Kidney transplantation

A total of 544 kidney transplant recipients from seven studies (Horvath 1975; Kauffman 1980; Kikic 2009; Osman 2007Ubhi 1989; Zhang 2005; Zhang 2009a) were included in this review; full details of each study can be found in the "Characteristics of Included Studies" section.

Zhang 2005 and Zhang 2009a looked at patients who were already experiencing an acute rejection episode and reported on levels of platelet activation, graft recovery time, and rates of patient and graft survival.

Kikic 2009 looked at the type of anticoagulant used in pre‐transplant dialysis on graft function, graft survival and bleeding events.

Horvath 1975 and Kauffman 1980 both note in their introductory paragraphs that intravascular coagulation is often seen in rejected grafts and that manipulation of coagulation systems may affect outcomes (such as graft function) but neither study set out to explicitly investigate rates of graft thrombosis in their patients. Horvath 1975 was particularly interested in the role of heparin in improving post‐operative graft function and also reported on rates of haemorrhagic complications, and graft survival. Kauffman 1980 was “spurred by internal and external pressures for cost containment” and primarily aimed to compare aspirin to dipyridamole as part of the routine care of patients post transplant. Kauffman 1980 reported on graft function, rates of gastrointestinal haemorrhage, and transfusion requirements.

Ubhi 1989 and Osman 2007 were the only studies that mentioned investigating rates of thrombosis as a main aim of their study. However, both looked at very particular groups of kidney transplant recipients; Ubhi 1989 was interested in patients receiving cyclosporin A as immunosuppression and the role of heparinisation in this group. Osman 2007 aimed to investigate the role of routine heparinisation in “non‐risky” kidney transplant recipients. 

Only four studies investigated ongoing anti‐thrombosis strategies which were started at the time of kidney transplantation, and all but one were completed over 30 years ago (Horvath 1975; Kauffman 1980; Osman 2007; Ubhi 1989). As such, these studies were conducted in an environment that was very different to that of modern transplantation where certain variables such as cold ischaemic times, preservation methods, and immunosuppressive protocols were different to those used today.

Studies were conducted in China (Zhang 2005; Zhang 2009a), Australia (Horvath 1975), UK (Ubhi 1989), USA (Kauffman 1980), Egypt (Osman 2007), and Austria (Kikic 2009).

One study was performed in live donor transplants (Osman 2007), one in grafts donated following brain death (Kikic 2009), and one in grafts donated following circulatory death (Zhang 2009a). The remaining four studies (Horvath 1975; Kauffman 1980; Ubhi 1989; Zhang 2005) all specify looking at patients who had received deceased donor donations and we have inferred from the timing of these studies that they likely donated following brainstem death.

Heparin was the most commonly studied thromboprophylactic therapy, being the intervention in six of the seven studies. Three studies examined unfractionated subcutaneous heparin (Horvath 1975; Ubhi 1989; Osman 2007), two investigated a combination of heparin and Lipo‐PGE₁ (Zhang 2005; Zhang 2009a), and one examined LMWH (Osman 2007). Kikic 2009 was conducted slightly differently as their intervention was the type of anticoagulant used in pre‐transplant dialysis. (see Characteristics of included studies).

Four of the seven studies reviewed their intervention in all patients post kidney transplantation (Horvath 1975; Kikic 2009; Osman 2007Ubhi 1989) while the remaining three studies only looked at patients experiencing an acute rejection episode (Kauffman 1980; Zhang 2005; Zhang 2009a).

Six studies (Horvath 1975; Kauffman 1980; Osman 2007; Ubhi 1989; Zhang 2005; Zhang 2009a) compared their intervention to a control group, although in three studies (Ubhi 1989; Zhang 2005; Zhang 2009a) it was not clear whether the control group received a placebo. Only Kauffman 1980 compared the use of aspirin to dipyridamole without the use of a placebo or no treatment.

Liver transplantation

Two studies (Mansourian 2014; Villamil 2016) (168 patients) investigated interventions to prevent thrombosis in liver transplantation. Both were conference abstracts and no full publications of either study were available at the time of writing this review. As full manuscripts were not available it was often difficult to elicit full details about study methods and outcome reporting due to lack of available information. We contacted authors of both studies to request further information, but we received no response.

Both studies stated using "randomised methodology", however further details about their randomisation methods were unavailable. Neither study mentioned blinding or use of a placebo. Similarly it was not mentioned in either study what type of donor the organs were retrieved from.

Mansourian 2014 was an Iranian study of 139 patients undergoing orthotopic liver transplantation who had a portal vein thrombosis immediately prior to transplantation. Following thrombectomy, initiation of heparin therapy and confirmation of portal vein patency by Doppler study, patients were randomised to receive either aspirin or warfarin therapy for three months. Patients with Budd‐Chiari and those where thrombectomy was not possible were excluded. Their study was mainly aimed at investigating rates of rethrombosis in these patients. 

Villamil 2016 looked at 29 patients undergoing orthotopic liver transplantation in Argentina who had high levels of anti‐phospholipid antibodies or lupus anticoagulant and were also assessed to be at “high risk of thrombosis” (factors used to determine 'high risk' were not described). These patients were randomised to receive either low dose aspirin with or without LMWH post‐transplantation or pre‐operative plasmapheresis followed by postoperative anticoagulation for three months with the aim of investigating differences in postoperative thrombotic complications.

Both studies reported specifically on rates of thrombotic complications. Villamil 2016 reported on the rates of death and graft loss in their patients. Mansourian 2014 only reported on patients who died due to haemorrhagic complications. The number of patients who died from other causes and the number of non‐lethal haemorrhagic complications were not mentioned in the conference abstract.

Non‐liver, non‐kidney solid organ transplantation

No RCTs were identified that reviewed thromboprophylaxis (anticoagulant or antiplatelet) therapies in recipients of heart, lung, pancreas, bowel, or any other solid organ transplantation. The remainder of this review will focus solely on studies pertaining to liver or kidney transplantation.

Kidney transplantation

Two studies (Horvath 1975; Zhang 2009a) did not discuss their methods of random sequence generation or allocation concealment and so it was unclear to what extent their methods introduced selection bias. 

Zhang 2005 reported on patients undergoing acute rejection episodes confirmed by needle biopsy. Their methods of randomisation were not well described, but as they stated that patients were randomised based on biopsy results this was deemed to be at high risk of bias. 

Osman 2007 used a randomisation method that was deemed to be at low risk of bias. They described using sealed opaque envelopes that were kept in the operating theatres and were only opened once the patient was enrolled.

Kauffman 1980 and Ubhi 1989 both used patients hospital numbers to randomise their patients and were therefore judged to be at high risk of bias (quasi‐RCT) as clinicians would have known which group the patient would have been entered into prior to enrolment. 

Kikic 2009 used a paired design to their study where patients were initially divided into two groups based on serum potassium levels. Each of those groups were then randomised using a blind envelope technique into treatment and control arms. As the initial division was based on laboratory values, this was assessed to be at high risk of selection bias, however as the remainder of their randomisation methods were low risk, we deemed it unlikely that this would have had a major impact on the quality of evidence.

Liver transplantation

Both Villamil 2016 and Mansourian 2014 stated that their patients were randomly placed into the two treatment groups, however their methods of randomisation were not described. As such, it was difficult to assess the potential degree of bias. However, Villamil 2016 had two treatment groups, one with 12 patients and one with 17 patients which raised suspicions about their randomisation methods and consequently we assessed this study to be at high risk of bias. Mansourian 2014 seems to have included patients with portal vein thrombosis at the time of transplantation and were started on heparin therapy subsequently. These patients were then only included in the study if Doppler ultrasound studies confirmed portal vein patency. We believed this study to be at high risk of selection bias, as this decision relied on a significant degree of subjective operator assessment.

Kidney transplantation

Horvath 1975 and Kikic 2009, described adequate double blinding methods and were deemed to be at low risk of bias.

Zhang 2005 and Zhang 2009a were single blind studies where the assessors were not blinded to the outcomes. However, as the only outcome relevant to this review was patient survival, lack of blinding was unlikely to bias this outcome so these studies were deemed to be at low risk of bias.

Kauffman 1980 used a single blind methodology, however one of the outcomes that they were investigating was transfusion requirements and it was certainly possible for the treating physician to be aware of the treatment allocation and influence the decision to transfuse. We judged this study to be at high risk of bias.

Ubhi 1989 appears to be an unblinded study, although details of specific methodology were lacking from the text. The authors did report on a number of thromboembolic events including renal vein thrombosis however, it was not mentioned whether renal vein thrombus was partial or complete, or whether there were any differences in the number of scans performed in the different groups. Due to lack of blinding, surgeons were likely to have a lower threshold to perform Doppler US in those not receiving heparin than those receiving heparin. We judged this study to be at high risk of detection bias.

Osman 2007 was unblinded and we had concerns about their methods of patient selection. As such, this study was judged at being at high risk of bias.

Liver transplantation

Neither Mansourian 2014 nor Villamil 2016 specifically mentioned blinding in their conference abstract and it is unclear to what extent, if any, that blinding was used. Both studies reported on death and thrombotic complications and while it is very unlikely that lack of blinding would not have influenced mortality reporting, it is certainly possible that lack of blinding may have affected the clinicians' threshold for investigating for a thrombosis. In essence, if a clinician knew a patient was receiving a particular therapy, they may have been more or less assiduous in requesting investigations to evaluate the graft vasculature. 

Neither study described their criteria for investigating thrombotic complications in detail, with Villamil 2016 simply stating: “clinical and Doppler USS evaluations were performed immediately postoperatively and at different time points for the first 6 months.” Furthermore, Villamil 2016 neither predefined what they considered to be a thromboembolic event nor what threshold of severity was used.

Although the information was limited, we deemed it likely that blinding was not used to a sufficient level to prevent the introduction of performance bias and therefore judged these studies as being at high risk of bias.

Kidney transplantation

There were no concerns about any of the included studies being at high risk of selective reporting.

Liver transplantation

Mansourian 2014 reported on death due to haemorrhagic complication; however, they do not report on death from other outcomes or other haemorrhagic complications. This raised concerns that this study may have been at high risk of reporting bias. However, as all our assessors had available to them was the conference abstract, it was concluded that there was insufficient information available to permit judgement. 

Kidney Transplantation

Osman 2007 excluded 45 “risky” patients for a number of reasons including young age, extent of atherosclerotic disease, history of thromboembolic phenomena, and intraoperative technical difficulty. However, none of these exclusion criteria seemed to be predefined and it appears to be down to the surgeon's discretion as to which patients were deemed “risky”. For these reasons, we judged this study to be at high risk of inclusion bias.

There were no other potential sources of bias identified across the remainder of the included studies.

Liver Transplantation

Villamil 2016 stated that they included patients at “high thrombotic risk” however, there was no further detail given on how this decision was made. As such, we judged this study to be at potentially high risk of inclusion bias. 

Anticoagulants versus placebo

Three studies compared heparin (unfractionated or low molecular weight) to placebo (Horvath 1975; Osman 2007; Ubhi 1989).

• Horvath 1975: 18 patients were each administered 2500 U of subcutaneous heparin pre‐operatively and continued at 2500 U every 12 hours for 17 days

• Osman 2007: three‐armed study; 25 patients received 5000 IU of UFH twice daily for 1 week, 25 received a dose of prophylactic subcutaneous LMWH (tinzaparin sodium, once daily 3500 anti‐Xa IU for 1 week post‐operatively), and 25 received no heparin.

• Ubhi 1989: 32 patients received 5000 IU of subcutaneous heparin twice daily for 7 days or until the patient became fully mobile. 

Antiplatelet interventions

Kauffman 1980, Zhang 2005 and Zhang 2009a studied antiplatelet interventions.

• Kauffman 1980 compared buffered aspirin and dipyridamole following deceased donor kidney transplantation. Patients randomised to the aspirin group received buffered aspirin (5 mg twice/day). The dipyridamole group received dipyridamole (100 mg 4 times/day).

• Zhang 2005 and Zhang 2009a included only patients experiencing acute rejection and randomised into a group receiving lipo‐PGE₁ plus low dose heparin and a "control" group. The aim of these studies was to investigate the impact of platelet aggregation on acute rejection, not to investigate graft thrombosis.

  • In Zhang 2005, patients with confirmed acute rejection episodes following renal transplantation were randomly assigned Lipo PGE₁ (plus low dose heparin) therapy or a control group. In the intervention group, 20 μg/day of Lipo PGE₁ and 5 mg/day of heparin were administered for 2 weeks if no bleeding was present.

  • In Zhang 2009a, the intervention group received Lipo‐PGE₁ (20 μg/day) and heparin (10 mg/day), administered until reversal of rejection if no bleeding was present.

Anticoagulant during pre‐transplant dialysis

Kikic 2009 was the only study which aimed to assess the impact of pre‐operative haemodialysis. One of the two sub‐studies in Kikic 2009 studied anticoagulation in pre‐transplant haemodialysis. One group (66 patients) received heparin (1000 U/hour) and the control group (44 participants) received trisodium citrate (25 to 50 mmol/hour via arterial line). The study did not include early‐graft thromboses or non‐graft thromboses as an outcome, so the relative thromboprophylactic efficacies of heparin and citrate dialysis are very unclear. This study had a high risk of random sequence generation bias, as allocation was based on tests (pretransplant levels of serum potassium).

Kikic 2009 reported that citrate dialysis may make little or no difference versus heparin dialysis to the rate of bleeding complications (P = 0.56), the one‐year death‐censored graft survival rates (P = 0.44), delayed graft function (P = 0.28), or cellular rejection (P = 0.29) (very low certainty evidence). Seven bleeding complications occurred post‐transplant in each group. One‐year death‐censored graft survival rates were 88% in heparin group and 90% in citrate group respectively (P = 0.44). The study reported five cases of early C4d‐positive graft dysfunction in the heparin group, but there were none in the citrate group (P = 0.08).

Hospital stay, death, blood transfusion requirements, readmissions to intensive care, and patient survival rates were not reported. Overall, there is insufficient evidence to assess the impact of heparin and citrate anticoagulation in preoperative haemodialysis (very low certainty).

Aspirin versus warfarin

Mansourian 2014 (139 patients) compared the risk of portal vein rethrombosis in patients receiving aspirin with patients receiving warfarin post liver transplantation. They focused only on patients with pre‐existing portal vein thrombosis (9.6% of screened patients). Portal vein rethrombosis did not occur in either group. There were two deaths in the warfarin group due to bleeding complications (one with intracranial haemorrhage and one with internal bleeding) and none in the aspirin group. The abstract did not mention the number of patients in each group so no statistical comparisons can be made.

The rate of major bleeding complications, hospital stay, blood transfusion requirements, readmissions to intensive care, patient and graft survival rates were not reported. Mansourian 2014 had a high risk of random sequence generation bias and detection bias, with unclear bias in most other domains. Overall, there is insufficient evidence to compare the impact of aspirin and warfarin therapy post liver transplantation (very low certainty).

Pre‐transplant plasmapheresis plus post‐transplant anticoagulation versus post‐transplant aspirin ± LMWH

Villamil 2016 (29 patients) compared the impact of pre‐transplant plasmapheresis plus post‐transplant anticoagulation therapy versus postoperative standard low dose aspirin ± LMWH therapy. They looked exclusively at patients with antiphospholipid antibodies or positive lupus anticoagulant who they additionally deemed to be at high risk of thrombotic complications (no definition of high risk was given); 9.0% of screened patients. The study did not describe which anticoagulation therapy was used in the plasmapheresis arm and it was also unclear the dose of LMWH in the control arm.

Villamil 2016 reported that pre‐transplant plasmapheresis plus post‐transplant anticoagulation therapy may reduce the rate of "thrombotic complications" (not defined).Thrombotic complication rates were 37.9% in aspirin ± LMWH group and 10.3% in plasmapheresis group, respectively (P < 0.0001). They report 3/17 patients in the plasmapheresis group developing antiphospholipid associated complications (catastrophic antiphospholipid syndrome (2), and hepatic artery thrombosis (1)) compared to 11/12 patients in the aspirin ± LMWH group (cerebrovascular ischaemia (3), humeral thrombosis (2), hepatic artery thrombosis (1), intestinal ischaemia (1), retinal artery thrombosis (1), portal vein thrombosis, catastrophic antiphospholipid syndrome (4)). Comparison of early graft thrombosis could not be completed as the number of portal vein thromboses in the plasmapheresis group was not given. This study had high performance bias and high detection bias; "thrombotic complications" was not defined. The number of scans performed in the different groups is not mentioned and the method of randomisation is not described.

The study reported that pre‐transplant plasmapheresis plus post‐transplant anticoagulation therapy may make little or no difference to graft survival (P = 0.41) or death (P = 0.09). Graft survival was 91.7% in the aspirin ± LMWH group and 100% in the plasmapheresis group. There was no difference in the development of catastrophic antiphospholipid syndrome between groups. There were more deaths in the aspirin ± LMWH group than the plasmapheresis group (5 and 2 respectively).

Hospital stay, bleeding events, blood transfusion requirements, and readmissions to intensive care were not reported. Overall, there is insufficient evidence to compare plasmapheresis plus anticoagulation with aspirin ± LMWH therapy in patients at high risk of thrombosis plus antiphospholipid antibodies or positive lupus anticoagulant (very low certainty).