Induction immunosuppression with polyclonal rabbit antithymocyte globulins (rATGs) has been used in kidney transplantation to prevent rejection in patients at higher risk (1). The long-term kidney graft function remains stable after rATG induction in the majority of patients; however, the immunologic mechanisms responsible for these effects are not fully understood. Several immunomodulatory mechanisms of rATG have been proposed, including T-cell depletion (by apoptosis, complement-dependent lysis, or Fc receptor-mediated antibody-dependent cellular lysis), modulation of cell surface antigens, and leukocyte-endothelium interactions (2–4). The effects of rATG have been studied primarily in T and B cells, although more recent data suggest that dendritic and natural killer cells are relevant targets for the immunosuppressive action of rATGs (5–7). Recently, rATG was shown to induce the expansion and enrichment of CD4+CD25+FoxP3+ regulatory T cells (Tregs) in vitro, which down-regulate the alloimmune response and have been implicated in tolerance induction in an experimental setting (8, 9). Similarly, Minamimura et al. (10) suggested that rATG may preferentially deplete conventional T cells but spare Tregs. Thymoglobulin and ATG-Fresenius S (ATG-F) represent rATGs that are approved for clinical use but differ in the manufacturing process: human thymocytes were used as the immunogen for creating Thymoglobulin, while Jurkat human cells were used to immunize rabbits in the development of ATG-F. Moreover, both rATGs have different immunomodulatory potential that is independent of their capacity to deplete T cells (11–13).
The efficacy of both Thymoglobulin and ATG-F has also been studied in clinical trials. As an induction agent, Thymoglobulin was shown to be superior in the prevention of rejection (14) and improvement of patient survival and graft acceptance (15). Other studies have reported that the efficacy of both biologics was similar (16, 17), although some differences in the observed side effects have been reported (14–18).
To date, little is known about the effects of rATG induction on the transcriptome of renal allografts. We hypothesized that Thymoglobulin and ATG induce unique transcriptome profiles, including genes involved in the maintenance of transplantation tolerance, when used as inductive agents in immunologically high-risk patients who had not experienced rejection and had normal kidney graft function 3 months postinduction. To investigate this possibility, we quantified the intragraft messenger RNA (mRNA) expression of 376 genes involved in tolerance, inflammation, T- and B-cell immune response, and apoptosis in 3 months protocol biopsy in patients who received induction immunosuppression with Thymoglobulin or ATG-F.
RESULTS
Patients' Demographic Data
Twenty-four patients were enrolled in the study. The demographics and distribution of clinical variables are summarized in Table 1. Nine low-risk patients did not receive induction therapy (control group), 10 high-risk patients received induction immunosuppression with Thymoglobulin, and an additional 5 patients received induction with ATG-F. Delayed graft function (DGF), which was defined as the requirement for dialysis within the first week after transplantation, developed in 3 of 5 grafts in the ATG-F group, 1 of 10 grafts in the Thymoglobulin group, and 1 of 9 grafts in the control group (P<0.05). Despite these differences in graft function development, the renal function did not differ among the studied groups at the time of biopsy.
TABLE 1: Patient's demographics and clinical characteristics
Clinical Follow-Up
The patients were followed for 2 years after transplantation, and no decrease in function, graft loss, or patient death was observed in either group (Table 1). One patient from the control group experienced a subsequent episode of clinical acute T-cell-mediated rejection at day 186 after transplantation because of noncompliance but was successfully treated with methylprednisolone.
T-Cell Depletion
T-cell depletion was effectively sustained in both studied groups, although absolute T-cell (CD3+) counts on day 6 were significantly lower in the Thymoglobulin group than the ATG-F group (9±6.2 vs. 16.4±6.42 cells/μL, respectively; P<0.01). However, the clinical relevance of this difference was negligible.
Transcriptional Analysis Distinguishes ATG-F Grafts From Thymoglobulin Grafts
Twenty-four renal biopsy samples were analyzed and the relative expression levels of 376 candidate genes were measured and normalized using the housekeeping gene glyceraldehyde 3 phosphate dehydrogenase (GAPDH). Despite stable renal function and identical histology with normal findings from protocol biopsies of patients in all three groups, the transcriptional pattern of the biopsies was markedly different in the ATG-F group compared with the Thymoglobulin and control groups. A comparison of mRNA expression profiles of the ATG-F group with the Thymoglobulin group identified 18 genes that were differentially expressed (Table 2; Fig. 1).
TABLE 2: Target genes differentially expressed in biopsies from patients treated with Thymoglobulin vs. ATG-Fresenius
FIGURE 1: Transcript level differences in biopsies of renal allografts from patients treated with Thymoglobulin vs. ATG-F. The relative transcript prevalence for each patient biopsy sample is shown in transcripts with significantly differential expressions when comparing Thymoglobulin and ATG-F group. Individual, log-scale, quantitative fold levels of RNA transcripts are shown for the control group (triangles), Thymoglobulin group (full circles), and ATG-F group (ATG, squares) biopsies referenced to the calibrator sample biopsy. Symbols located below the graph baseline indicate undetectable levels. Red bars represent the median fold level of RNA transcripts for each biopsy group.
The most significant finding was the differential expression of nuclear factor-κB (NF-κB)-associated genes, which suggested higher activity of this pathway in the ATG-F group. Specifically, the gene expressions of Toll-like receptor 4 (TLR4), transmembrane receptor of dendritic cells (CD209), which is an initiator of innate immunity through modulation of TLR, and myeloid differentiation primary response gene 88 (MYD88), which is an adapter protein used by TLRs to activate NF-κB, were all increased in the ATG-F group. The ATG-F group showed a significant decrease in NF-κB repressing factor (NKRF) expression, which is a mediator of transcriptional repression of certain NF-κB-responsive genes, prostaglandin E receptor 3 (PTGER3), which is a suppressor of proinflammatory factor production through inhibition of NF-κB activity (19), mitogen-activated protein kinase 8 (MAPK8), which is required for TNF-α induced apoptosis, and TNF-receptor superfamily 21 (TNFRSF21), which is an activator of NF-κB. The observed upregulation of the NF-κB pathway was validated by an enrichment analysis (P<0.05).
ATG-F induction also resulted in the upregulation of hyaluronan-mediated motility receptor (RHAMM) and thromboxan a synthase 1 (TBXAS1) and genes involved in dendritic cell function (CLEC4C), costimulation (CD80 and CTLA4), apoptosis (NLRP1), and chemoattraction (CCR10) compared with Thymoglobulin. In contrast, Thymoglobulin induced the upregulation of adenylate kinase 3-like 1 (AK3L1), which is a novel immune tolerance marker (20), CXCL14, which is a chemotactic factor for immature dendritic cells, TMEM176B, which is a factor that suppresses the maturation of dendritic cells, and MTHFR. In regards to the described tolerogenic potential of rATG in vitro, we only detected FoxP3 transcripts in a few biopsies, and FoxP3 mRNA expression did not differ between the studied groups (data not shown).
Hierarchical Clustering
Transcriptome analysis confirmed that the transcripts in the Thymoglobulin group were distinct from the ATG-F group. Supervised hierarchical clustering analysis revealed that the main split in the dendrogram clearly distinguished the ATG-F group from the Thymoglobulin group. Moreover, gene expression profiles in the control group were similar to the Thymoglobulin group (see Figure, SDC 1,https://links.lww.com/TP/A595). Unsupervised hierarchical clustering analysis using 18 differentially expressed genes from a univariate analysis demonstrated a different gene expression pattern for ATG-F samples based on clear separation from the other samples (Fig. 2).
FIGURE 2: Unsupervised hierarchical clustering analysis using 18 differentially expressed genes. Unsupervised hierarchical clustering analysis demonstrated a different gene expression pattern for ATG-F samples by their clear separation from the others.
The Observed Difference Between Two rATG Agents Is Not a Consequence of DGF
To determine whether the different gene expression profiles were related to DGF, an additive experiment was performed. In total, 33 renal allografts were included in the additive study, including 19 grafts with primary graft function and 14 grafts with DGF. The patient demographics and clinical characteristics are shown in Supplemental Table 1 (see Table, SDC 2,https://links.lww.com/TP/A596). The additive experiment did not show a statistically significant difference in the expression of the investigated genes between grafts with and without DGF (Table 3). These data provided additional evidence that the differences observed in our study were not the consequence of DGF but were due to the different rATGs used.
TABLE 3: Expression profiles of target genes from patients with delayed graft function vs. primary function
DISCUSSION
Differences in the efficacy and safety profile between rATGs have been observed in previous clinical trials. In this study, we showed that Thymoglobulin induction was associated with down-regulation of NF-κB signaling, dendritic cell function, chemoattraction, apoptosis, and costimulation in rejection-free grafts from patients with normal histology and stable function compared with ATG-F. We found that the transcriptome of patients treated with these two regimens differed in a cluster of genes associated with the NF-κB pathway and dendritic cell activity, which are involved in innate and adaptive immunity, respectively.
It is possible that kidney transplant recipients with normal histological findings in protocol biopsy may differ in their molecular phenotype, which could ultimately determine the clinical variability observed posttransplantation. The ongoing allogeneic immune response, which was only manifested by distinct molecular changes and the absence of morphological and functional alterations, may explain the different posttransplant outcomes in patients with identical histological findings. Therefore, it is hypothesized that the molecular heterogeneity of such biopsies is a consequence of the different immunosuppressants used.
In our study, induction therapy with Thymoglobulin in high-risk patients was shown to induce an identical transcriptome profile as low-risk subjects who had not received induction therapy but differed from patients treated with ATG-F. The homogeneity of the gene expression profile from patients in the Thymoglobulin and control groups suggests a similar level of immune response to the graft despite the apparent difference in immunologic risk. These observations strengthen previous studies showing that Thymoglobulin induction is a safe and efficient therapy for sensitized patients to prevent rejection after kidney transplantation. Importantly, this therapy option is recommended by Kidney Disease: Improving Global Outcomes guidelines (1, 21, 22).
Thymoglobulin and ATG-F were previously shown to have different immunomodulatory potential, and Thymoglobulin was hypothesized to have stronger antirejection effects (11–13). The segregation of ATG-F from the Thymoglobulin pattern in unsupervised hierarchical clustering suggests that different mechanisms are involved in the regulation of an allogeneic response. Here, we show that the differences in gene expression patterns predominantly involved the upregulation of the innate immune system through activation of the NF-κB signaling pathway and genes involved in dendritic cell function, which were observed in the ATG-F group. Recently, increasing evidence has supported a role for the innate immune system in acute allograft rejection (23). Dendritic cells, the most potent antigen-presenting cells of the innate immune system, are regarded as potential targets for suppressing alloreactivity and induction of allograft tolerance. In addition, it has been reported that rATGs interfere with DC differentiation, maturation, and immune function through NF-κB activation (24).
NF-κB regulates the expression of genes involved in ischemic-reperfusion injury, graft rejection, and transplant tolerance. NF-κB is also activated in acute kidney injury, and its activation correlates with the severity of inflammation (25). Moreover, inhibitors and antagonists of NF-κB activation have been shown to have a beneficial effect in an experimental model of kidney injury (26, 27). Together, these findings suggest that NF-κB activation plays a critical role in renal inflammation.
In this study, we showed that the expression level of genes involved in activation of the NF-κB signaling pathway was lower in renal allografts after induction treatment with Thymoglobulin compared with rATG. We also observed the upregulation of AK3L1 in Thymoglobulin allografts, which seems to be a potential novel tolerance marker, as fragments of this gene were identified in the serum of a rat model of acceptance (20). In contrast to Lopez et al. and Feng et al., who found an expansion of CD4+CD25+FoxP3+ Tregs in vitro (8, 9), we did not find differences in FoxP3 gene expression after rATG induction, which is a defining functional marker of this cell type. These results are in agreement with the observation by others showing that FoxP3 mRNA transcripts were not upregulated 2 years posttransplantation (28). Moreover, the optimal concentration of rATG for the in vitro generation of Tregs seems to be markedly lower than the serum concentration achieved after treatment with standard rATG doses (29).
Although inductive agents are excessively used in kidney transplantation, little is known about how these drugs maintain silencing of the immune system during the initial posttransplant period. Therefore, we chose to analyze biopsies 3 months posttransplantation to explore ongoing molecular processes in stable kidney allografts after different induction treatments were administered, in patients who had not showed previous rejections. To the best of our knowledge, this study is the first attempt aimed at analyzing this specific clinical setting. Although the study was limited by the sample size and descriptive nature, these findings provide new information on how the different rATGs control the immune response in vivo. In conclusion, we have demonstrated that Thymoglobulin and ATG-F induction induce a different intrarenal transcript profile in patients 3 months posttransplantation despite normal morphology and stable kidney graft function. Moreover, induction therapy with Thymoglobulin in high-risk patients was shown to induce an identical transcriptome profile in low-risk subjects who had not received induction therapy. The differential gene expression, which resulted in lower activity of NF-κB-dependent pathways, seems to favor better alloimmune regulation in Thymoglobulin induction.
MATERIALS AND METHODS
Study Design
To analyze the tolerance potential of rATGs, we first identified all patients receiving induction immunosuppression with Thymoglobulin or ATG-F treatment who had normal histology in protocol biopsy, stable kidney graft function without proteinuria or infection, and no rejection history. Protocol renal biopsies were performed at the third month posttransplantation during 2005 to 2008. Rigorous sample selection was based on normal morphologic findings according to the Banff 2005 classification and absence of acute tubular necrosis, interstitial fibrosis and tubular atrophy, and inflammatory infiltrate. All patients treated with rATGs had panel reactivity antibody more than 50% and were considered to be at high risk for rejection. The control group consisted of patients considered to be at low risk for rejection (panel reactivity antibody<20%; first kidney transplantation) and received calcineurin inhibitor-based immunosuppressants with mycophenolate mofetil (MMF) and steroids. All patients were monitored for 24 months. The Ethics Committee of the Institute for Clinical and Experimental Medicine approved the study protocol and all patients signed informed consent to participate in the study (Institutional Review Board approval number G08-08-10).
Therapy
Recipients at high risk of rejection received induction therapy with Thymoglobulin (Thymoglobulin, Genzyme Corporation, Cambridge, MA) or ATG-F (Fresenius AG, Bad Homburg, Germany) according to our center practice at certain time periods (ATG-F: 10/2005–03/2007; Thymoglobulin 03/2007 to present). The rATG therapies were initiated intraoperatively and followed daily until the sixth postoperative day according to the manufacturer's recommendation. In case of side effects, the rATG doses were reduced. The final cumulative doses of Thymoglobulin and ATG were 6.03±0.89 and 11.82±1.29 mg/kg, respectively. The efficacy of rATGs for depleting peripheral T cells was evaluated by determining the absolute CD3+ counts in peripheral blood on day 6. It is accepted that CD3+ T-cell counts below 50 cells/μL represent profound and adequate depletion (30). All patients received maintenance immunosuppressant therapy with Tacrolimus (Prograf; Astellas Pharma, Inc., Deerfield, IL), MMF (Cellcept; Roche Laboratories, Nutley, NJ), and prednisone.
Protocol Biopsy
Protocol kidney graft biopsy was performed in accordance with our center practice 3 months posttransplantation. Histological examination was interpreted according to the 2005 Banff working classification criteria. A small portion (2 mm) of the cortical zone of the biopsy specimen was immediately placed in the RNA later Stabilization Reagent (Qiagen) and stored at −20 or −80°C for RNA extraction.
RNA Extraction and Complementary DNA Synthesis
Renal tissue was homogenized and total RNA was isolated using RNA blue reagent (Top-Bio s.r.o., Czech Republic) according to the manufacturer's instructions. RNA was eluted in 30 μL of RNase-free water. The purity and concentration of the RNA were assessed on an ultraviolet-visible spectrophotometer (NanoDrop 2000, Thermo Scientific). The RNA isolation method routinely used in our laboratory was validated and standardized on reference samples, thereby eliminating errors and ensuring the same standard across all measurements. The quality of RNA samples obtained by standard isolation protocol was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies). An RNA integrity number of 8 to 10 indicated high-quality RNA that was suitable for further analysis. Samples were stored at −80°C. A total amount of 2 μg RNA was used for complementary DNA synthesis using Superscript Reverse Transcriptase (Nitrogen, CA) according to the manufacturer's instructions. RNA was treated with DNase, and only complementary DNA samples without genomic DNA contamination were used for TaqMan low-density array (TLDA) analysis.
Real-Time RT-PCR Analysis Using TLDAs
Gene expression profiling was performed using a custom-made TLDA to analyze 376 candidate genes known to have implications in the immune response (chemokine defense, apoptosis, inflammation, tolerance, and TGFβ signaling [see Table, SDC 3,https://links.lww.com/TP/A597]). A quantitative real-time Reverse-transcriptase polymerase chain reaction assay based on TLDA technology was performed as previously described (31) and data were quantified using the SDS 2.4 software package (Applied Biosystems, Foster City, CA).
Expression Data Analysis
Relative gene expression values were generated from TLDA analysis using the comparative 2−ΔΔCt method for relative quantification (RQ) (32), which is implemented in the Applied Biosystems RQ Manager Software v1.2.1 (Applied Biosystems). To calculate the RQ of target genes in transplanted kidney after induction with Thymoglobulin compared with ATG, one sample from the control group was designated as the calibrator. Using the 2−ΔΔCt, the data are presented as the fold change in gene expression normalized to an endogenous reference gene and relative to the no induction control.
Additive Experiment
To exclude the effect of DGF on the study results, we additionally evaluated the expression patterns of genes that significantly differed between the Thymoglobulin and ATG-F groups in biopsies obtained from patients 3 months posttransplantation who had experienced DGF or primary graft function. Grafts experiencing DGF due to surgical complications or rejection episodes were not included. All patients received an immunosuppressive regimen consisting of Tacrolimus, MMF, and steroids without induction treatment.
Statistical Analysis
Statistical analyses were performed using SPSS v.17.0 (SPSS, Inc., Chicago, IL) and GraphPad InStat v. 3.05 for Windows (GraphPad Software, San Diego, CA). A Student's t test and the analysis of variance were applied to the variables following a normal distribution. Based on the distribution of the gene expression data, we performed nonparametric testing and data were expressed as median and interquartile ranges. The significance of differential gene expression among the control, Thymoglobulin, and ATG-F groups was determined using Kruskal-Wallis test. Post hoc intergroup comparisons were made using the Mann-Whitney test with adjustment for multiple comparisons. Significance was defined as a two-sided P value less than 0.05. Unsupervised hierarchical clustering that grouped together genes with similar expression patterns was performed with StatMiner 3.0.0 Software (Integromix, Madrid, Spain) and MultiExperiment Viewer 4.6.0 (TM4, Boston, MA). The putative enrichment of the NF-κB pathway was performed using the Global test method as previously described (33, 34).
ACKNOWLEDGMENTS
The authors thank the transplantation coordinators, nurses, and patients for their cooperation and help. Special thanks to Jiri Belohradsky for his help with the gene enrichment analysis, Romana Polackova for expert technical assistance, and Vera Dankova for her help with data collection.
REFERENCES
1. Kidney Disease: Improving Global Outcomes (KDIGO) Transplant Work Group. KDIGO clinical practice guideline for the care of kidney transplant recipients. Am J Transplant 2009; 9 (suppl 3): S1.
2. Preville X, Flacher M, LeMauff B, et al.. Mechanisms involved in antithymocyte globulin immunosuppressive activity in a nonhuman primate model. Transplantation 2001; 71: 460.
3. Michallet MC, Preville X, Flacher M, et al.. Functional antibodies to leukocyte adhesion molecules in antithymocyte globulins. Transplantation 2003; 75: 657.
4. Beiras-Fernandez A, Chappell D, Hammer C, et al.. Impact of polyclonal anti-thymocyte globulins on the expression of adhesion and inflammation molecules after ischemia-reperfusion injury. Transpl Immunol 2009; 20: 224.
5. Daniel V, Naujokat C, Sadeghi M, et al.. Association of IL-12+ DC with high CD3+CD4-DR+ lymphocyte counts in long-term HIV-infected hemophilia patients with clinically stable disease. J Clin Immunol 2008; 28: 58.
6. Dalle JH, Dardari R, Menezes J, et al.. Binding of thymoglobulin to natural killer cells leads to cell activation and interferon-gamma production. Transplantation 2009; 87: 473.
7. Naujokat C, Berges C, Fuchs D, et al.. Antithymocyte globulins suppress dendritic cell function by multiple mechanisms. Transplantation 2007; 83: 485.
8. Lopez M, Clarkson MR, Albin M, et al.. A novel mechanism of action for anti-thymocyte globulin:
Induction of CD4+CD25+Foxp3+ regulatory T cells. J Am Soc Nephrol 2006; 17: 2844.
9. Feng X, Kajigaya S, Solomou EE, et al.. Rabbit ATG but not horse ATG promotes expansion of functional CD4+CD25highFOXP3+ regulatory T cells in vitro. Blood 2008; 111: 3675.
10. Minamimura K, Gao W, Maki T. CD4+ regulatory T cells are spared from deletion by antilymphocyte serum, a polyclonal anti-T cell antibody. J Immunol 2006; 176: 4125.
11. Rebellato LM, Gross U, Verbanac KM, et al.. A comprehensive definition of the major antibody specificities in polyclonal rabbit antithymocyte globulin. Transplantation 1994; 57: 685.
12. Bourdage JS, Hamlin DM. Comparative polyclonal antithymocyte globulin and antilymphocyte/antilymphoblast globulin anti-CD antigen analysis by flow cytometry. Transplantation 1995; 59: 1194.
13. Müller TF. Long-term T cell dynamics following the use of polyclonal and monoclonal antibodies. Graft 1999; 2.1 (suppl): 15.
14. Zuckermann AO, Grimm M, Czerny M, et al.. Improved long-term results with thymoglobuline
induction therapy after cardiac transplantation: A comparison of two different rabbit-antithymocyte globulines. Transplantation 2000; 69: 1890.
15. Opelz G, Naujokat C, Daniel V, et al.. Disassociation between risk of graft loss and risk of non-Hodgkin lymphoma with
induction agents in renal transplant recipients. Transplantation 2006; 81: 1227.
16. Schnetzler B, Leger P, Volp A, et al.. A prospective randomized controlled study on the efficacy and
tolerance of two antilymphocytic globulins in the prevention of rejection in first-heart transplant recipients. Transpl Int 2002; 15: 317.
17. Krogsgaard K, Boesgaard S, Aldershvile J, et al.. Cytomegalovirus infection rate among heart transplant patients in relation to the potency of antithymocyte immunoglobulin
induction therapy. Copenhagen Heart Transplant Group. Transplant Proc 1994; 26: 1718.
18. Ducloux D, Kazory A, Challier B, et al.. Long-term toxicity of antithymocyte globulin
induction may vary with choice of agent: A single-center retrospective study. Transplantation 2004; 77: 1029.
19. Ogawa M, Suzuki J, Kosuge H, et al.. The mechanism of anti-inflammatory effects of prostaglandin E2 receptor 4 activation in murine cardiac transplantation. Transplantation 2009; 87: 1645.
20. Sawitzki B, Bushell A, Steger U, et al.. Identification of gene markers for the prediction of allograft rejection or permanent acceptance. Am J Transplant 2007; 7: 1091.
21. Thibaudin D, Alamartine E, de Filippis JP, et al.. Advantage of antithymocyte globulin
induction in sensitized kidney recipients: A randomized prospective study comparing
induction with and without antithymocyte globulin. Nephrol Dial Transplant 1998; 13: 711.
22. Charpentier B, Rostaing L, Berthoux F, et al.. A three-arm study comparing immediate tacrolimus therapy with antithymocyte globulin
induction therapy followed by tacrolimus or cyclosporine A in adult renal transplant recipients. Transplantation 2003; 75: 844.
23. Obhrai J, Goldstein DR. The role of toll-like receptors in solid organ transplantation. Transplantation 2006; 81: 497.
24. De Creus A, Abe M, Lau AH, et al.. Low TLR4 expression by liver dendritic cells correlates with reduced capacity to activate allogeneic T cells in response to endotoxin. J Immunol 2005; 174: 2037.
25. Rangan GK, Wang Y, Tay YC, et al.. Inhibition of nuclear factor-kappaB activation reduces cortical tubulointerstitial injury in proteinuric rats. Kidney Int 1999; 56: 118.
26. Sanz AB, Sanchez-Nino MD, Ramos AM, et al.. NF-kappaB in renal inflammation. J Am Soc Nephrol 2010; 21: 1254.
27. Hutchings A, Wu J, Asiedu C, et al.. The immune decision toward allograft
tolerance in non-human primates requires early inhibition of innate immunity and
induction of immune regulation. Transpl Immunol 2003; 11: 335.
28. Louis S, Audrain M, Cantarovich D, et al.. Long-term cell monitoring of kidney recipients after an antilymphocyte globulin
induction with and without steroids. Transplantation 2007; 83: 712.
29. AbouZahr C, Cleland J, Coullare F, et al.. The way forward. Lancet 2007; 370: 1791.
30. Midtvedt K, Fauchald P, Lien B, et al.. Individualized T cell monitored administration of ATG versus OKT3 in steroid-resistant kidney graft rejection. Clin Transplant 2003; 17: 69.
31. Sanchez-Espiridion B, Sanchez-Aguilera A, Montalban C, et al.. A TaqMan low-density array to predict outcome in advanced Hodgkin's lymphoma using paraffin-embedded samples. Clin Cancer Res 2009; 15: 1367.
32. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative
PCR and the 2(-Delta Delta C(T)) method. Methods 2001; 25: 402.
33. Goeman JJ, van de Geer SA, van Houwelingen HC. Testing against a high dimensional alternative. J R Stat Soc Ser B Stat Methodol 2006; 68: 477.
34. Goeman JJ, Buhlmann P. Analyzing gene expression data in terms of gene sets: methodological issues. Bioinformatics 2007; 23: 980.