S t u d y o f s e v e r a l a c q u i r e d a n d g e n e t i c
f a c t o r s i n r e l a t i o n w i t h o u t c o m e i n k i d n e y t r a n s p l a n t a t i o n
U n i v e r s i t é L i b r e d e B r u x e l l e s
D e p a r t m e n t o f N e p h r o l o g y
L i d i a G h i s d a l
2 Thesis submitted in fulfilment of the requirements for the degree of Doctor in Medical Sciences ‐ Université Libre de Bruxelles‐ 2011
3 Promotor
Prof. Dr. Marc Abramowicz (Department of Medical Genetics, Hôpital Erasme, ULB, ruxelles)
B
Copromotor
rof. Dr. Daniel Abramowicz (Nephrology Department, Hôpital Erasme, ULB, Bruxelles) P
Jury
Chair: Prof. Dr. Alain Lemoine (Nephrology Department, Hôpital Erasme, ULB, Bruxelles) Secretary: Prof. Dr. Marc Abramowicz
Copromotor: Prof Dr. Daniel Abramowicz Jury Members:
s) Prof. Dr. Françoise Fery (Endocrinology Department, Hôpital Erasme, ULB, Bruxelle Prof. Dr. Denis Franchimont (Gastro‐enterology Department, Hôpital Erasme, ULB, Bruxelles)
Prof. Dr. Massimo Pandolfo (Neurology Department, Hôpital Erasme, ULB, Bruxelles) External Experts :
, France) Prof. Dr. Didier Ducloux (Nephrology Department, CHU Saint‐Jacques, Besançon
Prof. Dr. Eric Thervet (Nephrology Department, Hôpital Necker, Paris, France)
4 Table of contents
CHAPTER 1: INTRODUCTION ...6
1.1 KIDNEY TRANSPLANTATION FOR ENDSTAGE RENAL DISEASE...6
1.2 OUT ...6
OMBOPHILIC FACTORS AND RENAL GRAFT OUTCOM COMES AFTER KIDNEY TRANSPLANTATION 1.3 THR ES...9
ORMAL HAEMOSTASIS AND THROMBOPHILIA... 1.3.1 N ...9
1.3.2 HAEMOSTASIS DISORDERS IN CHRONIC RENAL FAILURE...10
1.3.3 PREVALENCE AND IMPACT OF THROMBOPHILIC FACTORS IN KIDNEY ALLOGRAFT RECIPIENTS...10
WONSET DIABETES AFTE PLANTATION... 1.4 NE R TRANS ... 11
.... 1.4.1 DEFINITION ...11
1.4.2 INCIDENCE AND IMPACT OF NODAT IN RENAL TRANSPLANT PATIENTS...12
ISK FACTORS... 1.4.3 R ...13
S O THE THESIS... ... ... 1.5 AIM F ... ... ... 15
1.5.1 THROMBOPHILIC FACTORS IN KIDNEY RECIPIENTS...16
1.5.2 NEW ONSET DIABETES AFTER TRANSPLANTATION...16
CHAPTER 2 THROMBOPHILIC FACTORS IN KIDNEY TRANSPLANT RECIPIENTS ... 18
2.1 ARTICLE 1 ... 19
2.2 ARTICLE 2 ... 20
CHAPTER 3 NEWONSET DIABETES AFTER RENAL TRANSPLANTATION... 21
3.1 ARTICLE 3 ... 22
3.2 ARTICLE 4 ... 23
3.3 ARTICLE 5 ... 24
CHAPTER 4 DISCUSSION ... 25
4.1 THROMBOPHILIC FACTORS IN KIDNEY TRANSPLANT RECIPIENTS... 25
4 4 .2 NEWONSET DIABETES AFTER RENAL TRANSPLANTATION... 31
.3 CONCLUSIONS AND PERSPECTIVES... 36
SUMMARY (FRENCH)... 39
TABLES AND FIGURES... 45
ABREVIATION LIST... 53
REFERENCES ... 54
AKNOWLEDGEMENTS... 68
CURRICULUM VITAE... 69
5 Chapter 1: Introduction
nal disease 1.1 Kidney transplantation for end‐stage re
1.2 Outcomes after kidney transplantation
omes 1.3 Thrombophilic factors and renal graft outc
s after transplantation 1.4 New‐onset diabete
1.5 Aims of the thesis
6 Chapter 1: Introduction
1.1 Kidney transplantation for end‐stage renal disease
Chronic kidney disease (CKD) is a major public health problem. Adjusted incidence of stage V CKD is reaching 350 per million per year in United States of America (USA) (www.usrds.org), and 190 per million per year in Belgium (www.era‐edta‐reg.org).
Renal replacement therapy is available in three different modalities: haemodialysis, peritoneal dialysis and kidney transplantation from either a living or a deceased donor.
Although the number of renal transplantations remained stable for 20 years, the number of patients registered on waiting lists has increased dramatically. At the end of 2010, 10768 patients were registered on the Eurotransplant waiting list for a kidney transplantation (including 914 patients from Belgium), whereas only 3705 were transplanted in 2010. The waiting time for patients on active waiting list is increasing, with a median time of 42 months at the end of 2010 (www.eurotransplant.org).
Currently, renal transplantation is the renal replacement therapy associated with the best patient survival and quality of life, in adults and in children and is considered as the optimum treatment for eligible patients [1‐3]. This survival benefit is also observed in the elderly population (> 70 years old) that is growing progressively [4]. On a health economic point of view, renal transplantation is cost‐effective when compared with haemodialysis [5]. In the context of organ shortage, it is important to improve renal graft survival. The identification of risk and prognosis factors may help the clinicians to customize the management and the immunosuppression therapy of kidney transplant patients.
O m k io
1.2 utco es after idney transplantat n
In the sixties, the association of azathioprine with corticosteroids as immunosuppression was the first step to achieve graft survival that makes renal transplantation a good option as renal replacement therapy. The introduction of cyclosporine in the late seventies improved dramatically the outcome of renal transplantation and allowed corticosteroid dose reduction, resulting in a major increase in the number of renal transplantations performed [6]. The widespread use of cyclosporine associated with a marked decline in acute rejection episodes in the early
7 transplantation period mainly contributed to the improvement of short term graft survival.
Since their introduction during the nineties, tacrolimus and mycophenolic acid (MPA) progressively replaced cyclosporine and azathioprine respectively. The association of anti‐CD25 monoclonal antibodies (IL2 receptor antagonists) as induction with tacrolimus+MPA+corticosteroids as maintance therapy is currently the most prescribed combination therapy in the USA [7]. The switch to tacrolimus and MPA was based on early reports showing lower acute rejection rates in comparison with cyclosporine and azathioprine. However, recent reports could not confirm the superiority of tacrolimus over cyclosporine in reducing acute rejection rate and improving 5 years graft survival.
Likewise, recent comparisons between azathioprine and MPA showed no difference in acute rejection rate, or showed lower acute rejection rate with MPA, not translated in a significant benefit of graft survival [8;9].
However, long‐term renal graft survival remained largely unchanged over the last 2 decades. At first sight, it might be surprising that long ‐term graft survival rates have not improved over this period. Several factors might contribute to this observation. First, as described above, new immunosuppressive drugs (tacrolimus and MPA) failed to improve graft survival. Second, the aging of the recipient population and the higher proportion of diabetic recipients have affected the predominant cause of graft failure:
death with a functioning graft. Third, the average quality of grafted organs declined (older deceased donor, extended criteria used) [10]. On the other hand, improvement of surgical procedures, better allocation system as well as better management of post‐
transplant complications contributed to improve patient and graft survival.
Our limited understanding of the causes of graft failure is a major obstacle in building strategies to improve long‐term outcome, which is the goal of transplantation. The causes of kidney allograft loss were recently dissected in a cohort of 1317 kidney recipients, transplanted between 1996 and 2006 and receiving standard quadruple immunosuppressive therapy (Figure 1). Death with a functioning graft accounted for nearly half of the graft losses. The leading cause of mortality is cardiovascular, in diabetic patients particularly. The accelerated cardiovascular disease observed is due to the cumulative effect of residual pre‐existing cardiovascular risk factors acquired before transplantation and post‐transplant complications such as new onset diabetes after
8 transplantation (NODAT), dyslipidemia, hypertension, obesity and suboptimal renal function [11]. Early graft loss by primary non function is mainly due to non‐
immunologic events such as vessel graft thrombosis. The progressive graft failures are due to immunologic processes such as transplant glomerulopathy, fibrosis/atrophy due to recurrent rejections, acute rejection or to non‐immunological processes such as recurrent or de novo glomerulopathy, fibrosis/atrophy associated with non‐
immunological events (BK polyoma virus, toxicity of calcineurin inhibitors) and intercurrent medical/surgical conditions [12].
These data showing the causes of graft failure underscore the need of progress in the understanding and in the prevention of non‐immunological as well as immunological causes of graft loss. The Nephrology Department of ULB‐Erasme Hospital has spent efforts to identify clinical and innate as well as acquired biological factors that associate with major outcomes in renal transplant patients. In this context, the Nephrology Department has set up a biobank since 2001, prospectively collecting genomic DNA and serum from kidney transplant recipients. Clinical data from more than 2500 renal transplantations performed in our institution since 1965 were collected in an electronic database. Both clinical data and biological samples were used to perform association studies with major outcomes in renal transplant recipients. The research reported in the present thesis was initially started up in order to address issues raised by several clinical observations.
Historically, we started to study coagulation disorders in kidney transplant recipients from the observation of kidney graft vessels thrombosis after the prophylactic administration of high doses of muromonab (anti‐CD3 monoclonal antibody;OKT3) [13;14]. On the other hand, several studies reported surprisingly high incidence of thrombosis and acute rejection rates in recipients with thrombophilic factors. In this context, we initiated in 2001 a prospective study evaluating the prevalence and the impact of thrombophilic factors in our kidney transplant population, in collaboration
l
with the Hemato ogy department of ULB‐Erasme Hospital.
Likewise, based on the early observation that patients receiving tacrolimus as main immunosuppression drug frequently developed de novo diabetes, we decided to switch such patients to cyclosporine and we reported our single‐centre experience. In order to better understand risk factors associated with this metabolic complication, we decided
9 in 2007 to study the genetic susceptibility of NODAT, in collaboration with the experimental genetics unit (Laboratoire de génétique, Institut de Recherche Interdisciplinaire en Biologie Humaine et Moléculaire, Prof. Marc Abramowicz). This study was made possible by the setting up of a belgo‐french consortium, including at this time 4 different renal transplant centres (ULB‐ Erasme Hospital, Brussels, CHU of Tours, CHU of Limoges, and CHRU of Lille, France). The collection of clinical data and the
make this trial successful.
merging into a joined database were critical steps to omes 1.3 Thrombophilic factors and renal graft outc 1.3.1 Normal haemostasis and thrombophilia
The primary hemostasis phase includes vasoconstriction, platelet adhesion followed by aggregation to the damaged vascular surface. The secondary hemostasis (coagulation phase) is leading to the clot formation through the tissue factor (extrinsic) and contact activation (intrinsic) pathways both converging in the final common pathway leading to the activation of factor X, thrombin and fibrin. This cascade is down‐regulated by natural anticoagulant mechanisms. The tissue‐factor‐pathway inhibitor forms a quaternary structure with tissue factor, factor VIIa, and factor Xa. The thrombomodulin–protein C–
protein S pathway inactivates factors Va and VIIIa. Antithrombin III inactivates factors XIa, IXa, Xa, and IIa in a reaction that is accelerated by the presence of heparan sulfate. In the fibrinolytic pathway, tissue‐type plasminogen activator (t‐PA) and urokinase‐type plasminogen activator (u‐PA) convert plasminogen to plasmin. Once generated, plasmin proteolytically degrades fibrin [15].
Thrombophilia is defined as a tendency to develop thromboembolism as a consequence of predisposing factors that may be genetically determined, acquired or both. The frequency of each single inherited thrombophilic factor and the associated relative risk of thrombo‐embolic event have been well described in the general population [16]. On the other hand, several conditions can lead to the development of transient acquired thrombophilic factors, possibly associated with an increased risk of thrombo‐embolic events, while other circumstances are leading to false positive thrombophilic factors, not associated with thrombo‐embolic events. Factor V (FV) Leiden and prothrombin (G20210A) mutations are the most frequent inherited thrombophilia (5% and 2 % respectively). Antithrombin, protein C and protein S deficiencies due to rare mutations
10 in the general population are clearly associated with thrombo‐embolic events. Several conditions might lead to acquired deficiencies or might result in false positive tests, like liver disease (deficiency of the 3 factors), the use of oral anti‐coagulants (protein C and protein S deficiencies), oral contraceptive drugs (resistance to activated protein C and protein C deficiency), heparin (antithrombin deficiency), the presence of auto‐
antibodies (protein C and protein S deficiencies), or a nephrotic syndrome (protein S and antithrombin deficiencies) [17]. High level of factor VIIIc in the general population may be at least in part congenital but may also be acquired (stress, acute phase response, older age). Anti‐phospholipid antibodies and lupus anticoagulant are associated with primary or secondary antiphospholipid antibody syndrome. False positive lupus anticoagulant tests may be observed during oral anticoagulation therapy while antiphospholipid antibodies may be promoted by infectious diseases in the general population. Hyperhomocysteinemia can be caused by genetic disorders or be acquired (folic acid, vitamin B12 and vitamin B6 deficiencies, CKD, hypothyroidism, increasing age and smoking). A recent meta‐analysis, showed no significant thrombo‐
embolic risk associated with the methylenetetrahydrofolate reductase (MTHFR) C677T casians [18].
and platelet glycoprotein IIIa polymorphisms in Cau 1.3.2 Haemostasis disorders in chronic renal failure
Patients with end‐stage renal disease (ESRD) are prone to both hemorrhagic complications related to platelet dysfunction and thrombotic complications, mainly thrombosis of the haemodialysis vascular access. Early reports have described complex abnormalities of coagulation factors (quantity and/or functionality), as well as impairment of inhibitory and fibrinolytic pathways in ESRD [19]. Underlying factors such as type, location or stenosis of vascular access are well‐established risk factors of haemodialysis access thrombosis. On the other hand, studies evaluating the potential role of inherited and acquired thrombophilic factors are retrospective and produced conflicting results [20]. Large prospective cohort studies are needed to evaluate the risk of haemodialysis access thrombosis associated with thrombophilic factors.
1.3.3 Prevalence and impact of thrombophilic factors in kidney allograft recipients Before 2001, several retrospective (case‐control and cohort) studies have reported a high prevalence of specific thrombophilic factors and a spectacular impact on graft
11 survival, acute rejection and/or thrombo‐embolic events rates (Table 1). However, these studies were suffering from several limitations that precluded extrapolation for the current management of kidney allograft recipients. First, these studies were often based on small samples of patients and on a retrospective design, leading to potential biases.
Second, the number and the type of factors assayed was limited and differed from one study to another. Third, the use of anti‐aggregation or anti‐coagulation drugs is rarely reported. Moreover, it is questionable whether these results are still valid in the current era of quadruple modern immunosuppression characterized by much lower acute
rejection and early graft loss rates.
The guidelines developed in 2001 by the Clinical Practice Guidelines Committee of the American Society of Transplantation, concerning the evaluation of renal transplant candidates are very limited concerning the screening and the management of coagulopathies. The authors emphasized the lack of evidence of 1/the exact prevalence of coagulation disorders among renal transplant candidates 2/the impact of pre‐
transplant coagulation abnormalities on post‐transplant events and, 3/ the beneficial effects of anticoagulation[21].
In this context, we have set up in 2001 a prospective study to better define the prevalence and the impact of thrombophilic factors in kidney allograft recipients (Articles 1 and 2).
iabetes after transplantation 1.4 New‐onset d
1.4.1 Definition
NODAT is a serious and frequent metabolic complication after renal transplantation.
Over the last 50 years, the concept of NODAT has evolved in terms of denomination and definition. Before 2003, de novo diabetes that developed after transplantation was described in various terms, most frequently “Post‐transplantation diabetes mellitus” and suffered from a lack of consensus regarding its definition. The most commonly used clinical definition was the requirement of insulin for a minimum period post‐
transplantation (often 30 days). This definition, however, identified only the most severe cases, leaving out the majority of patients with glucose metabolism disorders.
This entity is currently well defined since the publication of the International consensus guidelines in 2003. It is recommended that the definition and diagnosis of NODAT
12 should be based on the American Diabetes Association (ADA) criteria for type 2 diabetes published in 2003 (endorsing revised diagnostic guidelines from World Health Organization (WHO) 1999) [22;23]. The definition has not been updated since this publication. The use of standardized A1C assay is recommended for the diagnosis of type 2 diabetes since 2009. However, several aspects argue against its implementation for the screening of NODAT.
1.4.2 Incidence and impact of NODAT in renal transplant patients
The reported incidence of NODAT greatly depends on the length of follow‐up, the diagnostic criteria and the immunosuppression used. The true incremental incidence of diabetes occurs mainly during the first 6 months post‐transplantation, when patients are treated with high doses of immunosuppressive drugs. After 6 months, the annual incidence of diabetes is similar to that observed in patients on the graft waiting list (approximately 6% per year) [24]. Thus, late‐onset cases of NODAT may be difficult to distinguish from genuine cases of type 2 diabetes mellitus. The most accurate incidence of NODAT under calcineurin inhibitor (CNI) therapy is provided by the DIRECT (Diabetes Incidence after Renal Transplantation: Neoral C2 Monitoring Versus Tacrolimus) study. This study prospectively evaluated the incidence of glucose metabolism disorders, within the first 6 months post‐renal transplantation, defined by ADA 2003/WHO criteria, in recipients treated by a standard regimen including CNI, MPA, steroids, and induction with basiliximab (IL2 receptor antagonist). The incidence of NODAT or impaired fasting glucose (IFG) (primary endpoint) was 29.8% (incidence of NODAT alone: 20.5%) [25].
Renal transplant recipients with NODAT exhibit diabetic complications that are similar to those seen in the general population with type 2 diabetes (ketoacidosis, hyperosmolarity, peripheral circulatory disorders, renal, ophthalmic, and neurological complications). These complications seem however to develop at an accelerated rate. A large US registry study showed that 58.3% of patients with NODAT developed at least 1 complication by 3 years post‐transplantation [26]. As a consequence, NODAT is associated with worse outcomes after renal transplantation, such as a higher risk of major cardiovascular events , graft failure , death‐censored graft failure and death [27;28] In addition, this metabolic complication substantially increases medical costs ($
11,797 additional Medicare payments during the first post‐transplant year) [24].
13 1.4.3 Risk factors
Several risk factors are commonly associated with both NODAT and type 2 diabetes, while others are related to immunosuppression.
1.4.3.1 Pre‐transplant risk factors commonly shared with type 2 diabetes in the general population
Older age and higher body mass index (BMI) are strong independent risk factors of NODAT [28;29]. African American and Hispanic recipients have a higher risk to develop NODAT in comparison with white patients [28;29]. The higher frequency of HLA*A28 might contribute to this higher risk in black recipients [30]. A family history of type 2 diabetes emerged as a significant risk factor in multivariate analysis of several studies [31;32]. Hepatitis C virus (HCV) is increasing the risk, through impaired insulin sensitivity mediated by alterations in peroxisome proliferator‐activated receptor α and γ pathways [28;29;33‐35]. Conflicting results exist regarding the relationship between cytomegalovirus infection and type 2 diabetes in general population as well as in transplant patients. Two retrospective studies reported a higher incidence of NODAT in patients with metabolic syndrome at baseline [36;37]. The risk of NODAT increases stepwise with pre‐transplant fasting plasma glucose (FPG) level (FPG:101‐110 mg/dL, odds ratio (OR):1.5 and FPG: 110‐125 mg/dL, OR:7.6) [38]. There is also a relationship between the pre‐transplant 2hour plasma glucose level observed after an oral glucose tolerance test (OGTT) and the risk of post‐transplant hyperglycaemia (OR:1.26 per 1mmol/L or 18 mg/dL) [39]. Pre‐transplantation hypertriglyceridemia was also shown to correlate with NODAT [40;41]. Hypomagnesemia has been associated with metabolic syndrome, type 2 diabetes and obesity in the general population [42].
Hypomagnesemia induced by CNI (more frequently with tacrolimus), is due to renal magnesium wasting occurring through transcriptional inhibition of the renal magnesium transporter in the distal collecting tubule (TRPM6 gene) [43]. Recently, post‐transplantation hypomagnesaemia was found to be an independent predictor of NODAT in both renal and liver transplant recipients [41;43]. The mechanisms whereby hypomagnesemia may induce or worsen existing diabetes are not well understood.
Hypomagnesemia is involved in the secretion of insulin (competition with calcium influx), in the transduction of insulin signal (phosphorylation of insulin receptor thyrosine kinase and other protein kinases in the signal transduction cascade) and in the
14 secretion of various effectors involved in insulin resistance (adipokines and other interleukins). On the other hand, type 2 diabetes could also be involved in low magnesium levels found in the diabetic population, as insulin has been implicated in enhancing renal magnesium reabsorption [44].
ted to immunosu
1.4.3.2 Specific factors rela ppression
The diabetogenic effect of corticosteroids, mainly due to insulin resistance is mediated by both impaired insulin‐dependent glucose uptake in the peripheral tissues and enhanced gluconeogenesis in the liver [45]. High‐dose steroid regimens used during the 70’s of the previous century were associated with a very high incidence of so‐called
“steroid diabetes”, which declined when cyclosporine was introduced as an immunosuppressant in the 80’s [46]. Pulse corticosteroid therapy given in the context of acute rejection is still an independent risk factor of NODAT (OR: 2.8) [47]. Calcineurin inhibitors are diabetogenic by inducing a defect in insulin secretion by interfering with nuclear factor of activated T (NFAT) cell signalling in pancreatic βcells. This pathway triggers the expression of genes critical for βcell function, including at least 6 genes that are known to be mutated in monogenic forms of hereditary diabetes [48]. Tacrolimus induces a reversible suppression of insulin secretion mediated by an inhibition of insulin gene transcription. This effect is mediated by the binding of tacrolimus (FK506 was the name of the investigational medicinal product) to FK506 binding protein‐12 (FKBP‐12) and a subsequent inhibition of calcineurin in the beta‐cells [49]. The high level of FKBP‐12 in pancreatic βcells might explain why tacrolimus more profoundly inhibits insulin secretion than cyclosporine. Registry analyses, meta‐analyses and the prospective DIRECT study showed that the risk of NODAT was significantly higher in patients on tacrolimus versus cyclosporine [25;28;29;50;51]. In the prospective DIRECT study, 26.0% of patients on cyclosporine developed NODAT or IFG versus 33.6% on tacrolimus (P=0.046) and 8.9% of patients on cyclosporine required hypoglycaemic drugs versus 16.8% on tacrolimus treated patients (P=0.005) [25]. The risk of NODAT related to tacrolimus is dose‐dependent and high trough‐levels increase this risk, in particular during the early post‐transplant period [47;51;52]. Based on early studies reporting a two‐fold increase of the risk under tacrolimus versus cyclosporine, we started to switch recipients from tacrolimus to cyclosporine, after the occurrence of NODAT. We retrospectively reviewed a cohort of 54 renal transplant recipients who
15 developed NODAT while being treated with TRL and we compared glucose metabolism parameters of patients switched to cyclosporine to patients maintained on tacrolimus (Article 3). There is now strong evidence that mammalian target of rapamycin (m
TOR) inhibitors increase the risk of NODAT [53;54]. This diabetogenic effect is on of an insulin secretion defect and insulin resistance [55].
probably due to a combinati 1.4.3.3 Genetic background
Heritability of type 2 diabetes is strong in the general population. During the last decade, several studies evaluated the possible association of NODAT with single nucleotide polymorphisms (SNPs) in genes of interest, e.g., genes already related to type 2 diabetes in the general population (Table 2). Small sample sizes and the absence of replication studies in independent cohorts precluded any definitive conclusions. With Genome Wide Association (GWA) studies evaluating type 2 diabetes phenotype, the genetic background of NODAT could be evaluated more precisely. GWA studies dissect genetic susceptibilities in complex diseases by scanning the whole genome for associated markers. Since the first GWA study in 2007, a total of approximately 40 confirmed loci have been associated with type 2 diabetes in the general population. The effect size of genetic variants discovered is modest, with an OR ranging from 1.10 to 1.20 and reaching in non‐obese patients 1.55 for rs7903146 SNP (T allele), a common variant in TCF7L2 (transcription factor 7‐like 2) gene [56]. Most of these genes are associated with decreased insulin secretion due to defective beta‐cell function (TCF7L2, KCNJ11, MTNR1) or beta‐cell mass (CDKAL1, CDKN2A, CDKN2B). Focusing on results from the first wave of GWAs [57], we have evaluated the association between 11 type 2 diabetes susceptibility genes (1 SNP per gene) and the risk of NODAT during the first 6 months
n a large Caucasian cohort (Article 4).
post‐transplantation, i 1.5 Aims of the thesis
Long term graft survival did not improve during the last 2 decades despite huge progress achieved in the knowledge and the management of immunological risk factors.
Improvement in patient and graft outcomes should also focus on non‐immunological risk factors. These observations gave rise to the formulation of the following
nvestigational aims.
i
16 1.5.1 Thrombophilic factors in kidney recipients
Although trials have investigated coagulation disorders in ESRD patients at the time of renal transplantation, several issues remained unclear and raised the following questions:
What is the impact of thromboplilic factors on patient and graft outcomes in the current era of immunosuppression?
What is the exact prevalence of a large panel of known thrombophilic factors in a stage V CKD population?
tion?
What is the proportion of thrombophilic factors corrected after renal transplanta Should we perform a full screening before kidney transplantation in all patients?
In order to answer these questions, our research evaluated the impact of thrombophilic factors in stage V CKD patients on patient and renal graft outcomes (Article1) as well as the prevalence and the proportion of thrombophilic factors corrected one month after renal transplantation (Article2).
1.5.2 New Onset Diabetes After Transplantation
As previous reports showed that cyclosporine‐treated patients had a lower rate of NODAT, than tacrolimus‐treated patients, we started to switch patients who developed NODAT under tacrolimus to cyclosporin. After near a decade of switch experience, we reported our single centre experience. The aim of the study was to retrospectively evaluate the efficacy and the safety of a switch strategy from tacrolimus to cyclosporine in kidney recipients developing NODAT (Article 3).
As NODAT and type 2 diabetes were known to share several epidemiological risk factors, we have searched for an association between NODAT and the genes recently associated with type 2 diabetes. The aim of our study was to analyze the potential association between 11 well‐
established type 2 diabetes susceptibility genes and the occurrence of NODAT within the first 6 months post‐transplantation in a large cohort of 1229 predominantly Caucasians kidney recipients (Article 4). The choice of candidate genes was mainly based on the results of the first wave of GWAs revealing unsuspected genes associated with type 2 diabetes.
17 Based on our experience and on literature, we reviewed the factors contributing to the risk of NODAT and the strategies related to modifiable factors, with emphasis on practical issues. We elaborated recommendations in order to improve both risk assessment and management of NODAT (Article 5).
18 Chapter 2 Thrombophilic factors in kidney transplant recipients
2.1 Article 1
Thrombophilic factors do not predict outcomes in renal transplant recipients under acetylsalicylic acid [58].
prophylactic 2.2 Article 2
Thrombophilic factors in stage V chronic kidney disease patients are largely corrected by renal transplantation [59].
19 2.1 Article 1
20 2.2 Article 2
21 Chapter 3 New‐onset diabetes after renal transplantation
3.1 Article 3
Conversion from tacrolimus to cyclosporine A for new‐onset diabetes after
transplantation: a single‐centre experience in renal transplanted patients and review of [60].
the literature 3.2 Article 4
orphism associates with New‐Onset Diabetes after Transplantation [61].
TCF7L2 polym 3.3 Article 5
New‐onset Diabetes after renal Transplantation: Risk assessment and management [62].
22 3.1 Article 3
23 3.2 Article 4
24 3.3 Article 5
25 Chapter 4 Discussion
s
4.1 Thrombophilic factors in kidney tran plant recipients
The prospective study evaluating the prevalence and the impact of thrombophilic factors in renal transplant patients initiated in 2001 (Article 1 and 2) showed three important results. First, stage V CKD patients have a high prevalence (80.6%) of at least 1 thrombophilic factor when a large panel of 11 factors is systematically screened.
Second, the majority of these thrombophilic factors is corrected 1 month after transplantation. Third, the presence of at least 1, 2 and even 3 thrombophilic factors is not associated with the occurrence of thrombo‐embolic event or acute rejection in a cohort of renal transplant patient treated with standard modern immunosuppression and receiving acetylsalicylic acid (or LMWH in specific conditions) as prophylaxis.
The results of our prospective study have contributed to clarify the exact prevalence and the impact of pre‐transplant coagulation abnormalities on post‐transplant events. The lack of evidence concerning these important issues has been highlighted in the previous guidelines of the American society of Transplantation[21].
We reported a very high global prevalence of thrombophilic factors after a screening of 11 common thrombophilic factors (80.6%). The prevalence of single thrombophilic factors was however similar to the prevalence reported in other stage V CKD populations. The prevalence of antithrombin, protein C and protein S deficiencies, of elevated factor VIIIc, antiphospholipid antibodies and positive lupus anticoagulant was higher than in the general population, while the prevalence of prothrombin (G20210A), GPIIIa (T1565C) and FV Leiden variants was similar (Table 3). It is the first time that a simultaneous screening of 11 factors was assayed in such population, allowing the calculation of a global prevalence of thrombophilic factors. The study of Knoll et al.
reported a global prevalence of 43.2% after a screening of 8 thrombophilic factors [20].
Hypercoagulability state seen in stage V CKD might contribute to the higher rate of athero‐thrombotic cardiovascular events (myocardial infarction, stroke, peripheral vascular disease) and haemodialysis access thrombosis observed. Such events are promoted by well‐known risk factors (such as diabetes and hypertension) and probably by haemostatic changes. Coagulation disorders seen in CKD patients are not fully described and clearly understood to date and there is a lack of up‐dated research
activation that we and other observed in stage V CKD patients [66].
Taking our results and previous published data together, we can postulated that the higher prevalence of thrombophilic factors observed in stage V CKD patients is caused by a general coagulation activation and other specific conditions resulting in the detection of acquired thrombophilic factors. Protein C, protein S and antithrombin deficiencies are due to rare mutations in the general population but may be acquired in certain circumstances. In our cohort, the use of oral anticoagulation explained a significant proportion of false positive protein C deficiency (33.3%) and protein S deficiency (50%). The prevalence of antithrombin deficiency was higher in hemodialyzed patients in comparison with patients on peritoneal dialysis, while the deficiency was absent in stage V CKD patients not on dialysis, suggesting the potential 26 focused on this topic. It has been shown that coagulation is enhanced (reflected by increased levels of prothrombin fragments F1+2 and of tissue factor) in CKD patients, and even more in hemodialyzed patients. The haemodialysis therapy itself is enhancing coagulation, while the type of membrane used seems not to play a role. The turbulent blood flow resulting into platelet aggregation and leukocyte activation (with tissue factor expression) might trigger this activation. We showed that the prevalence of thrombophilic factors (at least one) was significantly higher in patients under dialysis (74%) than in stage V CKD patients not yet on dialysis (52%). This difference might be explained by both the lack of residual renal function in dialyzed patients and the direct role of dialysis in triggering coagulation activation. There is evidence of both intrinsic and extrinsic pathway activation of coagulation. The endothelial dysfunction markers that have been positively correlated with plasma creatinine might also play a key role in coagulation activation trough the pivotal role of platelet in the regulation of coagulation and fibrinolysis. Impairment of natural anticoagulant and fibrinolytic processes have been also reported [63;64]. Microparticles with procoagulant activity (MP‐PCA) have also been involved in coagulation activation. MP resulting from cell activation and apoptosis are increased by uremic toxins, inflammatory state, oxidative stress and complement activation. In addition of being a marker of endothelial cell damage, platelet‐ derived MP can directly activate coagulation via the exposition of phosphatidylserine and tissue factor on outer membrane and the interaction with factor Va, VIII and IXa, thereby facilitating assembly of prothrombinase complex [65].
Therefore, one may hypothesize that MP‐PCA account for the global coagulation
27 role of heparin administered during haemodialysis (false positive results). High level of factor VIIIc in the general population is at least in part congenital but may also be acquired. The high prevalence of high factor VIIIc observed in our population might be explained by stress, acute phase response and older age often observed in stage V CKD patients. We reported a high proportion of stage V CKD patients with positive lupus anticoagulant (37.7%) or positive antiphospholipid antibodies (29.3%). The majority of patients with positive lupus anticoagulant were not taking oral anticoagulant (small proportion of false positive results). Since the first report in 1990, many studies have reported the high prevalence of positive lupus anticoagulant and/or antiphospholipid antibodies in patients under dialysis [67‐77]. The heterogeneity in the prevalence reported (Table 3) is obviously related to the heterogeneity of populations studied and the methodology used for the screening. We have used 2 different tests (PTT‐LA and DRVVT) with both a screening and a neutralization procedure, as recommended [78].
However, in our study, as well as in most of the other published studies, we did not perform a second test after 6 weeks, as recommended. The cause of enhanced antiphospholipid antibody synthesis and positive lupus anticoagulant test in stage V CKD patients remains unclear. It has been shown that the level of antiphospholipid antibody titers observed in hemodialyzed patients is much lower than those observed in patients with systemic lupus erythematous [67;68]. The type of membrane used for haemodialysis seems not to play a role [69;72;73;79]. Several hypotheses have been suggested to explain the high prevalence of positive lupus anticoagulant and/or antiphospholipid antibodies in patients under dialysis, such as: regular exposition to endotoxin present in the dialysate, autoimmunity exacerbated by bioincompatible surfaces as dialyser and a marker of oxidant stress [72]. In our cohort, we could not show a role of dialysis method, while Sitter et al. reported a higher prevalence of anti‐
69].
phospholipid antibodies associated with haemodialysis versus peritoneal dialysis [
We have evaluated the correction rate of thrombophilic factors other than the 3 constitutional mutations (FV Leiden, prothrombin and GPIIIa) genotyped at baseline (Article 2). One month after renal transplantation, the global prevalence of thrombophilic factors dropped from 74% to 44.7% (P<0.001). The prevalence of antithrombin, proteinC and protein S deficiencies decreased significantly, reaching low prevalence as observed in the general population. While lupus anticoagulant prevalence dropped close to the prevalence observed in the general population, antiphospholipid
28 antibodies were still present in 12.6%, one month after transplantation. A more extended correction might be possible with a longer follow‐up period, given the long half‐life of IgG. Near half of patients normalized factor VIIIc level, while 18.6% of transplanted patients developed de novo a high level of factor VIIIc. This observation is in agreement with the study of Huser et al. showing a transient elevation of factor VIIIc after renal transplantation, with normalization 4 months after transplantation [80]. One can hypothesize that the acute phase response associated with surgery may trigger the synthesis of factor VIIIc.
The correction of thrombophilic factors after renal transplantation might be explained by both the restoring of renal function and the absence of exposition to dialysis. One month after transplantation, the mean glomerular filtration rate was 65mL/min in our cohort. The suboptimal renal function seen in renal transplant patients might explain the slightly higher incidence of thrombophilic factors in comparison with general population (Table 3). Interestingly and similar to our observations related to coagulation factors, a group evaluating kinetics of MP‐PCA during haemodialysis and 3, 6, 9 and 12 months after renal transplantation showed that 1/ the level of MP‐PCA in hemodialyzed patients was significantly higher than in healthy controls, 2/ MP‐PCA decreased significantly at 3 months after transplantation and remained stable but higher than controls thereafter, 3/ MP levels were negatively correlated with renal function [66]. Along this line, the presence of inflammatory and pro‐coagulant biomarkers (fibrinogen, factor VIIc, factor VIIIc, D‐dimer, plasmin‐antiplasmin complex levels) has also been correlated with renal impairment in several reports [81‐83].
We showed the lack of significant association between thrombophilic factors and the occurrence of thrombo‐embolic events or acute rejection episodes during the first year post‐transplantation (Article 1). While the two most recent series also failed to show any detrimental effect of thrombophilic factors on acute rejection, thrombo‐embolic events or graft survival [84;85], some earlier studies have reported major deleterious impact of thrombophilic factors on acute rejection rates , thrombo‐embolic events, cardio‐vascular events and graft survival after renal transplantation (Table 1). Several factors probably contribute to explain these major differences in outcomes. First, we observed a low rate of events (acute rejection: 13.9%, thrombo‐embolic event: 3.5%), under modern quadruple immunosuppression. Second, our patients received 100mg of
29 acetylsalicylic acid orally started before surgery and continued ad vitam. Prophylaxis with low‐molecular‐weight heparin (LMWH) was considered only in case of an well‐
identified risk factor of thrombo‐embolism: a) a clinical history of hypercoagulable state, such as repeated thromboses of vascular access for haemodialysis, or a history of deep venous thrombosis (DVT); b) the presence of a lupus anticoagulant when associated with previous thrombosis; and c) a risk of graft vessels thrombosis, discovered at
r , i r
su gery which in the op nion of the su geon required prophylaxis.
A total of 11 patients developed a thrombo‐embolic event after transplantation: 2 arterial graft thromboses, 1 venous graft thrombosis and 8 DVT. Four of them (36%) had a previous history of DVT and received low molecular‐ weighted heparin after transplantation. One was carrier of the prothrombin mutation and another was carrier of Factor V Leiden mutation. In our cohort, prothrombin mutation or factor V Leiden mutation was associated with a 3 fold increase in the risk of thrombo‐embolic event (the
i
associat on reached not the significant level).
In our cohort, three patients with a previous history of lupus nephritis were transplanted. One had a positive lupus anticoagulant with a previous history of haemodialysis access thrombosis. The patient received LMWH after transplantation and developed a hemorrhagic shock. The second had a previous history of positive lupus anticoagulant and antiphospholipid antibodies without thrombosis. The patient received LMWH after transplantation and developed hemorrhagic complications. The third one had borderline anti‐phospholipid antibodies levels (13U), received no heparin and did not develop any thrombotic event. When we looked at the 7 patients presenting high titers of anti‐phospholipid antibodies (>40U), only one had a previous history of dialysis access thrombosis and none of them developed thrombo‐embolic event after transplantation. The presence of these antibodies has been related to the occurrence of thrombotic complications in early reports. However, the clinical significance of such antibodies as well as lupus anticoagulant in CKD patients free of systemic lupus erythematosous (SLE) or anti‐phospholipid syndrome is still far from established.
Multivariate analysis of more recent and larger studies could not confirm an association between anti‐cardiolipin antibodies, lupus anticoagulant and vascular access thrombosis [20;72]. It has been shown that patients with positive anti‐cardiolipin antibodies in the context of a true anti‐phospholipid antibody syndrome had a high risk of thrombo‐
30 embolic complications after transplantation, whereas, it was not the case for patients with isolated anti‐cardiolipin antibodies [86].
Should we perform a systematic or a selective screening in renal allograft candidates?
Although controversy still exists currently in the general population, it is admitted to perform a screening in case of personal history of thrombosis younger than 50 years, thrombosis at an unusual anatomic site or extensive thrombosis and in case of a striking family history of thrombo‐embolism. Women with repeated spontaneous abortions should also be screened for anti‐phospholipid syndrome [87;88]. In this situation, the screening is beneficial for the patient as the results may influence directly a (primary or secondary) prevention strategy [89]. It is probably not reasonable to perform a systematic large screening of thrombophilic factors in stage V CKD renal transplant candidates. First, such screening will be positive (at least one factor) in almost all patients, thus the results will not help clinician to discriminate high risk patients.
Second, the results might lead to misinterpretation and inadequate anticoagulation, possibly leading to hemorrhagic complications. Third, currently there is no high level evidence on how to treat patients at risk. Finally, the screening is expensive and there is no evidence of cost‐effectiveness of such approach. Our results as well as the current guidelines are supporting the realisation of a thrombophilic screening in case of a past history (personal or familial) of thrombo‐embolic events, recurrent haemodialysis vascular access thrombosis, spontaneous abortion or systemic lupus erythematosus (anti‐phospholipid antibodies and lupus anticoagulant screening) to identify patients having a high risk of thrombo‐embolic event [21].
The decision for prophylaxis should take into account the individual thrombo‐embolic risk and the risk of bleeding. The current guidelines are suggesting the use of perioperative aspirin and/or anticoagulation in particular conditions (history of thrombosis with coagulation abnormality). Currently, there is no recommendation for a systematic prophylaxis. Several strategies have been suggested to prevent thrombo‐
embolic events in renal transplant patients, but the current evidence is limited to restrospective series mostly. The study of Friedman et al. is showing a reduction of allograft thrombosis rate in patients with both clinical risk factors and demonstrated thrombophilia, receiving post‐operative intravenous unfractionated heparin, followed by warfarin. The authors showed in addition that the screening in patients without
31 clinical risk factor was not cost‐effective [90]. One study showed that post‐operative dalteparin (LMWH) administration was safe and efficient in preventing thrombo‐
embolic events after renal transplantation [91]. Two case control studies reported a lower incidence of graft vein thrombosis with 75mg of acetylsalicylic acid given daily systematically for one month after renal transplantation [92;93]. In our experience, systematic prophylaxis with 100mg of acetylsalicylic acid or subcutaneous enoxaparin with a switch to aspirin after 3 to 5 days in selected at‐ risk patients (if the patient had a history of thrombo‐embolic event or a graft thrombosis risk discovered at surgery) was safe and the rate of thrombo‐embolic event was low (3.5% at 1 year). Graft thrombosis risk factors related to surgical procedure make also part of the decision of anticoagulation. Well recognized risk factors are: poor surgical technique, vessel torsion, kinking or compression, multiple donor renal vessels, discrepancy in the size between donor and recipient vessels, especially in pediatric transplantation,
toma and stenosis [94].
hypotension/hypoperfusion, renal artery atheroma 4.2 New‐onset diabetes after renal transplantation
We observed an incidence of NODAT reaching 19% in our centre between 1997 and 2005. We showed that 42 % of patients switched from tacrolimus to cyclosporine (N=34) experienced a resolution of NODAT (ADA 2003 criteria) while this never occurred in patients remaining on tacrolimus (N=20), after a follow‐up of 1 year (Article 3). In the converted group, insulin therapy could be stopped completely in 31%
of insulin‐treated patients and the dose could be reduced significantly in the other patients. Five other single‐centre retrospective studies reported as our group either a complete resolution or a significant improvement of NODAT after conversion from tacrolimus to cyclosporine. In the first study, 50% of switched patients could stop glucose‐lowering therapy versus no patients continuing tacrolimus [95]. In the second study, 80% of patients could stop insulin after conversion to cyclosporine [96]. In a study that included both renal and liver recipients, resolution or improvement of NODAT occurred in 78% of patients [97]. Another study evaluating the same switch strategy in liver transplant recipients reported a resolution in 22% and an improvement of NODAT in 60% [98]. The last study performed in 39 liver transplant recipients reports a resolution of NODAT in 36%, 1 year after switching [99]. Most of these studies are however limited by the small sample size and the absence of a control arm.
32 Our study does not provide a definitive proof of efficacy, as patients were not randomized in two arms. However, several arguments are supporting a causality link between conversion to cyclosporine and the remission of NODAT. First, FPG and insulin dose rapidly declined early after conversion (with a slower A1C decline). Second, the remission cannot be completely attributed to the decrease of the prednisolone dose, as the dose lowering was more progressive in the converted group (lowering of 12.5% of the dose between t0 and 3months) than in the control group (lowering of 53.8% of the dose between t0 and 3months). Third, we did not observe spontaneous reversal of NODAT in patients remained on tacrolimus, although it was previously described [47].
One may argue that this strategy will result in an undesirable increase of the cardiovascular risk and decrease of graft outcome. In our cohort, blood pressure and lipid level remained stable, although the use of a statin drug increased significantly and the use of anti‐hypertensive increased numerically (P=NS). Early reports have claimed that a switch from cyclosporine to tacrolimus improved the cardiovascular profile. A recent study prospectively following 540 kidney recipients for a median period of 4.7 years reported both the theoretical cardio‐vascular Framingham risk score and the true number of events observed by CNI type. Although cyclosporine was associated with higher cholesterol, lower glomerular filtration rate and higher systolic blood pressure, the risk of major cardiovascular events was more than 2 fold higher in the tacrolimus‐
treated patients. This apparent paradox might be explained by the higher rate of NODAT in tacrolimus‐treated patients and the higher proportion of patients under statin, converting enzyme inhibitor/ angiotensin receptor blockers and acetylsalicylic acid in cyclosporine‐treated patients [100]. In our experience, graft function remained stable and 1 episode of acute rejection occurred after the switch. It is now established that long‐term graft survival is identical with cyclosporine and tacrolimus [8].
Several other strategies regarding modification of immunosuppression might be considered in NODAT. A recent meta‐analysis of 30 randomized controlled trials showed that glucocorticoid withdrawal (discontinuation after some months) was not associated with a reduction of NODAT incidence, while avoidance (no steroids at all after transplantation) resulted in less NODAT requiring any treatment. However, both steroid‐sparing strategies were associated with higher acute rejection rates and higher risk of graft loss excluding death [101]. The treatment of acute rejection with pulse
33 steroids (followed by higher steroid dose as maintenance) may lead paradoxically to the outbreak or the worsening of NODAT. Therefore, in patients at high risk of NODAT, steroid minimization strategy should be balanced with immunological risk profile to avoid acute rejection. In one study, the discontinuation of CNI with replacement by sirolimus (m‐TOR inhibitor) failed to improve glucose metabolism of kidney transplant recipients and was even associated with a worsening of insulin resistance and an inappropriately low insulin response [102]. The risk of NODAT related to tacrolimus is dose‐dependent and high trough‐levels enhance this risk, in particular during the early post‐transplant period. To date no large prospective study evaluated a CNI minimization strategy in patients with NODAT [47].
What is the place of a switch to cyclosporine in the management of immunosuppression in patients with NODAT? Based on the results of our retrospective study, we initiated the REVERSE prospective study (EudraCT number: 2006‐001765‐42). Kidney recipients with NODAT persisting 6 months after transplantation, despite a maximal acceptable reduction/withdrawal of steroids for a period of 3 months are randomized to be switched to cyclosporine or to remain on tacrolimus. The primary endpoint is the remission rate of NODAT, one year after inclusion. The results of the study might confirm the efficacy and safety of the strategy. Based on our experience and retrospective studies, we have proposed an algorithm for the management of immunosuppression in order to both minimize the risk to develop NODAT and to improve NODAT in kidney transplant recipients (Article5). The choice of immunosuppression should first take into account the immunological risk of the patients to avoid acute rejection. In patients with a low immunological risk and a high risk of NODAT, the first choice might be a cyclosporine‐based or a belatacept (co‐
stimulation blocker)‐based immunosuppressive regimen. In case of NODAT, the first step includes hygieno‐dietetic recommendations (weight control, diet and exercice) and a reduction in the exposure to diabetogenic drugs (corticosteroids and tacrolimus). MPA should be closely monitored to avoid rejection in the context of steroid tapering. If NODAT does not improve under tacrolimus, a switch to cyclosporine might be considered in patients with high immunological risk, while a switch to belatacept might also be considered in patients with a low immunological risk.
size) or might merely reflect different mechanisms involved in NODAT.
The mechanistic link between TCF7L2 and the pathogenesis of diabetes is still under investigation. TCF7L2 as well as 6 other genes linked to type 2 diabetes in GWAs code for transcription factors that act through Wnt signaling pathway involved in pancreas development, islet function, and insulin production and secretion. Wnt ligands might also be involved in the cross talk between adipocytes and pancreatic β cells. It has been shown that carriers of the T allele of rs7903146 have higher levels of both TCF7L2 and 34 Focusing on results from the first wave of GWAs, we have evaluated the association between 11 type 2 diabetes susceptibility genes (1 tag SNP per gene) and the risk of NODAT during the first 6 months post‐renal transplantation, in a large (N=1076) Caucasian cohort (Article 4). We found that the rs7903146 variant of TCF7L2 was independently associated with NODAT. TCF7L2 codes for a transcription factor, TCF4, involved in Wnt signalling pathway. The CT genotype of rs7903146 increased the odds of NODAT by 70% and the TT genotype by 142% (reference =CC genotype). None of the 10 other type 2 diabetes‐associated SNPs reached statistical significance for the association with NODAT. The risk conferred by the genotype was independent of the main immunosuppression used. Another group found a significant association with the same variant of TCF7L2 gene (OR: 2.20), in a cohort of 589 Korean transplant recipients [103]. The association between this variant and NODAT found in two cohorts of different ethnicities is very likely to represent true genetic association, and strongly suggests a common genetic background between NODAT and type 2 diabetes mellitus.
Unlike our study, this Korean group also found a significant association with other type2 diabetes genes discovered in GWAs: SLC30A8, HHEX, CDKAL1, CDKN2A/B. This discrepancy is probably due to the fact that we have appropriately included only early onset NODAT, while the other study also included very late onset NODAT that might be genuine cases of type 2 diabetes (mean follow‐up period of 10,3 years). The difference of allele frequencies between ethnicities might also contribute to this difference. A recent meta‐analysis of 8 GWAs, confirmed the association between the 11 loci that we tested and type 2 diabetes. TCF7L2 is the most risky gene. The SNP rs7903146 that we genotyped was associated with type 2 diabetes with an OR of 1.55 (p=9.72 x 10-45 ) in non‐obese and an OR of 1.34 (p=4.34 x 10-10) in obese patients. The effect size of the ten other SNPs was smaller [56]. The lack of association with the 10 other SNPs with NODAT in our analysis might be due to a type II error (insufficient power to detect small effect
