications?

Authors


Correspondence: Francesco Rodeghiero, Department of Cell Therapy and Haematology, San Bortolo Hospital, Via Rodolfi 37, 36100 Vicenza, Italy. E-mail: rodeghiero@hemato.ven.it

Summary

Splenectomy has represented a key treatment option in the treatment of many benign haematological diseases, including immune thrombocytopenia (ITP) and disorders associated with ongoing haemolysis (thalassaemia major and intermedia, sickle cell disease, and hereditary or acquired haemolytic anaemias). Improvements in surgical techniques have reduced perioperative complications and mortality. Preventive measures (new protein conjugate vaccines, antibiotic prophylaxis, and increased vigilance) are thought to greatly reduce the risk of overwhelming post-splenectomy infection (OPSI), although their implementation is inconsistent. Nevertheless, there is increasing documentation of the short- and long-term risks of splenectomy, which vary according to the underlying indication. Splenectomized patients are at increased risk of venous thromboembolism, particularly within the splenoportal system. The long-term thromboembolic risk is higher in haematological disorders associated with ongoing haemolysis, particularly in thalassaemia intermedia, which has led to a more conservative approach. In comparison, patients with ITP appear to be at lower risk of adverse effects of splenectomy, which maintains its place as the potentially most curative and safe second-line treatment. However, a splenectomy-sparing approach is also emerging for ITP, and recent guidelines recommend that this procedure is deferred until ≥12 months from ITP diagnosis, to allow sufficient time for possible remission.

Surgical splenectomy has represented a key treatment option for benign haematological disorders, including immune thrombocytopenia (ITP), haemolytic anaemias and particularly, thalassaemia. Improvements in surgical techniques, most notably the introduction of laparoscopic splenectomy in 1991 (Delaitre & Maignien, 1991), have greatly lowered the immediate risks of this procedure. However, a growing understanding of the role of the spleen, as well as the continuing emergence of data documenting the long- and short-term adverse events associated with splenectomy, require that the indication for this procedure is posed after more careful consideration of its relative merits and alternative treatments for the individual patient. It is now clear that the spleen is much less dispensable than was previously believed and that this organ, in addition to its blood-filtering functions, plays a pivotal role in innate and adaptive immunity (Mebius & Kraal, 2005; Cesta, 2006; Di Sabatino et al, 2011). The clinical relevance of the immunological function of the spleen was first highlighted by King and Shumacker (1952), who reported five cases of overwhelming post-splenectomy infection (OPSI), caused by encapsulated bacteria, in a series of 100 splenectomized children. Subsequently, in 1977, findings of a 28-year follow-up of 740 World War II veterans suggested an increased risk of mortality from ischaemic heart disease or pneumonia in those who had undergone splenectomy for war injuries (Robinette & Fraumeni, 1977).

This review examines the potential adverse events associated with splenectomy in patients with benign haematological disorders, in the light of recent evidence, and its role in the different disorders. We have focused mainly on ITP, thalassaemia major and intermedia, sickle cell disease, and hereditary spherocytosis, because these appear to be the areas of most active research over recent years. Data from other settings, including splenectomy for traumatic injury in otherwise healthy individuals and malignant haematological diseases, are also included, in order to provide a context.

A Pubmed search was conducted using the search terms ‘splenectomy’, ‘adverse events’, ‘complications’, ‘haematological disorders’, ‘immune thrombocytopenia’, ‘thalassaemia’, ‘sickle cell disease’, ‘hereditary spherocytosis’ and ‘guidelines’. Conference databases were also searched for relevant abstracts.

Overview of functions of the spleen

The spleen consists of three compartments, namely (i) the red pulp, a meshwork of splenic cords and venous sinuses, (ii) the white pulp, a reticular structure consisting of the peri-arteriolar lymphoid sheath (PALS) and follicles that surround central arterioles, and (iii) the marginal zone, which is located between the red and white pulp, outside the mantle layer of the lymphatic follicles.

Briefly, the red pulp functions as a blood filter, removing foreign material, including blood-borne micro-organisms, and aged and damaged erythrocytes, while providing a storage site for iron, erythrocytes and platelets. Its role in adaptive and innate immunity includes phagocytosis of antibody-coated cells or microorganisms; synthesis of immunoglobulin M by memory B cells; and production of the opsonins properdin and tuftsin. Opsonized bacteria are readily removed by macrophages within the spleen and liver, but poorly opsonized bacteria, such as encapsulated species, can be cleared only by the spleen. Within the marginal zone, binding and clearing of opsonized and nonopsonized bacteria and viruses takes place via several receptors, including Toll-like receptors on macrophages or dendritic cells. The white pulp is concerned with key aspects of adaptive immunity, including antibody production, antigen processing and preservation (Mebius & Kraal, 2005; Cesta, 2006; Di Sabatino et al, 2011).

Intriguingly, findings from a recent murine study suggest that the spleen may additionally be involved in storage and rapid deployment of monocytes to regulate inflammation (Swirski et al, 2009). It was previously thought that the bone marrow was the reservoir for these monocytes, but the investigators found that half of the body's monocytes were located within the splenic subcapsular red pulp of the mice. After induction of experimental myocardial infarction (MI), monocytes were found to exit the spleen and infiltrate and remodel the damaged myocardial tissue. These observations need to be validated in humans.

Infection and mortality

The term OPSI defines fulminating sepsis, meningitis or pneumonia mainly caused by encapsulated bacteria, such as pneumococci, meningococci and Haemophilus influenza type b (King & Shumacker, 1952; Schwartz et al, 1982; Ejstrud et al, 2000; Bisharat et al, 2001). Characterized by evolution in a matter of hours, in association with hypotension, alteration of consciousness or shock, OPSI is the major concern after splenectomy, with a high mortality risk of approximately 40–50% (Bisharat et al, 2001). The risk of OPSI and serious infections (here defined as those requiring hospitalization) is difficult to quantify, because of reporting biases and variability in selection and inclusion criteria, duration of follow-up and stratification of patients by age, underlying indication for splenectomy and operative technique (Di Sabatino et al, 2011), as well as implementation of vaccination or other prophylactic measures. Data sources for this section are summarized in Table SI.

Schwartz et al (1982) conducted a retrospective evaluation of patients who were splenectomized (n = 193) at the Mayo Clinic (Rochester, MN, USA) over a 25-year period. Most of the splenectomies were due to trauma (24%), incidental abdominal surgery for malignant (19%) or non-malignant disorders (27%) or haematological diseases including lymphoma (24%). Although most patients had severe underlying disease, only two cases of fulminant sepsis were recorded, for a cumulative incidence of 0·18 per 100 person-years, or one case for every 545 patient-years of observation. The first case resulted in the death of a young woman receiving intensive chemoradiotherapy for recurrent Hodgkin lymphoma who developed cryptococcal meningitis and fulminant Gram-negative sepsis. The second case was a young woman who developed Neisseria meningitidis bacteraemia 17 months after undergoing splenectomy for trauma. She subsequently recovered. The overall incidence of any type of serious infection following splenectomy was 7·16 per 100 person-years, with the highest incidence in patients undergoing incidental splenectomy in conjunction with abdominal operations for malignancies. These patients had a relative risk (RR) five times higher than splenectomized trauma patients. Previous radiotherapy, immunosuppression or chemotherapy greatly increased the risk in all patients.

An English-language literature review for the period 1966–1996 found that 209/6942 splenectomized patients developed invasive infection (crude rate 3%) and 106 (1·5%) died. The mean time interval between splenectomy and infection was estimated to be 22·6 months (Bisharat et al, 2001). The risk of infection or death was generally higher in children than in adults, depending on the underlying indication. The lowest risk was observed in ITP patients (n = 484; infection 2·1% and death 1·2%); indeed their risks were very similar to those who underwent splenectomy for trauma (2·3% and 1·1%). The highest rates of infection (12% and 9%) and all-cause mortality (7% and 6%) were found in adults and children with thalassaemia major (n = 293) and sickle-cell disease (n = 207). Death rates include all causes, not only those related to infection. It should be noted that in sickle-cell disease, complex multiorgan manifestations are frequent with a major impact on mortality (Hebbel, 2011) and functional asplenia ensues in most patients due to sequestration of red blood cells (RBC) trapped within the spleen and autoinfarction.

A population-based study conducted in Northern Jutland (Denmark), estimated that the crude overall risk of bacteraemia in 538 patients who were splenectomized for various causes, including malignant disorders, was 2·3 per 100 person-years at risk. While the mortality risk was 18·4 per 100 patient-years, no fatalities due to infections were recorded. Early postoperative bacteraemias (<30 d from splenectomy) accounted for 45% of all infections (Ejstrud et al, 2000). Most bacteraemias were due to Enterobacteria spp. and no bacteraemias due to Streptococcus pneumoniae, Neisseria meningitidis and Haemophilus influenzae type b were found after 30 d from splenectomy. Moreover, only one early bacteraemia was due to an encapsulated species (S. pneumoniae), although only 60% of patients had received pneumococcal vaccination. Among the non-trauma indications, haematological indications were associated with the lowest rates of mortality (7·2%) and early bacteraemia (1%).

A recently published nationwide analysis, based on data from the Danish National Patient Registry, compared infection risk in 3812 splenectomized persons, categorised into eight indication subgroups, versus three matched cohorts. Comparison cohorts (matched for age and sex) were the general population (n = 38120); appendectomized patients (n = 16962) and nonsplenectomized patients with the same underlying medical condition (matched-indication comparison cohort, n = 8310) (Thomsen et al, 2009). The overall incidence rate of infection requiring hospitalization in splenectomized patients was 7·7 per 100 person-years, versus 2 per 100 person-years for the general population. The adjusted RR was highest in the first 90 d after splenectomy, at 18·1, progressively declining to 4·6 at 3–12 months and 2·5 after the first year (Fig 1; details in Table SII).

Figure 1.

Relative risk (with 95% confidence intervals) for hospital contact for any infection in splenectomized patients by time since splenectomy: data from a Danish population-based study (Thomsen et al, 2009). Results are shown for any indication, immune thrombocytopenia (ITP; n = 269), hereditary haemolytic anaemias (HHA; n = 142) and trauma (n = 765), in comparison with age- and sex-matched subjects from the general population (A) and up to 5 matched-indication nonsplenectomized patients (B). For details, see Table SIII.

All subgroups of splenectomized patients were at much higher risk of infection than both appendectomized patients and the general population, particularly during the first 90 d post-splenectomy. The subgroups at highest risk versus the general population at this timepoint were those with splenomegaly/splenic disease (RR = 118), other/nonspecified thrombocytopenia (RR = 47) (not shown) and hereditary haemolytic anaemias (RR = 32), while the lowest risk was seen in ITP patients (RR = 15) (Fig 1). When compared to their nonsplenectomized matched-indication cohort, patients with ITP or hereditary haemolytic anaemias were at slightly higher infection risk (RR = 2·6 and 3·6) during the first 90 d after splenectomy, but not after this timepoint (Table SII). Splenectomized trauma patients had a similar risk to their matched-indication cohort at all three timepoints. For the first time, this important study clearly separated the deleterious effects of the different underlying conditions from the effects of splenectomy itself on the risk of infection and demonstrated that the magnitude of infection risk in splenectomized patients critically depends on the underlying disease and time since splenectomy. In particular, in patients with ITP or hereditary haemolytic anaemias, the risk of early, but not late, infections was significantly increased versus their matched-indication nonsplenectomized cohort.

In patients with sickle-cell disease, the increased risk of infection and death appears to be mainly attributable to the disease itself rather than splenectomy (Wright et al, 1999), as patients are already functionally hypo/asplenic, with impaired phagocytic function and defective activation of the alternative complement pathway (Bisharat et al, 2001). They are seldom splenectomized nowadays. Patients with thalassaemia major are also predisposed to infection via altered complement activation and immunoglobulin levels, cardiopulmonary disease and haematochromatosis, with splenectomy increasing the risk (Bisharat et al, 2001). Indeed, an infection incidence as high as 40% (4/10) was reported in splenectomized paediatric patients with thalassaemia (Ein et al, 1977). It should be noted that data for ITP patients may include patients with secondary ITP e.g. common variable immunodeficiency (CVID; also known as acquired hypogammaglobulinaemia), who may be more susceptible to infection than are patients with primary ITP (Park et al, 2008).

An analysis of mortality data from the Danish National Patient Registry (Yong et al, 2010) found that the overall adjusted relative risk of death (estimated as odds ratio) during the first 90 d after splenectomy (3812 cases) was 30-fold higher than in the general population, if all indications were included. Indications in the analysis included abdominal cancers (712), ITP (269), haematopoietic cancer (252), and hereditary haemolytic anaemias (142), as well as some less well-defined categories: traumatic rupture (765), splenomegaly/splenic disease (201), other nonspecified thrombocytopenia (53) and no indication recorded (1418). Unfortunately, causes of death were not reported and in particular, fatalities due to infection were not listed separately.

While the risk declines substantially after 90 d (RR = 5·1), patients remain at increased risk at 1 year post-splenectomy (RR = 2·3). Patients with traumatic spleen rupture or splenomegaly/splenic disease had the highest 90-d mortality risk (RR = 84·5 and 51·7, respectively) compared with the general population. Similarly, patients with ITP had an RR of 33·6 during the first 90 d after splenectomy. These high mortality risks appear to be inflated by uncontrolled associated morbidities, because the excess risk almost disappeared when compared with matched-indication patients who were not splenectomized. Notably, after 1 year, splenectomized patients with ITP had a lower mortality risk than nonsplenectomized ITP patients (RR = 0·4; 95% CI 0·2–0·7), consistent with the curative potential of splenectomy in this setting. No excess mortality was found in hereditary haemolytic anaemias, but splenectomized patients with haematopoietic cancers were at increased mortality risk compared to nonsplenectomized indication-matched patients for all three of the time periods.

Vascular events

Data sources for this section are summarized in Table SI.

Acute venous thromboembolism

The occurrence of deep vein thrombosis (DVT) outside the splenic and venous portal (SPV) system or pulmonary embolism (PE) in the immediate postsurgical period is not a matter of major concern after laparoscopic splenectomy, unless specific risk factors are present. Data from the Danish National Patient Registry (Thomsen et al, 2010) on thromboembolism risk in the first 90 d after splenectomy indicated that the overall frequency of DVT or PE was very low (0·63% and 0·73%, respectively). Nevertheless, the overall adjusted odds ratio for venous thromboembolism in splenectomized patients was 32·6 versus the general population and 3·2 versus appendectomy patients in the first 90 d after surgery, falling to 7·1 and 2·8 respectively at 91–365 d, and 3·4 and 3·2, respectively, at >1 year (Table SIII). No case series reporting the incidence of DVT or PE in patients splenectomized for benign haematological disorders are available. A high risk of acute thromboembolism has been reported in thalassaemia intermedia patients undergoing splenectomy, leading to the proposal of short-term antithrombotic prophylaxis during and after surgery (Cappellini et al, 2000).

Portal vein thrombosis

In contrast with the limited literature on acute thrombosis outside the SPV system, thrombosis or thromboembolism of the SPV system (SPVT) occurring within days to 2–3 months from splenectomy has raised major concerns and is still debated. The portal system forms an interactive venous system. Blood from the spleen is drained through the splenic vein, which after receiving the inferior mesenteric vein flows into the portal vein originating by the confluence of the splenic with the superior mesenteric vein. General factors contributing to SPVT are surgery itself, which is known to favour a pro-thrombotic state, and the need to induce a pneumoperitoneum in laparoscopic splenectomy, with consequent venous stasis within the SPV system. Most thromboses originate from the splenic vein stump left by the surgeon after ligation and cutting. They may remain limited to this area without consequences, extend beyond the splenoportal confluence to different sites of the system or appear isolated within separate areas of the SPV. This explains the wide spectrum of severity, ranging from asymptomatic to fatal (Stamou et al, 2006; Targarona, 2008).

The risk of SPVT was negligible in the Danish National Patient Registry analysis, with the number of events too small to obtain a meaningful risk measure (Thomsen et al, 2010). A small prospective study using contrast-enhanced computerized tomography (CT) scan found that SPVT occurred in 55% (n = 12; four symptomatic) of patients who underwent laparoscopic splenectomy, versus 19% (n = 4; none symptomatic) of those who underwent open splenectomy (P = 0·03)(Ikeda et al, 2005). However, in two large retrospective studies only nine of 563 (1·6%) (van't Riet et al, 2000) and 6 of 688 (0·9%) (Fujita et al, 2003) splenectomized patients developed symptomatic PVT and none of these cases occurred in ITP patients. More recently, an SPVT incidence of 3·4% (1/29) was seen in a small, well-conducted study of oral anticoagulation after splenectomy (Wang et al, 2011). The risk of SPVT very clearly correlates with the presence of splenomegaly, or lymphoproliferative or myeloproliferative disorders. Hereditary haemolytic anaemias and thalassaemia also carry a distinctly higher risk (van't Riet et al, 2000; Fujita et al, 2003; Krauth et al, 2008). A comprehensive compilation of the various studies describing systemic thrombosis or SVPT after splenectomy for haematological disorders, including the potential pathophysiological mechanisms, has been published recently (Crary & Buchanan, 2009).

Mesenteric vein thrombosis, the most severe complication of SPVT in the immediate weeks to months after splenectomy, has been reported in benign indications including ITP, haemolytic anaemia and haemoglobin H disease and has often been linked to post-splenectomy thrombocytosis as an aggravating factor (Balz & Minton, 1975; Tso et al, 1982). Diagnostic procedures, such as Doppler ultrasonography, helical contrast ST or magnetic resonance imaging (MRI), should be promptly initiated in patients who develop vague, cramping abdominal pain, depressed bowel sounds and generalized bowel tenderness, because mesenteric vein thrombosis can lead to potentially fatal intestinal infarction.

Late vascular events

An increased long-term risk (in terms of years) of vascular complications after splenectomy performed for benign haematological diseases or trauma was first reported in ex-servicemen who underwent splenectomy for war injuries (Robinette & Fraumeni, 1977) and subsequently in patients with hereditary spherocytosis (Schilling, 1997). In World War II veterans splenectomized for trauma between 1944 and 1945, with splenectomy coded as the first or second operative procedure, the RR for death from ischaemic heart disease was 1·9, versus a control group of nonsplenectomized men (hospitalized for acute nasopharyngitis). Of the 41 deaths (acute myocardial infarction 30: chronic ischaemic heart disease 11), 36 occurred more than 15 years after splenectomy (Robinette & Fraumeni, 1977). As well as being based on unvalidated administrative data, the value of this study is severely hampered by the lack of a matched control group. Excess risk of death from pneumonia, but not from thromboembolism, in the asplenic men was seen in this analysis. Schilling (1997) found that in patients with hereditary spherocytosis, the rate of arteriosclerotic endpoints (stroke, myocardial infarction, coronary or carotid artery surgery) was 5·6 fold higher in asplenic patients, with the first event occuring one or more decades after splenectomy. Anecdotal cases of late occurrence of thrombosis after splenectomy in hereditary spherocytosis have also been reported (Hayag-Barin et al, 1998).

More recent data from the Danish Registry showed that the long-term (>1 year) risk of VTE remained approximately 3-fold higher in trauma patients (n = 765) versus the general population, and almost 4-fold higher versus appendectomized patients (Table SIII) (Thomsen et al, 2010). However, the contribution of wounds, fractures, crush injuries, orthopaedic surgery and immobility to thrombotic risk is unknown. The long-term RRs were highest in patients splenectomized for malignant haematological disorders, haemolytic anaemia or splenomegaly (8–22 versus the general population and 6–11 versus appendectomized patients; data not shown). In ITP patients the long-term risk was slightly increased versus the matched general population [RR = 2·7 (1·1–6·3)] but not versus appendectomized patients [2·6 (0·9–7·1)].

Patients undergoing splenectomy for benign haematological disorders may be at additional risk for late vascular events, depending on the type of disorder, as noted by Crary and Buchanan (2009). The risk of thromboembolism and pulmonary arterial hypertension appears to be greatly increased after splenectomy in patients with disorders that are associated with ongoing haemolysis, including thalassaemia, particularly thalassaemia intermedia, hereditary stomatocytosis and sickle cell disease (Crary & Buchanan, 2009). Indeed, thalassaemia major and intermedia are now considered to be associated with a hypercoaguable state that may be aggravated by splenectomy (Eldor & Rachmilewitz, 2002). A large survey of patients with thalassaemia found an overall 1·7% (146/8860) incidence of thromboembolism, with 93% of these thrombotic events occurring in splenectomized patients. Patients with thalassaemia intermedia were approximately 4-fold more likely to experience thromboembolism than those with thalassaemia major [85/2190 (4%) vs. 61/6670 (1%); P < 0·001] (Taher et al, 2006). Subsequent analysis of data from 584 thalassemia intermedia patients also found that thrombosis was approximately six times more frequent among the splenectomized patients (23% vs. 4%). Other complications of the disease, including osteoporosis, cholelithiasis, pulmonary hypertension, leg ulcers, hypothyroidism, and diabetes, were also approximately 4- to 6-fold more common in those who had undergone splenectomy (Taher et al, 2010a). However, as splenectomy is usually reserved for patients with growth retardation, poor health, leucopenia, thrombocytopenia, increased transfusion demand or symptomatic splenomegaly, the findings of this retrospective study are difficult to interpret (Cappellini et al, 2008). The same investigators (Taher et al, 2010b) found that in 73 patients with thalassaemia intermedia and thromboembolism (DVT 63%, PE 18%, PVT 15%, superficial vein thrombosis 16%, stroke 6%) the median time to thrombosis was 8 years (range 1–33 years). Comparison with data from splenectomized and nonsplenectomized patients who did not develop thromboembolism indicated that increased platelet count (>500 × 109/l) was a major risk factor, as were increased nucleated RBC and transfusion naivety.

Potential mechanisms of increased risk of thrombosis

Potential underlying mechanisms for an increased risk of thromboembolism after splenectomy have been discussed (Crary & Buchanan, 2009). Loss of the spleen's filtering activities may allow particulate matter and damaged cells to persist in the circulation, leading to changes in the endothelium that result in hypercoagulability. Other changes that have been reported to occur after splenectomy that might potentially contribute to thrombosis risk include increased platelet and leucocyte counts, and haemoglobin, cholesterol, and C-reactive protein levels. In patients with haemolysis, loss of the spleen leads to a shifting of haemolysis from extravascular to intravascular, resulting in increased plasma haemoglobin levels (Crary & Buchanan, 2009). As haemoglobin scavenges nitric oxide, this could promote a generalized vasculopathy, including vasoconstriction, smooth muscle proliferation and activation of platelets and endothelial cells, particularly within the pulmonary vasculature (Crary & Buchanan, 2009; Kato, 2009).

The intertwined contribution of splenectomy and ongoing haemolysis and/or RBC abnormalities to thrombosis is not yet fully clarified and both conditions appear to contribute to the increased risk. Abnormal RBC may function as activated platelets in patients with thalassaemia intermedia, providing a procoagulant surface for increased thrombin formation. A reduced rate of removal of these structurally abnormal cells after splenectomy may aggravate hypercoagulability (Mannucci, 2010). Other investigators noted that the spleen may have a role in scavenging non-transferrin-bound iron (Tavazzi et al, 2001). These findings call into question the place of splenectomy as a procedure of choice in thalassaemia intermedia. Similar mechanisms may be less important in regularly transfused patients with thalassaemia major, in whom the majority of circulating RBC are normal because of current intensive transfusion regimens.

The pathogenic mechanisms leading to the multiorgan manifestations of sickle-cell disease are complex and still controversial (Hebbel, 2011). However, it is established that adherence of sickled RBC to the vascular endothelium leads to vasculopathy and organ damage. The contributing role of the underlying disease and/or of its treatment to thromboembolism cannot be separately assessed.

Nonsplenectomized adults with hereditary spherocytosis (which is not usually associated with significant haemolysis) are reported to be at markedly lower risk of thrombosis than the general population, possibly reflecting a protective effect of chronic mild anaemia, but splenectomy appears to raise their risk to a similar level to healthy individuals (Schilling, 2009). Delayed thromboembolism has important clinical implications for the design of effective prophylaxis.

Data from a murine study suggesting that the spleen is involved in storage and rapid deployment of monocytes to regulate inflammation (Swirski et al, 2009) raise questions as to whether splenic monocytes play a key role in other types of stress or injury (Jia & Pamer, 2009) and whether loss of the splenic monocyte reservoir might provide some explanation for the vascular events associated with splenectomy.

Cancer risk

Splenectomy might be expected to affect cancer risk via its effects on immunological function. Several reports suggest that in the setting of Hodgkin lymphoma, splenectomy independently increases the risk for secondary leukaemias (Chung et al, 1997; Henry-Amar, 1992; Kaldor et al, 1990; Tura et al, 1993; van Leeuwen et al, 1987). However, the risk of cancer does not appear to be increased in patients who undergo splenectomy for trauma (Robinette & Fraumeni, 1977; Mellemkjoer et al, 1995; Linet et al, 1996). For instance, a Danish population-based study of 6315 splenectomized patients followed for a mean of 6·8 years found no excess risk in trauma patients (Mellemkjoer et al, 1995).

Perioperative complications

Perioperative complications of splenectomy include complications of general anaesthesia, bleeding and thrombosis, as well as postoperative pain, pneumonia and atelectasis, wound/other infection and ileus. The increasing use of minimally invasive laparoscopic techniques, together with refinements in optics, camera technology and instrumentation, have greatly reduced perisurgical complications and also shortened hospitalization and convalescence times compared with conventional open splenectomy (Winslow & Brunt, 2003; Kojouri et al, 2004; Dolan et al, 2008). A meta-analysis of data from 51 published series in which splenectomy was carried out for various indications, found that the overall complication rate for laparoscopy (n = 2119 patients) was 15·5%, vs. 26·6% for open surgery (n = 821 patients) (P < 0·0001) (Winslow & Brunt, 2003). Laparoscopy was associated with a significantly (P < 0·001) lower risk of pulmonary (3·8% vs. 9%), wound (1·6% vs. 4·3%), and infectious (1·0% vs. 3·8%) complications, as well as subphrenic abscess (0·1% vs. 2·4%). However, laparoscopy was associated with more haemorrhagic complications than open surgery (4·8% vs. 2·4%), when conversions for bleeding were taken into account. Mortality rates did not differ significantly between the two techniques (laparoscopy 0·6% vs. open surgery 1·1%).

In an analysis of 3386 splenectomies, all in ITP patients (Kojouri et al, 2004), the complication rate was 9·6% for laparoscopic splenectomy and 12·9% for open splenectomy, while associated mortality rates were very low, at 0·2% and 1%. Recently, more favourable results of laparoscopic splenectomy have been published. For example, a multicentre analysis conducted via the Italian Registry of Laparoscopic Surgery of the Spleen (Casaccia et al, 2010) found only three fatalities, all in patients with haematological malignancies, among 676 splenectomies performed between 1993 and 2007. No deaths were recorded among 246 patients with ITP and 142 patients with congenital or acquired haemolytic anaemia. Respective morbidity rates were 13% and 14%, while median hospital stay was 5 d. Furthermore, it has been suggested (Keidar et al, 2005) that laparoscopic splenectomy may be carried out in patients with ITP without measures to increase platelet count preoperatively, if platelets are above 20 × 109/l. Below this threshold, however, severe bleeding may occur.

Measures to reduce the adverse effects of splenectomy

Table 1 summarizes current recommendations to prevent serious infections in patients without a functional spleen, while Table 2 summarizes recommendations for preventing acute venous thromboembolism and late vascular complications.

Table 1. Summary of recommended measures to prevent infection in patients without a functional spleen

Immunization against pneumococcal, Haemophilus influenzae type b and meningococcal infections should be initiated at least 2 weeks before splenectomy, or as soon as possible after emergency

splenectomy. Immunization schedules should be determined on the basis of country-specific official recommendations or local practices.

A common scheme for pneumococcal vaccination is: 1 dose of 13-valent (or 7-valent if not available) protein conjugate vaccine (PCV) 14 d before splenectomy, followed by a dose of 23-valent

pneumococcal polysaccharide vaccine (PPV) 2 months later, with a 23-valent PPV booster after 5 years and 5–10 years thereafter or based on antibody titre. Despite the lack of formal proof of an

additional benefit in adults, we recommend that in all patients, and particularly in children aged <5 years, vaccination should be complemented with protein conjugate vaccines. Detailed

recommendations by age group and previous vaccination history are provided in the recently updated British Committee for Standards in Haematology (BCSH) guidelines (Davies et al, 2011). Patients

with a good serological response to PPV may not derive additional benefit from PCV revaccination. Antibody levels may decline rapidly in some patient groups, e.g. those with sickle-cell anaemia.

A single dose of the recently available tetravalent conjugate meningococcal vaccine, with booster doses at 5-year intervals, is suggested by some authorities (Advisory Committee on Immunization

Practices (ACIP) Centers for Disease Control and Prevention (2009).

A single dose of conjugated H influenzae type b vaccine appears to be sufficient, at least in adults.

Patients should also be offered annual vaccination against influenza, because this has been shown to reduce overall mortality, despite possibly not reducing the rate of overwhelming post-

splenectomy infection (Langley et al, 2010).

Vaccinations may not be effective in patients who have received treatment with the anti-CD20 antibody rituximab in the previous 6 months (Makris et al, 1994; Provan et al, 2010; Davies et al, 2011).

Antibiotic prophylaxis is required after tooth extraction/dental procedures or surgery in general. Despite the suggestion by some authorities and most conservative guidelines that antibiotic

prophylaxis should be continued long-term, if not for life, this recommendation has never been validated and is usually not followed by patients. Nevertheless, the recently updated BCSH guidelines

recommend that patients at high risk of infection should receive lifelong prophylaxis (Davies et al, 2011). It is important that patients are educated to be aware of their increased risk and remain

vigilant for signs of infection (Davies et al, 2002, 2011; Provan et al, 2010).

Based on anecdotal reports, Gram-negative bacteria, Capnocytophaga canimorsus and intraerythrocytic parasites, such as Malaria spp., may also pose extra risk to asplenic individuals (Green et al,

1986). Several cases of babesiosis have been reported (White et al, 1998). Thus, patients should be aware of their increased risk when travelling to tropical areas. If bitten by dogs or other animals,

they should receive appropriate antibiotic cover.

Table 2. Recommendations for preventing acute venous thromboembolism and late vascular events in splenectomized patients

Routine pharmacological thromboprophylaxis after laparoscopic surgery is not standardized. Its efficacy after laparoscopic splenectomy for benign haematological disorders in order to prevent deep

vein thrombosis of the lower limbs and/or pulmonary embolism and splenic and venous portal system thromboembolism (SPVT) is unproven (Krauth et al, 2008; Wang et al, 2011).

The American College of Chest Physicians guidelines (8th edition) recommend against routine thromboprophylaxis (grade B recommendation) in patients without additional risk factors (Geerts

et al, 2008), whereas in the 9th edition no recommendations are given. Mechanical or heparin-based thromboprophylaxis is suggested, according to patients' age below or above 40 years (Gould

et al, 2012). The clinical practice guidelines of the European Association for Endoscopic Surgery recommend perioperative anticoagulant prophylaxis with subcutaneous heparin for all patients.

Patients at high risk for SPVT should receive anticoagulant prophylaxis for 4 weeks (grade C recommendation) (Habermalz et al, 2008).

However, additional risks may be present in patients with thalassaemia intermedia or haemolytic anaemia if they have a very large spleen. Strict surveillance for SVPT is recommended and colour

Doppler ultrasound investigation for the detection of SPVT is advisable before discharge.

For patients who develop thrombosis extending beyond the splenic vein, therapeutic doses of low molecular weight heparin, followed by oral anticoagulation for 3-6 months, are recommended, even

in asymptomatic cases. Treatment of symptomatic SVPT is mandatory and a protracted course of oral anticoagulation is suggested (Krauth et al, 2008).

Prevention of late vascular events is based on the same principles adopted in the general population at risk, i.e. reducing modifiable thrombotic risk factors. Long-term aspirin administration is

advised for those with thalassaemia intermedia or additional risk for arterial thrombosis, although it is not proven that elevated platelet count is a risk factor for thrombosis in patients at high risk

(Mannucci, 2010).

In splenectomized patients requiring elective surgery, robust perioperative thromboprophylaxis is recommended, even in young subjects. Patients splenectomized for haemolytic anaemias or

thalassaemia who are exposed to other situations carrying an added risk of thrombosis should receive antithrombotic prophylaxis with low molecular weight heparin (Geerts et al, 2008; Gould et al, 2012).

Strategies to prevent serious infections include patient education, vaccination and antibiotic prophylaxis. Such measures are universally felt as critical to reduce the risk of OPSI, but despite a plethora of data and guidelines, lack supporting evidence from unbiased trials or cohort studies. Indeed, most studies have major limitations, due to inclusion of patients with malignant and nonmalignant disorders, or different age groups or geographic settings. Moreover, in almost all studies patients received suboptimal vaccination (El-Alfy & El-Sayed, 2004; Okabayashi & Hanazaki, 2008; Langley et al, 2010). Suboptimal vaccination is particularly relevant for young children. The immaturity of their immune system and diminished activity of IgM memory cells lead to the inability to mount an effective response against purified polysaccharide vaccines (Di Sabatino et al, 2011). The newer conjugated vaccines elicit a T-lymphocyte-dependent response and are thus preferred over polysaccharide vaccines for use in young children (Di Sabatino et al, 2011). Indeed, they have been widely integrated into routine childhood vaccination programmes. Nevertheless, because of the increased infectious risks in children, splenectomy is usually deferred where possible until they are aged at least 5 years.

Treatment with the anti-CD20 monoclonal antibody rituximab results in rapid and prolonged depletion of normal B cells lasting at least 6 months. Conceivably, vaccinations may not be effective in patients who have received treatment with rituximab in the previous 6 months (Makris et al, 1994; Davies et al, 2002; British Committee for Standards in Haematology General Haematology [BCSH] Task Force, 2003; Provan et al, 2010). Thus, whenever possible, patients should be vaccinated before starting rituximab treatment. Although these recommendations are not based on evidence from the setting of benign haematological disorders, it is noteworthy that in a recent study none of 67 lymphoma patients who were receiving this agent, or had received it in the previous 6 months, were able to mount protective antibody responses to the influenza A(H1N1) 2009 virus vaccine (Yri et al, 2011). In contrast, 82% (42/51) of a control group of healthy volunteers responded adequately (Yri et al, 2011).

Antibiotic prophylaxis is required after splenectomy. Despite the suggestion by some authorities and most conservative guidelines that this should be continued long-term, if not for life, this recommendation has never been validated and is usually not followed by patients. Nevertheless, the recently updated BCSH guidelines recommend that patients at high risk of infection, including those aged ≤16 or >50 years, those with inadequate response to pneumococcal vaccination or a previous episode of invasive pneumococcal disease should receive lifelong prophylaxis (Davies et al, 2011). It is important that patients are educated to be aware of their increased risk and remain vigilant for signs of infection. Antibiotic prophylaxis is required after tooth extraction/dental procedures or surgery in general (Davies et al, 2002, 2011; Provan et al, 2010).

Reassessing the role of splenectomy in benign haematological disorders

Primary immune thrombocytopenia

Primary ITP is an autoimmune disease characterised by low platelet counts due to both increased platelet destruction and suboptimal platelet production (Rodeghiero et al, 2009; Provan et al, 2010). An International Working Group (IWG) recently defined three phases of primary ITP: (i) newly-diagnosed (first 3 months since diagnosis), during which treatment is aimed at rapidly obtaining a safe platelet count; (ii) persistent ITP, for patients not achieving spontaneous remissions or not maintaining their response after stopping treatment between 3 and 12 months from diagnosis, and (iii) chronic ITP lasting longer than 12 months. The IWG considered spontaneous remission most likely to occur during the persistent phase and hence suggested deferral of splenectomy until 12 months after diagnosis, whenever possible (Rodeghiero et al, 2009).

In newly diagnosed patients, corticosteroid treatment achieves an initial response in approximately 70–80% of patients (Provan et al, 2010), but relapse is common and repeated or prolonged treatment courses are often required with frequent burdensome toxicity, hence splenectomy may eventually be considered. Splenectomy has been regarded as the gold standard in the setting of chronic ITP, as it achieves an initial and prolonged response in approximately two-thirds of patients (Kojouri et al, 2004). Data presented earlier in this review suggest that splenectomy carries a much lower risk of infection and thrombosis in ITP patients than in patients with other benign haematological disorders. While the rate of infection is higher than in the general population, it is not increased versus nonsplenectomized ITP patients after the first 3 months (Thomsen et al, 2009). The low risk of sepsis and thrombosis in patients splenectomized for ITP is also supported by findings of representative series, as shown in Table 3.

Table 3. Fatal and non–fatal sepsis and thrombotic events in representative cohorts of patients splenectomized for ITP
ReferencesSplenectomy, N/n (%)Follow-up, median (range)Septic deaths (splenectomized patients)Non-fatal sepsis (splenectomized patients)Arterial thrombosisVenous thrombosis
  1. CV, cardiovascular; NR, not reported.

  2. a

    Study of outcome in splenectomized patients.

  3. b

    Only 47 refractory patients were followed.

  4. c

    Six cases from Vianelli study attributed to each type of thrombosis.

Stasi et al (1995)63/208 (30)92 (48–151) months00NRNR
Mazzucconi et al (1999)94/94 (100)a83 (22–180) months20NRNR
Portielje et al (2001)78/152 (51)10·5 years (2 months–22·6 years)2 (1 early)2 (+2 pneumonia) 2 (early)
Bourgeois et al (2003)183/183 (100)a7·5 (5–15) yearsb00NRNR
Neylon et al (2003)30/245 (12)66 (6–78) months1 (early)0NRNR
Schwartz et al (2003)75/75 (100)a7·5 (5–15) years00Sudden death 
McMillan and Durette (2004)114/114 (100)a110 months (for 105 cases)3 (+1 pancreatitis and sepsis)NR1 + 6 death for CV events11
Vianelli et al (2005)402/402 (100)a92 (1–502) months0012 (type not specified) 
Total n (%) (95% confidence interval)1039 10 (0·96) (0·46–1·7)2 (0·19) (0·02–0·69)14 (1·3)c (0·7–2·2)19 (1·8)c (1·1–2·8)

New medical treatment options for splenectomized and nonsplenectomized patients have emerged over recent years. The recently introduced thrombopoietin receptor agonists (TPOra) romiplostim (Kuter et al, 2008, 2010a) and eltrombopag (Bussel et al, 2007, 2009a; Cheng et al, 2011) have proved effective in producing sustained increases in the platelet count in 70–80% of patients, while concomitantly ameliorating bleeding manifestations and improving their quality of life (George et al, 2009; Sanz et al, 2011; Kuter et al, 2012). TPOra appear to be well tolerated, with few patients discontinuing treatment, even during prolonged treatment periods of up to 5 years with romiplostim (Bussel et al, 2009b; Kuter et al, 2010b). The clinical significance of newly apparent/increased bone marrow reticulin found in a very small proportion of patients treated with TPOra (Kuter et al, 2008, 2009; Saleh et al, 2009; Brynes et al, 2011) and the hypothetical increased risk of vascular events, particularly in patients with pre-existing risk factors, cannot be excluded at present and need to be further investigated in more prolonged studies. The need for continuous treatment and monitoring, the remaining hypothetical long term risks and their cost suggest that TPOra may be best reserved for patients failing splenectomy or with contraindications to surgery. An International Consensus Report placed splenectomy in a second-line group of treatments, which also included TPOra, without expressing a preference (Provan et al, 2010). Recent guidelines from the American Society of Hematology also recommend both splenectomy and TPOra as possible second-line treatment choices in patients failing corticosteroids (Neunert et al, 2011), but noted that the authors were ‘unable to make specific evidence-based recommendations such as prioritizing treatments for patients who have failed first line-therapy’. The use of TPOra in persistent ITP in place of more toxic treatments, to bridge the gap while waiting for ITP to enter its chronic form (>12 months from onset), is an attractive strategy that is gaining support. It should be noted that use of TPOra before splenectomy is currently off-label in Europe (although not in the US), unless surgery is contraindicated http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Product_Information/human/000942/WC500039537.pdf, http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Product_Information/human/001110/WC500089964.pdf).

Rituximab, which selectively depletes B-cell lymphocytes, is also used as a second-line treatment in ITP (Arnold et al, 2007). However, it is associated with significant toxicities, including a risk (albeit low) of fatal adverse events (estimated incidence 2·9%) (Arnold et al, 2007), and its use in ITP is off-label (Arnold et al, 2007). Moreover, the long-term response rates compare unfavourably with those of splenectomy, at no more than 20% (Patel et al, 2006). Furthermore, patients requiring splenectomy after rituximab treatment are expected to be unresponsive to vaccines for at least 6 months, and some concerns have also been raised regarding pregnancy outcomes after maternal exposure to rituximab (Chakravarty et al, 2011). Thus, rituximab is rarely indicated before splenectomy.

Our current policy is to maintain splenectomy as the preferred choice after failure of initial treatment, preferably after 12 months from diagnosis, followed by TPOra, and lastly by rituximab, in refractory patients (Rodeghiero & Ruggeri, 2011). However, we recognize that the approach needs to be individualized, in order to meet the needs of the patient (Rodeghiero & Ruggeri, 2008; Stasi et al, 2010).

Haemolytic anaemias and red blood cell disorders with enhanced haemolysis

In haemolytic anaemias or other RBC disorders with haemolysis, circulating RBC are recognized as abnormal by the reticulo-endothelial system and consequently rapidly removed by the spleen. This may lead to/aggravate anaemia and result in severe splenomegaly, which may be accompanied by jaundice and gallstones. Thalassaemia and sickle cell disease are additionally characterized by skeletal abnormalities due to extramedullary haematopoiesis, failure to thrive and growth retardation in children. Management approaches in congenital forms of these disorders are still mainly supportive within a comprehensive treatment programme, based on blood transfusions and iron chelation therapy to reduce transfusional iron overload and its associated complications. Medical therapies are mainly used in acquired disorders, although the antimetabolite hydroxycarbamide (previously termed hydroxyurea), which increases levels of fetal haemoglobin (HbF), is increasingly being prescribed to prevent complications of sickle cell disease, including vaso-occlusive painful crises and acute chest syndrome (Steinberg et al, 2003). Drawbacks of hydroxycarbamide include neutropenia and thrombocytopenia, thus regular monitoring for cytopenias is required. A further concern is possible effects on growth and development in children. However, such risks should be interpreted in the context of the mortality risks associated with complications of sickle cell disease: 9-year follow-up of patients in the Multicenter Study of Hydroxyurea in Sickle Cell Anemia suggests that hydroxycarbamide treatment may reduce overall mortality by 40% in this setting (Steinberg et al, 2010).

Splenectomy was previously widely undertaken in patients with haemolytic anaemias or other RBC disorders with haemolysis, in order to reduce the extravascular haemolysis. However, given the high risk of long-term vascular complications, a more conservative approach has been advocated in the settings of hereditary spherocytosis by the BCSH (Bolton-Maggs et al, 2004) and by Schilling (2009), and in thalassaemia intermedia (Taher, et al 2010a) and thalassaemia major (Cappellini et al, 2008). The current indication for splenectomy in thalassaemia major remains a progressive increase in transfusion requirement due to hypersplenism and difficulty in controlling iron overload. However, a more conservative approach is presently suggested and a large spleen alone should not represent an indication for splenectomy. In thalassaemia intermedia, splenectomy is even more sparingly undertaken. Transfusion therapy, which is not usually recommended in current practice, may be worthwhile to reduce the risk of thromboembolic events (Mannucci, 2010).

In children with sickle-cell crisis, splenectomy may be considered after the crisis has been abated by transfusion and medical treatment. The immediate risks seem similar to those associated with other indications for benign haematological disorders, but the long-term outcome is burdened by many vascular complications (Kokoska et al, 2004; Ghantous et al, 2008; Kalpatthi et al, 2010). Advice from expert centres is recommended.

There are currently no treatment guidelines for autoimmune haemolytic anaemias (AIHA). Corticosteroids are still the first choice in AIHA due to warm antibodies, while splenectomy remains the most effective and best-evaluated second-line therapy. However, there are few data on the long-term outcomes of these patients. Rituximab is another second-line therapy, but has limited long-term efficacy and safety data. AIHA due to cold antibodies does not respond to steroids or splenectomy and rituximab is usually the preferred first-line treatment (Lechner & Jager, 2010).

Conclusions

Some decades ago it was assumed that the spleen was relatively unimportant and could be removed without any significant adverse consequences. However, ensuing years have seen a growing understanding of the spleen's role and accumulation of a large body of data documenting the risks associated with splenectomy, with evidence that these risks may strongly depend on the underlying disorder.

Improvements in surgical techniques have reduced perioperative complications and surgical mortality is negligible, while the risk of OPSI has been reduced considerably – but not eliminated – since the introduction of pneumococcal, Haemophilus influenzae and meningococcal vaccinations, long-term antibiotic prophylaxis and increased vigilance. The recent introduction of protein conjugate vaccines is expected to further reduce the infectious risk. Nevertheless, in children splenectomy should still be deferred until they are aged at least 5 years.

Splenectomized patients are now known to be at increased risk of venous thromboembolism, particularly within the splenoportal system. SPVT, while uncommon, has emerged as a concern and may have devastating consequences, thus any abdominal pain, diarrhoea, ascites, ileus or unexplained low fever manifesting days or months after splenectomy should be investigated by CT, MRI or colour Doppler flow imaging, in order to initiate appropriate treatment.

In the setting of moderate blunt splenic injury in adults, a trend towards spleen preservation has been seen over recent decades, with splenectomy now usually being reserved for those who are haemodynamically unstable or have massive bleeding (Gauer et al, 2008; Heuer et al, 2010). Splenectomy maintains a central role in the treatment of many benign haematological disorders, particularly ITP, although a more conservative approach is emerging, particularly in those disorders that are associated with ongoing haemolysis. In ITP patients, an approach that is gaining support is to defer splenectomy for at least 12 months after diagnosis, in order to allow the patient the best chance of spontaneous remission.

Table 4 summarizes the suggested place of splenectomy in the management algorithm for the benign haematological disorders covered in this review. In all settings, the indication for splenectomy should be carefully assessed against alternative strategies as they emerge from recent or ongoing clinical studies, taking into account the priorities and needs of a well-informed patient.

Table 4. Suggested place of splenectomy in the management algorithm of benign haematological disorders
DiseaseSequence of treatmentComments
Primary immune thrombocytopenia, adultsSecond-lineAfter 12 months from diagnosis in adult patients at risk of bleeding, or earlier in cases requiring prolonged steroids or other treatments with unacceptable toxicity.
Primary immune thrombocytopenia, childrenThird- or fourth- lineRarely recommended in paediatric patients, unless at high risk of bleeding and unresponsive to medical treatment or experiencing unacceptable toxicity.
Immune haemolytic anaemia (warm antibodies)Third-lineIn patients unresponsive to medical treatments, after failure of steroid therapy and anti-CD20 antibody administration.
Thalassaemia majorRarely indicatedProgressive increase in transfusion requirement due to hypersplenism and difficulty in controlling iron overload. A large spleen alone should not represent an indication for splenectomy.
Thalassaemia intermediaSecond-lineIn patients requiring massive transfusion treatment and not responsive or intolerant to iron chelation therapy or in case of massive splenomegaly.
Sickle cell diseaseRarely indicatedUsually only in case of massive splenomegaly or with sequestration crisis. Patients often have functional asplenia.
Hereditary spherocytosisSelective indicationIn patients aged ≥6 years with moderate/severe haemolysis and anaemia and/or gallstones.

Acknowledgements

F.R. designed the research study. F.R. and M.R. identified and reviewed the literature, analysed the data and wrote the paper. Julia Balfour, Medical Writer, Dundee, Scotland provided assistance with the preparation of the manuscript, with financial support from Amgen (Europe) GmbH, Zug, Switzerland.

Conflicts of interest disclosures

F.R. receives honoraria for participation in advisory boards and/or as a speaker at medical education events supported by Amgen, GSK, Shionogi, Celgene, Suppremol and LFB. MR receives honoraria for participation in advisory boards and/or as a speaker at medical education events supported by Amgen, GSK, and Celgene. The authors state that they have made an independent search for relevant literature, reviewed all references and take full responsibility for the content of this article, which represents their viewpoint and clinical expertise. Amgen did not influence the content of the manuscript, nor did the authors receive financial compensation for authoring the manuscript.

Ancillary