CLINICAL
AND
TRANSLATIONAL RESEARCH
Assessing the Metabolic Effects of Calcineurin Inhibitors in Renal Transplant Recipients by Urine Metabolic Profiling Binta Die´me´,1 Jean Michel Halimi,2,3 Patrick Emond,1,4 Matthias Bu¨chler,2,3 Lydie Nadal-Desbarat,1,5 He´le`ne Blasco,1,4 and Chantal Le Guellec3,4,6 Background. Biomarkers that can predict graft function and/or renal side effects of calcineurin inhibitors (CNI) at each stage of treatment in kidney transplantation are still lacking. We report the first untargeted GC-MSYbased metabolomic study on urines of renal transplant patients. This approach would bring insight in biomarkers useable for graft function monitoring. Methods. All consecutive patients receiving a kidney allograft in our transplantation department over a 6-month period were prospectively included and followed up for 12 months. We collected urine samples on the seventh day (D7) after transplantation, then at month 3 (M3) and month 12 (M12), and obtained mass-spectrometryYbased urinary metabolic profiles. Multivariate analyses were conducted to compare metabolic profiles at the 3 different periods and to assess potential differences between cyclosporine and tacrolimus. Differences in metabolic signatures were also assessed according to graft function at D7 and renal function at M3 and M12. Results. The urinary metabolic patterns varied over time in cyclosporine- and tacrolimus-treated patients and were somewhat different at D7, M3, and M12 between the 2 treatment groups. Principal metabolites that differed, regardless of the treatment used, were mainly sugars, inositol, and hippuric acid. Interestingly, among tacrolimus-treated patients, different metabolic signatures were found between patients with immediate or delayed graft function at D7. Conclusion. Urinary metabolomics represents a noninvasive way of monitoring immunosuppressive therapy in renal transplant patients. Although it is too early to consider it as a biomarker of CNI-induced injury or graft function, metabolomics appears a promising evaluation tool in this area. Keywords: Biomarkers, Metabolomics, Kidney transplantation. (Transplantation 2014;98: 195Y201)
alcineurin inhibitor (CNI)Yrelated nephrotoxicity remains a major cause of long-term graft loss in kidney transplantation. Specific biomarkers could be used not only to predict the occurrence of graft injury but also to allow a better mechanistic understanding of the underlying pathologic processes (1). Recently, innovative ‘‘omics’’ approaches have been evaluated, identifying sets of proteins or mRNA associated with clinical events (2Y4). Metabolomics focuses on low-molecular
C
Part of the financial support for chemical analysis has been provided by the Association pour le De´veloppement de la Recherche en Immunologie Clinique (ADRIC). The authors declare no conflicts of interest. 1 INSERM U930, Universite´ Franc¸ois-Rabelais, Tours, France. 2 Service de ne´phrologie-immunologie Clinique, CHU Tours, France. 3 EA 4245, Universite´ Franc¸ois-Rabelais, Tours, France. 4 Laboratoire de biochimie et biologie mole´culaire, CHU Tours, France. 5 PPF ‘‘Analyses des Syste`mes Biologiques’’, Universite´ Franc¸ois Rabelais, Tours, France. 6 Address correspondence to: Chantal Le Guellec, PharmD, Ph.D., Laboratoire de biochimie et biologie mole´culaire, CHU Tours, France. E-mail: [email protected] B.D. performed data and statistical analysis and contributed in writing the paper. J.M.H. designed the research, contributed in writing the paper, and was responsible for collecting the data and patients’ follow-up. P.E.
Transplantation
& Volume 98, Number 2, July27, 2014
weight compounds, allowing analysis of metabolite level changes in biological samples. Surprisingly, although these methods are widely applied to identify biomarkers in as many fields as cancer (5), neurologic diseases (6, 7), renal diseases (8), or druginduced nephrotoxicity (9), only few metabolomic studies have been conducted in renal transplantation. Some studies conducted in rats identified biomarkers of ischemia/reperfusion injury (10) or immunosuppressant-induced nephrotoxicity in transplanted (11) or nontransplanted (12Y14) animals. Conversely, studies in human are very scanty. We found 2 study participated in data analysis and in writing the article. M.B. participated in collecting the data and in the patients’ follow-up. L.N.D. participated in data analysis and writing the article. H.B.R. participated in research design, data analysis, and preparing the article. C.L.G. designed the research, participated in analyzing the data, and writing the paper. Supplemental digital content (SDC) is available for this article. Direct URL citations appear in the printed text, and links to the digital files are provided in the HTML text of this article on the journal’s Web site (www.transplantjournal.com). Received 28 October 2013. Revision requested 19 November 2013. Accepted 23 December 2013. Copyright * 2014 by Lippincott Williams & Wilkins ISSN: 0041-1337/14/9802-195 DOI: 10.1097/TP.0000000000000039
www.transplantjournal.com
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
195
196
www.transplantjournal.com
evaluating either short-term toxicodynamic effects of cyclosporine (CsA) on serum and urinary metabolite profiles in healthy volunteers (15) or time-dependent changes of serum metabolite profiles in patients treated with a CNI (16) and only a small study in five transplanted patients relating urinary metabolome to acute tubular injury (17). Very recently, Calderisi et al. used 1H-NMR (proton nuclear magnetic resonnance) to profile urines of kidney transplant patients during the 2 weeks after transplantation. We used gas chromatographyYmass spectrometry (GC-MS) to profile urine samples of kidney transplant patients receiving a triple-drug immunosuppressive regimen.
TABLE 1.
Transplantation
& Volume 98, Number 2, July 27, 2014
Using an untargeted approach, we analyzed whether the metabolic profile differed according to the CNI used (tacrolimus (Tac) or CsA) and, under a given treatment, if it varied over time during the first 12 months after grafting. In an attempt to derive potential biomarkers of kidney function, metabolic profiles were compared according to renal function at Day 7 but also at Month 3 (M3) and Month 12 (M12).
RESULTS Demographic and Clinical Characteristics The characteristics of the patients at the time of inclusion and during follow-up are given in (Table 1).Thirty-five patients
Patient characteristics
Characteristics Baseline characteristics of all patients Male/female Age median (range), years Weight median (range), kg Primary/secondary kidney transplant, number Type of donnor (cadaveric/living), number Cold ischemia duration, median (range), minutes Induction therapy anti-lymphocyte / basiliximab, number Acute rejection during the 12 months of follow-up , number (delay in days (D)) Characteristics of patients analyzed at D7 Creatininemia median (range), Kmol/L Creatininuria median (range), mmol/24 h Proteinuria median (range), g/24 h IGF, number of patients (%) SGF, number of patients (%) DGF, number of patients (%) UTI, number of patients (%) Viral infections, number of patients (%) Characteristics of patients analyzed at M3 Creatininemia median (range), Kmol/L Creatininuria median (range), mmol/24 h Proteinuria median (range), g/24 h eGFR median (range), mL/min/1.73 m2 eGFRQ60 mL/min/1.73 m2, number of patients (%) UTI, number of patients (%) Viral infections, number of patients (%) Diabetes mellitus, number of patients (%) Patients off-prednisone, number (%) Characteristics of patients analyzed at M12 Creatininemia median (range), Kmol/L Creatininuria median (range), mmol/24 h Proteinuria median (range), g/24 h eGFR median (range), mL/min/1.73 m2 eGFRQ60 mL/min/1.73 m2, number of patients (%) UTI, number of patients (%) Viral infections, number of patients (%) Diabetes mellitus, number of patients (%) Patients off-prednisone, number (%)
Tac group
CsA group
n=23 15/8 59 (16Y74) 73 (40Y105) 19/4 23/0 1086 (520Y1926) 10/13 1 (D10) n=18 195 (92Y1490) 10.4 (4.8Y18.9) 0.62 (0.22Y4.2) 10 (55%) 7 (39.5%) 1 (5.5%) 0 0 n=19 137 (78Y369) 11.1 (3.1Y29) 0.29 (0Y6.4) 54 (17.5Y105) 5 (26%) 0 0 0 1 (5.2%) n=19 129 (89Y356) 13 (6.5Y42) 0.26 (0Y1.28) 48 (14.7Y115) 6 (31.5%) 0 0 1 (5.2%) 11 (57%)
n=12 6/6 50 (20Y67) 75 (59Y111) 11/1 12/0 928 (731Y1960) 1/11 4 (D53, D69, D95, D140) n=9 146 (80Y464) 9 (6.2Y13.8) 0.84 (0.24Y2.9) 6 (67%) 2 (22%) 1 (10%) 0 0 n=9 133 (77Y206) 10.4 (2.4Y18.8) 0.16 (0Y0.95) 55 (48.3Y106) 5 (56%) 1 (11%) 0 0 0 n=10 109 (82Y171) 13.7 (7.1Y41) 0.13 (0Y1.22) 70 (63Y118) 10 (100%) 0 1 (10%) 3 (30%)
IGF, immediate graft function; SGF, slow graft function; DGF, delayed graft function; eGFR, estimated glomerular filtration rate; UTI, urinary tract infection; Tac, tacrolimus; CsA cyclosporine.
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
* 2014 Lippincott Williams & Wilkins
were included in the study but, as some urinary samples were missing, each of them was not equally represented at each sampling time. Antilymphocyte globulins and basiliximab were equally balanced as induction therapy in the Tac group, but basiliximab was the most frequently used in patients of the CsA group (92%). One patient per group required dialysis after transplantation (delayed graft function [DGF]), but the majority of patients had a good initial renal function. Renal function improved gradually over time in both groups. A higher percentage of patients had recovered a good renal function (eGFRQ60 mL/min/1.73 m2) in the CsA group as compared with the Tac group at M3 (56% versus 26%), but the difference was not significant (P=0.407). In contrast, the difference (100% versus 31.5%) was significant at M12 (P=0.0004). Some clinical events occurred during follow-up (urinary tract infections, viral infections, acute rejection, and diabetes mellitus). Those occurring surrounding sample collection times are listed in Table 1 but the weak number of events precluded analysis of their potential impact on metabolomic profiles. Metabolite Profiling Analysis Variation Over Time Orthogonal partial least-squares discriminant analysis (OPLS-DA) scores scatter plots showed that samples drawn at the different time-points were well segregated into different regions, regardless of the CNI used, CsA or Tac (Fig. 1A and B). Urines collected in patients treated with tacrolimus showed a clear clustering when comparing D7 versus M3 (Fig. 2A) and D7 versus M12 (Fig. 2B). In contrast, the model did not differentiate well the samples obtained at M3 and M12. This indicates that metabolic profiles significantly change between D7 and M3 and between D7 and M12. The set of metabolites discriminating between D7 and M3 and between D7 and M12 are shown in Table S1, SDC, http://links.lww.com/TP/A940. These are mainly sugars, inositol, hippuric acid, and an unidentified metabolite (M100T444). Univariate analysis revealed that urines at D7
Dieme et al.
197
contained significantly more inositol than that of the later periods and that hippuric acid and some unidentified metabolites were over represented at M12. Similar results were obtained in patients treated with CsA (Fig. 2C and D). Inositol and M100T444 were more abundant at D7, whereas lactate, citrate, and hippurate predominated at M12 (Table S2, SDC, http://links.lww.com/TP/A940). Variation of the Metabolic Pattern With the CNI Used Visual group separation was also observed in OPLS-DA scores scatter plots obtained at each single sampling time between patients treated by CsA or by Tac. However, model parameters were not excellent, and we were not able to obtain clear separation between CNI groups, particularly at D7 and M12 (low R2X and Q2 values and presence of outliers), although the discrimination was quite better at M3 (Figure S1, SDC, http://links.lww.com/TP/A940). The 3 models included a huge number of discriminative metabolites, and we did not try to identify them because of the poor model performances. Metabolite Pattern and Renal Function A clear separation between patients with immediate graft function (IGF, n=10) versus slow graft function (SGF or DGF, n=8) was obtained in the Tac group (Fig. 3). A similar proportion of patients received thymoglobulin induction in each group, that is, 3/10 (30%) in the IGF group versus 3/8 (37%) in the SGF/DGF group. The metabolites responsible for the separation between groups were inositol, hippurate, D-furanose, and other unidentified metabolites (Fig. 4 and Table S3, SDC, http://links.lww.com/TP/A940). Inositol and hippurate are more important in urine of patients with SGF or DGF than in IGF group. No such difference was obtained in the nine patients treated with CsA exhibiting either IGF (n=6) or SGF/DGF (n=3). Multivariate analysis failed to find any difference in the metabolite profiles between patients with ‘‘good’’ and ‘‘poor’’ renal function at M3, regardless of the CNI used. At M12, no difference was found in the Tac group, and as all patients
FIGURE 1. Orthogonal partial least squares discriminant analysis (OPLS-DA) score plot of urine samples obtained in (A) the Tac group at 7 days (medium gray), 3 months (light gray), and 12 months (black) after kidney transplantation and in (B) the CsA group at 7 days (medium gray), 3 months (light gray) and 12 months (black). For explanations about the OPLS-DA method, the reader is referred to the SDC, http://links.lww.com/TP/A940. Each point stands for the projection of a single sample on the main principal axes, corresponding to principal components, named either t[1], t[2], t0[1], or Num (in descending order of importance on projection), showing clear discrimination between samples drawn at D7, M3, or M12, whatever the CNI considered. The 3D orientation of each figure is the one enabling the best vision of the clusters.
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
198
www.transplantjournal.com
Transplantation
& Volume 98, Number 2, July 27, 2014
FIGURE 2. Pairwise comparison of time-dependent changes by OPLS-DA score plots, allowing visual inspection of discrimination between groups. Each point stands for the projection of a single sample on the main principal axes, corresponding to principal components, named either t[1] or t0[1] (in descending order of importance on projection). Also are given objective criteria for judging the quality of the separation: R2X and Q2 (see SDC, http://links.lww.com/TP/A940 for explanations of the meanings of these parameters). A, urines of D7 versus urines of M3 in patients treated with Tac (R2X=0.67; Q2=0.705). B, urines of D7 versus urines of M12 in patients treated with Tac (R2X=0.507; Q2=0.727). C, urines of D7 versus urines of M3 in patients treated with CsA (R2X=0.66; Q2=0.71). D, urines of D7 versus urines of M12 in patients treated with CsA (R2X=0.754; Q2=0.85).
of the CsA group had a ‘‘good’’ renal function, no comparison was performed.
DISCUSSION The present study shows that there is a change of urinary biochemistry over time and that this may correlate with graft function in tacrolimus-treated patients. One of our principal observations is that the metabolite profile at D7 is different from that at M3 and M12 regardless of the CNI used. After transplantation, the timeframe of kidney injury typically involves ischemia-reperfusion injury, followed by changes in protein expression and histologic modifications.
CNI-related nephrotoxicity, inflammation, and fibrosis are also responsible for histologic changes and alteration of kidney function over time (2). The time-dependent changes in urinary metabolite pattern found in our study indicate that this approach has the potential to reflect changes in kidney state during the first year after transplantation. We identified pentoses and hexoses as the metabolites that discriminate between D7 and the later periods, whatever the CNI used. However, precise identification of the true structure of these sugars was not achieved. Aside from sugars, inositol was more frequent at D7 (Tac and CsA groups), hippuric acid more frequent at M3 (Tac group) and M12 (Tac and CsA groups), and lactate and
FIGURE 3. OPLS-DA score plot for the comparison of metabolic profiles in patients with immediate graft function (IGF, black dots) or either slow or delayed graft function (SGF/DGF, gray dots) in the subgroup of patients treated with Tac. (R2X=0.539; Q2=0.874)
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
* 2014 Lippincott Williams & Wilkins
Dieme et al.
199
FIGURE 4. Box plots of univariate comparison of metabolite amounts (referred as area) between patients with immediate graft function (IGF) and either slow or delayed graft function (SGF/DGF) in the subgroup of patients treated with Tac. *PG0.05/n, with n is the number of metabolites tested (n=6 in this case) and considered for correction for multivariate analysis.
citrate more frequent at M12 (CsA group). Interestingly, a recent paper also identified time-dependent changes in 1H-NMR metabolomic profiles during the first 2 weeks after transplantation, which were representative of two stages of kidney recovery, after an initial postoperative stage (18). In this study, the metabolites that discriminated between early (postoperative) and later periods (2 weeks of follow-up) were mainly lipoproteins, creatine, and creatinine and, to a lesser extent, mannitol, acetate, and hippurate. From a mechanistic point of view, the urinary metabolic profile at D7 has the potential to reflect ischemia-reperfusion injury. Serkova et al. described the shift of the metabolic composition of the kidneys after cold storage alone and cold storage followed by transplantation in rats using 1H-RMN (10). Native renal tissue contained carbohydrates (glycogen and glucose) and triglycerides but also high concentrations of cellular osmolytes such as inositol, TMAO, and taurine as well as intermediates of the Krebs cycle (citrate), amino acids (glutamate, alanine, glutamine, and aspartate) and end products of glycolysis (lactate). After cold storage, there was a significant increase in glycogen and other carbohydrates as well as lactate in kidney extracts. They also showed a dramatic increase of allantoin, the end-product of xanthine metabolism and of uric acid, a marker of oxidative stress, in kidneys exposed to reperfusion. Similarly, Hauet et al. performed a 1H-NMR analysis of selected metabolites in an animal model of autotransplantation, the kidneys being cold-preserved for 48 h at 4-C (19). They showed an increase of biochemical markers of renal medulla damage (trimethylamine-N-oxide, TMAO) and
mitchondrial dysfunction (citrate, lactate, and acetate) across the 14 days after transplantation. Although our results are not directly comparable to other studies because of the different analytical methods used (GC-MS as compared with 1H-MNR), some similarities are observed in that carbohydrates, lactate, and citrate were also shown discriminant in our patients at D7. This indicates that stigmas associated to ischemia/ reperfusion may be detected by this approach in urines of transplanted patients. In addition to ischemia/reperfusion injury, the urinary composition at D7 may also be dependent on the additional nephrotoxicity of CNI. Urinary metabolite patterns were previously studied in transplanted rats receiving either CsA or sirolimus over a period of 7 days (11). By using an orthotopic transplant model, the authors avoided interference from immunologic processes but were able to analyze changes because of surgical procedure, ischemia/reperfusion, or acute drug nephrotoxicity. Contrary to the above studies, no difference in metabolic pattern was seen at day 7 between transplanted and nontransplanted rats who received no treatment, allowing the authors to conclude that the metabolic changes that they observed in treated animals (variable according to the immunosuppressant used) strictly arose from drug nephrotoxicity. Further experimental studies evaluating the time-dependent effects of CsA on kidney metabolism in nontransplanted rats showed that the 1H-NMR biochemical signature reflects mitochondrial dysfunction in the early period after treatment initiation (change in 15-F2t-isoprostane and Krebs cycle intermediates) and was typical for proximal tubular damage
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
200
www.transplantjournal.com
(reduction in Krebs-cycle intermediates and TMAO, and increases of glucose, lactate, acetate and trimethylamine) after 9 or 28 days (13, 14, 20). The hypothesis is that CsA increases glucose and reduces Krebs cycle metabolites in urine as a consequence of defective transmembrane transport of metabolites by tubular proximal cells (13). The results found in our study indicate that similar processes occur in transplanted patients and that CNI-induced nephrotoxicity may be detected at different stages after transplantation. Our results do not allow concluding firmly that the 2 CNIs have different effects on renal metabolism over time. Both CNI have similar nephrotoxic properties, altering renal hemodynamics and structure but also tubular transport and ion homeostasis, translating in hyperuricemia, distal tubular acidosis, hyperkalemia, or magnesium wasting (for review, see (21)). There is evidence from clinical studies that Tac may be less nephrotoxic than CsA, but the exact mechanisms for the difference between the two drugs is not fully understood. Our findings suggest that cyclosporine and tacrolimus may result in different urine metabolomic profile in renal transplant. Only one study has compared the effects of CsA and Tac on the metabolic profile of transplanted patients and was conducted on serum (16). The main findings were that both CNI produce dramatic changes in the serum metabolic profile over time and that levels of metabolites such as glucose, glycerol, hypoxanthine, lactate, pyruvate, succinate, and taurine differed significantly between the CsA and Tac groups. In a previous study using urine profiling after 28 days of treatment with either Tac or CsA in nontransplanted rats, the comparison of metabolites reflecting tubular cells function showed a less pronounced effect of Tac and, particularly, the absence of modification of urinary glucose (13). The authors concluded that Tac affects the kidneys to a lesser extent than CsA. The P-glycoprotein expression in endothelial and renal tubules cells was differentially affected by cyclosporine and Tac in culture cells (22), and one may hypothesize that other transporters are also affected. It has also been suggested that CsA and Tac have different effects on mineral and water transport (23, 24). CsA inhibits the activity of the Na+/K+-ATPase and the Na+/K+/2Cl- cotransporter, whereas Tac stimulates the Na+/K+/2Cl- cotransporter (25). However, ions cannot be analyzed using GC-MS, and therefore, it is not possible to assess the implications of such changes on our own findings. The fact that no significant difference was found between urines of patients treated with either Tac or CsA also indicate that neither these drugs nor their metabolites have interfered on the results. Obviously, the presence of any drug-related metabolite would have created an artificial separation of the metabolic profiles, but this was not the case. Elimination of CsA is primarily biliary with only 6% of the dose excreted in the urine and 0.1% as unchanged drug. The same is true for tacrolimus with less than 1% of the dose administered excreted unchanged in the urine, fecal elimination being the major pathway (http://www.drugbank.ca). Any analysis of potential transplantation-related confounding factors that may have cause changes in metabolomic profiles over time (urinary tract infections, viral infections, or acute rejection) was hampered by the weak number of such events surrounding the urine collection times. Similarly, a number of patients were off-prednisone at M3 and M12 in each group of treatment (Table 1), and some had rapamycin
Transplantation
& Volume 98, Number 2, July 27, 2014
added to Tac (4 patients at M3 and M12). However, although this could have contributed, there were no means to explore the specific impact of changes in the immunosuppressive therapies on urinary metabolites because of the low number of patients. We also showed that in the Tac-group, the metabolic profile differed between patients with IGF as compared with patients with SGF/DGF, indicating that it may be possible to define parameters for the assessment of early lesions in this setting. Inositol and hippurate in urine are such potential candidates. The fact that there was no difference seen between patients with IGF or DGF/SGF in the CsA group could come from the small number of patients (only 9 patients). However, a confounding factor could be that most of them did not receive thymoglobulin as induction therapy, whereas thymoglobulin and basiliximab were equally represented in Tac-treated patients (10 and 13 patients, respectively). However, when comparing patients with IGF and DGF/SGF in the Tac group, we found that a similar proportion of them had received thymoglobulin. Thus, the different metabolite profiles found in patients with IGF or DGF/SGF (Tac group) is a fact not explained by the drug used for induction. After the early posttransplant period, monitoring urinary metabolites representative of tubular injury may confer critical information for patient follow-up. Our study identified lactate, citrate, hippurate, inositol, glucose, and other sugars as metabolites varying over time but, probably in part because of the small number of patients in each CNI group, they did not allow discriminating patients according to eGFR at M3 or M12. Overall, our results confirm that metabolomics is a relevant tool for biomarker development in renal transplantation. GC-MS and 1H-NMR are complementary analytical tools that should probably be used in combination to better characterize the metabolic profiles and explore further the underlying pathophysiologic mechanisms involved at each stage of transplantation.
MATERIALS AND METHODS Patient Selection All consecutive patients receiving a kidney allograft in our department over a 6-month period (July 2009 to February 2010) were included. They received induction therapy with antithymocyte globulin (Thymoglobulin; Amgen, Lyon, France) or interleukin-2 receptor antagonist (basiliximab, Simulect; Novartis, Rueil-Malmaison France) and maintenance immunosuppression with oral mycophenolate mofetil, a calcineurin inhibitor (CNI, either cyclosporine [CsA] or tacrolimus [Tac]), and prednisone. Prednisone at 1 mg/kg per day for the first 2 weeks was then progressively decreased and finally withdrawn within the first year after transplantation in patients with low immunologic risk. Dosage modifications of tacrolimus and cyclosporine were guided by therapeutic drug monitoring.
Collected Parameters We focused analysis on biological and clinical data at day 7 (D7), month 3 (M3), and M12. Graft function on the seventh day after transplantation (D7) was characterized based on serum creatinine value and the need for dialysis. Immediate graft function (IGF) corresponded to a patient whose creatininemia was less than 250 Kmol/L at D7. Patients requiring dialysis within the first week after transplantation were assigned to the delayed graft function (DGF) group, and those whose creatininemia was greater than 250 Kmol/L but who were not dialyzed were assigned to the slow graft function (SGF) group. Renal function at M3 and M12 was estimated (eGFR) using the MDRD formula.
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Dieme et al.
* 2014 Lippincott Williams & Wilkins
Morning urine specimens were systematically collected in fasting state at D7, M3, and M12 after transplantation for routine measurement of proteinuria. The remaining volume was used for the metabolomic study. Urine samples were stored at j20-C until analysis, in compliance with current recommendations to ensure correct conservation (26).
5. 6. 7.
Data Processing The urines were analyzed using a GC-MS method submitted to a series of validation tests to ensure robustness of the results (see SDC, http://links.lww.com/TP/A940). The chromatograms obtained were processed using the freely available XCMS software (27) and normalized according to standard methods in metabolomics (28) (see SDC, http://links.lww.com/TP/A940).
8. 9. 10.
Statistic Analysis Several multivariate models were fitted to explore particular issues. The first ones aimed at comparing the pattern of metabolites between patients of the CsA group and those of the Tac group, at each single time point (D7, M3, and M12). The followings evaluated the variation of urinary composition over time by comparing the metabolite patterns at the three different sampling times (D7, M3, and M12). They were applied separately on urines from patients belonging to each treatment group (CsA or Tac). Finally, to study if particular metabolic signatures may reflect renal function, patients were segregated based on graft function at day 7 (D7) and on eGFR at M3 and M12 after transplantation. These analyses were conducted separately in patients of the CsA or the Tac group. SGF and DGF patients were grouped to compensate for the limited number of patients in each treatment group. The comparison at D7 thus referred to IGF versus SGF or DGF. At M3 and M12, the segregation was made according to the cutoff value of 60 mL/min/1.73 m2, thus defining ‘‘good renal function’’ (Q60 mL/min/1.73 m2) and ‘‘poor renal function’’ (G60 mL/min/1.73 m2) groups. Principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were applied to the data using SIMCA-P software version 13.0 (Umetrics, Umea, Sweden) and R (www.r-project.org) (see SDC for further description of multivariate analysis, http://links.lww.com/TP/A940). Univariate analysis was performed with JMP statistical software version 7.0.2 (SAS Institute, Cary, NC).
Identification of Metabolites Contributing to Discrimination Between Groups
11. 12. 13. 14.
15. 16. 17. 18. 19.
We annotated metabolites from chromatographic peaks based on their MS/MS fragment patterns and retention times. We used the NIST05 (The National Institute of Standards and Technology) library to identify their possible chemical constitutions and ionic structures. The software proposes a list of molecules along with their percentage of matching with standard compounds, indicating the most probable chemical structure. It should be acknowledged that metabolite identification is not always straightforward because of the complexity of biological samples, structural similarities among metabolites, and attenuated spectra resulting from coeluting metabolites.
23.
ACKNOWLEDGMENTS The authors thank Bernadette Pilorge who has contributed to this work with drive and dedication.
24.
REFERENCES
25.
1. 2. 3. 4.
Perkins D, Verma M, Park KJ. Advances of genomic science and systems biology in renal transplantation: a review. Semin Immunopathol 2011; 33: 211. Christians U, Klawitter J, Klawitter J, et al. Biomarkers of immunosuppressant organ toxicity after transplantation: status, concepts and misconceptions. Expert Opin Drug Metab Toxicol 2011; 7: 175. Anglicheau D, Muthukumar T, Hummel A, et al. Discovery and validation of a molecular signature for the noninvasive diagnosis of human renal allograft fibrosis. Transplantation 2012; 93: 1136. Sigdel TK, Klassen RB, Sarwal MM. Interpreting the proteome and peptidome in transplantation. Adv Clin Chem 2009; 47: 139.
20. 21. 22.
26. 27. 28.
201
Hu JD, Tang HQ, Zhang Q, et al. Prediction of gastric cancer metastasis through urinary metabolomic investigation using GC/MS. World J Gastroenterol 2011; 17: 727. Blasco H, Corcia P, Moreau C, et al. 1H-NMR-based metabolomic profiling of CSF in early amyotrophic lateral sclerosis. PLoS One 2010; 5: e13223. Graham SF, Chevallier OP, Roberts D, et al. Investigation of the human brain metabolome to identify potential markers for early diagnosis and therapeutic targets of Alzheimer’s disease. Anal Chem 2013; 85: 1803. Goek ON, Doring A, Gieger C, et al. Serum metabolite concentrations and decreased GFR in the general population. Am J Kidney Dis 2012; 60: 197. Portilla D, Li S, Nagothu KK, et al. Metabolomic study of cisplatininduced nephrotoxicity. Kidney Int 2006; 69: 2194. Serkova N, Fuller TF, Klawitter J, et al. H-NMR-based metabolic signatures of mild and severe ischemia/reperfusion injury in rat kidney transplants. Kidney Int 2005; 67: 1142. Schmitz V, Klawitter J, Bendrick-Peart J, et al. Metabolic profiles in urine reflect nephrotoxicity of sirolimus and cyclosporine following rat kidney transplantation. Nephron Exp Nephrol 2009; 111: e80. Bohra R, Schoning W, Klawitter J, et al. Everolimus and sirolimus in combination with cyclosporine have different effects on renal metabolism in the rat. PLoS One 2012; 7: e48063. Klawitter J, Bendrick-Peart J, Rudolph B, et al. Urine metabolites reflect time-dependent effects of cyclosporine and sirolimus on rat kidney function. Chem Res Toxicol 2009; 22: 118. Lenz EM, Bright J, Knight R, et al. Cyclosporin AYinduced changes in endogenous metabolites in rat urine: a metabonomic investigation using high field 1H NMR spectroscopy, HPLC-TOF/MS and chemometrics. J Pharm Biomed Anal 2004; 35: 599. Klawitter J, Haschke M, Kahle C, et al. Toxicodynamic effects of ciclosporin are reflected by metabolite profiles in the urine of healthy individuals after a single dose. Br J Clin Pharmacol 2010; 70: 241. Kim CD, Kim EY, Yoo H, et al. Metabonomic analysis of serum metabolites in kidney transplant recipients with cyclosporine AY or tacrolimus-based immunosuppression. Transplantation 2010; 90: 748. Wang J, Zhou Y, Xu M, et al. Urinary metabolomics in monitoring acute tubular injury of renal allografts: a preliminary report. Transplant Proc 2011; 43: 3738. Calderisi M, Vivi A, Mlynarz P, et al. Using metabolomics to monitor kidney transplantation patients by means of clustering to spot anomalous patient behavior. Transplant Proc 2013; 45: 1511. Hauet T, Baumert H, Gibelin H, et al. Noninvasive monitoring of citrate, acetate, lactate, and renal medullary osmolyte excretion in urine as biomarkers of exposure to ischemic reperfusion injury. Cryobiology 2000; 41: 280. Bohra R, Klepacki J, Klawitter J, et al. Proteomics and metabolomics in renal transplantation-quo vadis? Transplant International 2013; 26: 225. Naesens M, Kuypers DR, Sarwal M. Calcineurin inhibitor nephrotoxicity. Clin J Am Soc Nephrol 2009; 4: 481. Hauser IA, Koziolek M, Hopfer U, et al. Therapeutic concentrations of cyclosporine A, but not FK506, increase P-glycoprotein expression in endothelial and renal tubule cells. Kidney Int 1998; 54: 1139. Higgins R, Ramaiyan K, Dasgupta T, et al. Hyponatraemia and hyperkalaemia are more frequent in renal transplant recipients treated with tacrolimus than with cyclosporin. Further evidence for differences between cyclosporin and tacrolimus nephrotoxicities. Nephrol Dial Transplant 2004; 19: 444. Watanabe S, Tsuruoka S, Vijayakumar S, et al. Cyclosporin A produces distal renal tubular acidosis by blocking peptidyl prolyl cis-trans isomerase activity of cyclophilin. Am J Physiol Renal Physiol 2005; 288: F40. Aker S, Heering P, Kinne-Saffran E, et al. Different effects of cyclosporine a and FK506 on potassium transport systems in MDCK cells. Exp Nephrol 2001; 9: 332. Bando K, Kawahara R, Kunimatsu T, et al. Influences of biofluid sample collection and handling procedures on GC-MS based metabolomic studies. J Biosci Bioeng 2010; 110: 491. Smith CA, Want EJ, O’Maille G, et al. XCMS: processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Anal Chem 2006; 78: 779. Chan EC, Pasikanti KK, Nicholson JK. Global urinary metabolic profiling procedures using gas chromatography-mass spectrometry. Nat Protoc 2011; 6: 1483.
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
AND
TRANSLATIONAL RESEARCH
Assessing the Metabolic Effects of Calcineurin Inhibitors in Renal Transplant Recipients by Urine Metabolic Profiling Binta Die´me´,1 Jean Michel Halimi,2,3 Patrick Emond,1,4 Matthias Bu¨chler,2,3 Lydie Nadal-Desbarat,1,5 He´le`ne Blasco,1,4 and Chantal Le Guellec3,4,6 Background. Biomarkers that can predict graft function and/or renal side effects of calcineurin inhibitors (CNI) at each stage of treatment in kidney transplantation are still lacking. We report the first untargeted GC-MSYbased metabolomic study on urines of renal transplant patients. This approach would bring insight in biomarkers useable for graft function monitoring. Methods. All consecutive patients receiving a kidney allograft in our transplantation department over a 6-month period were prospectively included and followed up for 12 months. We collected urine samples on the seventh day (D7) after transplantation, then at month 3 (M3) and month 12 (M12), and obtained mass-spectrometryYbased urinary metabolic profiles. Multivariate analyses were conducted to compare metabolic profiles at the 3 different periods and to assess potential differences between cyclosporine and tacrolimus. Differences in metabolic signatures were also assessed according to graft function at D7 and renal function at M3 and M12. Results. The urinary metabolic patterns varied over time in cyclosporine- and tacrolimus-treated patients and were somewhat different at D7, M3, and M12 between the 2 treatment groups. Principal metabolites that differed, regardless of the treatment used, were mainly sugars, inositol, and hippuric acid. Interestingly, among tacrolimus-treated patients, different metabolic signatures were found between patients with immediate or delayed graft function at D7. Conclusion. Urinary metabolomics represents a noninvasive way of monitoring immunosuppressive therapy in renal transplant patients. Although it is too early to consider it as a biomarker of CNI-induced injury or graft function, metabolomics appears a promising evaluation tool in this area. Keywords: Biomarkers, Metabolomics, Kidney transplantation. (Transplantation 2014;98: 195Y201)
alcineurin inhibitor (CNI)Yrelated nephrotoxicity remains a major cause of long-term graft loss in kidney transplantation. Specific biomarkers could be used not only to predict the occurrence of graft injury but also to allow a better mechanistic understanding of the underlying pathologic processes (1). Recently, innovative ‘‘omics’’ approaches have been evaluated, identifying sets of proteins or mRNA associated with clinical events (2Y4). Metabolomics focuses on low-molecular
C
Part of the financial support for chemical analysis has been provided by the Association pour le De´veloppement de la Recherche en Immunologie Clinique (ADRIC). The authors declare no conflicts of interest. 1 INSERM U930, Universite´ Franc¸ois-Rabelais, Tours, France. 2 Service de ne´phrologie-immunologie Clinique, CHU Tours, France. 3 EA 4245, Universite´ Franc¸ois-Rabelais, Tours, France. 4 Laboratoire de biochimie et biologie mole´culaire, CHU Tours, France. 5 PPF ‘‘Analyses des Syste`mes Biologiques’’, Universite´ Franc¸ois Rabelais, Tours, France. 6 Address correspondence to: Chantal Le Guellec, PharmD, Ph.D., Laboratoire de biochimie et biologie mole´culaire, CHU Tours, France. E-mail: [email protected] B.D. performed data and statistical analysis and contributed in writing the paper. J.M.H. designed the research, contributed in writing the paper, and was responsible for collecting the data and patients’ follow-up. P.E.
Transplantation
& Volume 98, Number 2, July27, 2014
weight compounds, allowing analysis of metabolite level changes in biological samples. Surprisingly, although these methods are widely applied to identify biomarkers in as many fields as cancer (5), neurologic diseases (6, 7), renal diseases (8), or druginduced nephrotoxicity (9), only few metabolomic studies have been conducted in renal transplantation. Some studies conducted in rats identified biomarkers of ischemia/reperfusion injury (10) or immunosuppressant-induced nephrotoxicity in transplanted (11) or nontransplanted (12Y14) animals. Conversely, studies in human are very scanty. We found 2 study participated in data analysis and in writing the article. M.B. participated in collecting the data and in the patients’ follow-up. L.N.D. participated in data analysis and writing the article. H.B.R. participated in research design, data analysis, and preparing the article. C.L.G. designed the research, participated in analyzing the data, and writing the paper. Supplemental digital content (SDC) is available for this article. Direct URL citations appear in the printed text, and links to the digital files are provided in the HTML text of this article on the journal’s Web site (www.transplantjournal.com). Received 28 October 2013. Revision requested 19 November 2013. Accepted 23 December 2013. Copyright * 2014 by Lippincott Williams & Wilkins ISSN: 0041-1337/14/9802-195 DOI: 10.1097/TP.0000000000000039
www.transplantjournal.com
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
195
196
www.transplantjournal.com
evaluating either short-term toxicodynamic effects of cyclosporine (CsA) on serum and urinary metabolite profiles in healthy volunteers (15) or time-dependent changes of serum metabolite profiles in patients treated with a CNI (16) and only a small study in five transplanted patients relating urinary metabolome to acute tubular injury (17). Very recently, Calderisi et al. used 1H-NMR (proton nuclear magnetic resonnance) to profile urines of kidney transplant patients during the 2 weeks after transplantation. We used gas chromatographyYmass spectrometry (GC-MS) to profile urine samples of kidney transplant patients receiving a triple-drug immunosuppressive regimen.
TABLE 1.
Transplantation
& Volume 98, Number 2, July 27, 2014
Using an untargeted approach, we analyzed whether the metabolic profile differed according to the CNI used (tacrolimus (Tac) or CsA) and, under a given treatment, if it varied over time during the first 12 months after grafting. In an attempt to derive potential biomarkers of kidney function, metabolic profiles were compared according to renal function at Day 7 but also at Month 3 (M3) and Month 12 (M12).
RESULTS Demographic and Clinical Characteristics The characteristics of the patients at the time of inclusion and during follow-up are given in (Table 1).Thirty-five patients
Patient characteristics
Characteristics Baseline characteristics of all patients Male/female Age median (range), years Weight median (range), kg Primary/secondary kidney transplant, number Type of donnor (cadaveric/living), number Cold ischemia duration, median (range), minutes Induction therapy anti-lymphocyte / basiliximab, number Acute rejection during the 12 months of follow-up , number (delay in days (D)) Characteristics of patients analyzed at D7 Creatininemia median (range), Kmol/L Creatininuria median (range), mmol/24 h Proteinuria median (range), g/24 h IGF, number of patients (%) SGF, number of patients (%) DGF, number of patients (%) UTI, number of patients (%) Viral infections, number of patients (%) Characteristics of patients analyzed at M3 Creatininemia median (range), Kmol/L Creatininuria median (range), mmol/24 h Proteinuria median (range), g/24 h eGFR median (range), mL/min/1.73 m2 eGFRQ60 mL/min/1.73 m2, number of patients (%) UTI, number of patients (%) Viral infections, number of patients (%) Diabetes mellitus, number of patients (%) Patients off-prednisone, number (%) Characteristics of patients analyzed at M12 Creatininemia median (range), Kmol/L Creatininuria median (range), mmol/24 h Proteinuria median (range), g/24 h eGFR median (range), mL/min/1.73 m2 eGFRQ60 mL/min/1.73 m2, number of patients (%) UTI, number of patients (%) Viral infections, number of patients (%) Diabetes mellitus, number of patients (%) Patients off-prednisone, number (%)
Tac group
CsA group
n=23 15/8 59 (16Y74) 73 (40Y105) 19/4 23/0 1086 (520Y1926) 10/13 1 (D10) n=18 195 (92Y1490) 10.4 (4.8Y18.9) 0.62 (0.22Y4.2) 10 (55%) 7 (39.5%) 1 (5.5%) 0 0 n=19 137 (78Y369) 11.1 (3.1Y29) 0.29 (0Y6.4) 54 (17.5Y105) 5 (26%) 0 0 0 1 (5.2%) n=19 129 (89Y356) 13 (6.5Y42) 0.26 (0Y1.28) 48 (14.7Y115) 6 (31.5%) 0 0 1 (5.2%) 11 (57%)
n=12 6/6 50 (20Y67) 75 (59Y111) 11/1 12/0 928 (731Y1960) 1/11 4 (D53, D69, D95, D140) n=9 146 (80Y464) 9 (6.2Y13.8) 0.84 (0.24Y2.9) 6 (67%) 2 (22%) 1 (10%) 0 0 n=9 133 (77Y206) 10.4 (2.4Y18.8) 0.16 (0Y0.95) 55 (48.3Y106) 5 (56%) 1 (11%) 0 0 0 n=10 109 (82Y171) 13.7 (7.1Y41) 0.13 (0Y1.22) 70 (63Y118) 10 (100%) 0 1 (10%) 3 (30%)
IGF, immediate graft function; SGF, slow graft function; DGF, delayed graft function; eGFR, estimated glomerular filtration rate; UTI, urinary tract infection; Tac, tacrolimus; CsA cyclosporine.
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
* 2014 Lippincott Williams & Wilkins
were included in the study but, as some urinary samples were missing, each of them was not equally represented at each sampling time. Antilymphocyte globulins and basiliximab were equally balanced as induction therapy in the Tac group, but basiliximab was the most frequently used in patients of the CsA group (92%). One patient per group required dialysis after transplantation (delayed graft function [DGF]), but the majority of patients had a good initial renal function. Renal function improved gradually over time in both groups. A higher percentage of patients had recovered a good renal function (eGFRQ60 mL/min/1.73 m2) in the CsA group as compared with the Tac group at M3 (56% versus 26%), but the difference was not significant (P=0.407). In contrast, the difference (100% versus 31.5%) was significant at M12 (P=0.0004). Some clinical events occurred during follow-up (urinary tract infections, viral infections, acute rejection, and diabetes mellitus). Those occurring surrounding sample collection times are listed in Table 1 but the weak number of events precluded analysis of their potential impact on metabolomic profiles. Metabolite Profiling Analysis Variation Over Time Orthogonal partial least-squares discriminant analysis (OPLS-DA) scores scatter plots showed that samples drawn at the different time-points were well segregated into different regions, regardless of the CNI used, CsA or Tac (Fig. 1A and B). Urines collected in patients treated with tacrolimus showed a clear clustering when comparing D7 versus M3 (Fig. 2A) and D7 versus M12 (Fig. 2B). In contrast, the model did not differentiate well the samples obtained at M3 and M12. This indicates that metabolic profiles significantly change between D7 and M3 and between D7 and M12. The set of metabolites discriminating between D7 and M3 and between D7 and M12 are shown in Table S1, SDC, http://links.lww.com/TP/A940. These are mainly sugars, inositol, hippuric acid, and an unidentified metabolite (M100T444). Univariate analysis revealed that urines at D7
Dieme et al.
197
contained significantly more inositol than that of the later periods and that hippuric acid and some unidentified metabolites were over represented at M12. Similar results were obtained in patients treated with CsA (Fig. 2C and D). Inositol and M100T444 were more abundant at D7, whereas lactate, citrate, and hippurate predominated at M12 (Table S2, SDC, http://links.lww.com/TP/A940). Variation of the Metabolic Pattern With the CNI Used Visual group separation was also observed in OPLS-DA scores scatter plots obtained at each single sampling time between patients treated by CsA or by Tac. However, model parameters were not excellent, and we were not able to obtain clear separation between CNI groups, particularly at D7 and M12 (low R2X and Q2 values and presence of outliers), although the discrimination was quite better at M3 (Figure S1, SDC, http://links.lww.com/TP/A940). The 3 models included a huge number of discriminative metabolites, and we did not try to identify them because of the poor model performances. Metabolite Pattern and Renal Function A clear separation between patients with immediate graft function (IGF, n=10) versus slow graft function (SGF or DGF, n=8) was obtained in the Tac group (Fig. 3). A similar proportion of patients received thymoglobulin induction in each group, that is, 3/10 (30%) in the IGF group versus 3/8 (37%) in the SGF/DGF group. The metabolites responsible for the separation between groups were inositol, hippurate, D-furanose, and other unidentified metabolites (Fig. 4 and Table S3, SDC, http://links.lww.com/TP/A940). Inositol and hippurate are more important in urine of patients with SGF or DGF than in IGF group. No such difference was obtained in the nine patients treated with CsA exhibiting either IGF (n=6) or SGF/DGF (n=3). Multivariate analysis failed to find any difference in the metabolite profiles between patients with ‘‘good’’ and ‘‘poor’’ renal function at M3, regardless of the CNI used. At M12, no difference was found in the Tac group, and as all patients
FIGURE 1. Orthogonal partial least squares discriminant analysis (OPLS-DA) score plot of urine samples obtained in (A) the Tac group at 7 days (medium gray), 3 months (light gray), and 12 months (black) after kidney transplantation and in (B) the CsA group at 7 days (medium gray), 3 months (light gray) and 12 months (black). For explanations about the OPLS-DA method, the reader is referred to the SDC, http://links.lww.com/TP/A940. Each point stands for the projection of a single sample on the main principal axes, corresponding to principal components, named either t[1], t[2], t0[1], or Num (in descending order of importance on projection), showing clear discrimination between samples drawn at D7, M3, or M12, whatever the CNI considered. The 3D orientation of each figure is the one enabling the best vision of the clusters.
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
198
www.transplantjournal.com
Transplantation
& Volume 98, Number 2, July 27, 2014
FIGURE 2. Pairwise comparison of time-dependent changes by OPLS-DA score plots, allowing visual inspection of discrimination between groups. Each point stands for the projection of a single sample on the main principal axes, corresponding to principal components, named either t[1] or t0[1] (in descending order of importance on projection). Also are given objective criteria for judging the quality of the separation: R2X and Q2 (see SDC, http://links.lww.com/TP/A940 for explanations of the meanings of these parameters). A, urines of D7 versus urines of M3 in patients treated with Tac (R2X=0.67; Q2=0.705). B, urines of D7 versus urines of M12 in patients treated with Tac (R2X=0.507; Q2=0.727). C, urines of D7 versus urines of M3 in patients treated with CsA (R2X=0.66; Q2=0.71). D, urines of D7 versus urines of M12 in patients treated with CsA (R2X=0.754; Q2=0.85).
of the CsA group had a ‘‘good’’ renal function, no comparison was performed.
DISCUSSION The present study shows that there is a change of urinary biochemistry over time and that this may correlate with graft function in tacrolimus-treated patients. One of our principal observations is that the metabolite profile at D7 is different from that at M3 and M12 regardless of the CNI used. After transplantation, the timeframe of kidney injury typically involves ischemia-reperfusion injury, followed by changes in protein expression and histologic modifications.
CNI-related nephrotoxicity, inflammation, and fibrosis are also responsible for histologic changes and alteration of kidney function over time (2). The time-dependent changes in urinary metabolite pattern found in our study indicate that this approach has the potential to reflect changes in kidney state during the first year after transplantation. We identified pentoses and hexoses as the metabolites that discriminate between D7 and the later periods, whatever the CNI used. However, precise identification of the true structure of these sugars was not achieved. Aside from sugars, inositol was more frequent at D7 (Tac and CsA groups), hippuric acid more frequent at M3 (Tac group) and M12 (Tac and CsA groups), and lactate and
FIGURE 3. OPLS-DA score plot for the comparison of metabolic profiles in patients with immediate graft function (IGF, black dots) or either slow or delayed graft function (SGF/DGF, gray dots) in the subgroup of patients treated with Tac. (R2X=0.539; Q2=0.874)
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
* 2014 Lippincott Williams & Wilkins
Dieme et al.
199
FIGURE 4. Box plots of univariate comparison of metabolite amounts (referred as area) between patients with immediate graft function (IGF) and either slow or delayed graft function (SGF/DGF) in the subgroup of patients treated with Tac. *PG0.05/n, with n is the number of metabolites tested (n=6 in this case) and considered for correction for multivariate analysis.
citrate more frequent at M12 (CsA group). Interestingly, a recent paper also identified time-dependent changes in 1H-NMR metabolomic profiles during the first 2 weeks after transplantation, which were representative of two stages of kidney recovery, after an initial postoperative stage (18). In this study, the metabolites that discriminated between early (postoperative) and later periods (2 weeks of follow-up) were mainly lipoproteins, creatine, and creatinine and, to a lesser extent, mannitol, acetate, and hippurate. From a mechanistic point of view, the urinary metabolic profile at D7 has the potential to reflect ischemia-reperfusion injury. Serkova et al. described the shift of the metabolic composition of the kidneys after cold storage alone and cold storage followed by transplantation in rats using 1H-RMN (10). Native renal tissue contained carbohydrates (glycogen and glucose) and triglycerides but also high concentrations of cellular osmolytes such as inositol, TMAO, and taurine as well as intermediates of the Krebs cycle (citrate), amino acids (glutamate, alanine, glutamine, and aspartate) and end products of glycolysis (lactate). After cold storage, there was a significant increase in glycogen and other carbohydrates as well as lactate in kidney extracts. They also showed a dramatic increase of allantoin, the end-product of xanthine metabolism and of uric acid, a marker of oxidative stress, in kidneys exposed to reperfusion. Similarly, Hauet et al. performed a 1H-NMR analysis of selected metabolites in an animal model of autotransplantation, the kidneys being cold-preserved for 48 h at 4-C (19). They showed an increase of biochemical markers of renal medulla damage (trimethylamine-N-oxide, TMAO) and
mitchondrial dysfunction (citrate, lactate, and acetate) across the 14 days after transplantation. Although our results are not directly comparable to other studies because of the different analytical methods used (GC-MS as compared with 1H-MNR), some similarities are observed in that carbohydrates, lactate, and citrate were also shown discriminant in our patients at D7. This indicates that stigmas associated to ischemia/ reperfusion may be detected by this approach in urines of transplanted patients. In addition to ischemia/reperfusion injury, the urinary composition at D7 may also be dependent on the additional nephrotoxicity of CNI. Urinary metabolite patterns were previously studied in transplanted rats receiving either CsA or sirolimus over a period of 7 days (11). By using an orthotopic transplant model, the authors avoided interference from immunologic processes but were able to analyze changes because of surgical procedure, ischemia/reperfusion, or acute drug nephrotoxicity. Contrary to the above studies, no difference in metabolic pattern was seen at day 7 between transplanted and nontransplanted rats who received no treatment, allowing the authors to conclude that the metabolic changes that they observed in treated animals (variable according to the immunosuppressant used) strictly arose from drug nephrotoxicity. Further experimental studies evaluating the time-dependent effects of CsA on kidney metabolism in nontransplanted rats showed that the 1H-NMR biochemical signature reflects mitochondrial dysfunction in the early period after treatment initiation (change in 15-F2t-isoprostane and Krebs cycle intermediates) and was typical for proximal tubular damage
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
200
www.transplantjournal.com
(reduction in Krebs-cycle intermediates and TMAO, and increases of glucose, lactate, acetate and trimethylamine) after 9 or 28 days (13, 14, 20). The hypothesis is that CsA increases glucose and reduces Krebs cycle metabolites in urine as a consequence of defective transmembrane transport of metabolites by tubular proximal cells (13). The results found in our study indicate that similar processes occur in transplanted patients and that CNI-induced nephrotoxicity may be detected at different stages after transplantation. Our results do not allow concluding firmly that the 2 CNIs have different effects on renal metabolism over time. Both CNI have similar nephrotoxic properties, altering renal hemodynamics and structure but also tubular transport and ion homeostasis, translating in hyperuricemia, distal tubular acidosis, hyperkalemia, or magnesium wasting (for review, see (21)). There is evidence from clinical studies that Tac may be less nephrotoxic than CsA, but the exact mechanisms for the difference between the two drugs is not fully understood. Our findings suggest that cyclosporine and tacrolimus may result in different urine metabolomic profile in renal transplant. Only one study has compared the effects of CsA and Tac on the metabolic profile of transplanted patients and was conducted on serum (16). The main findings were that both CNI produce dramatic changes in the serum metabolic profile over time and that levels of metabolites such as glucose, glycerol, hypoxanthine, lactate, pyruvate, succinate, and taurine differed significantly between the CsA and Tac groups. In a previous study using urine profiling after 28 days of treatment with either Tac or CsA in nontransplanted rats, the comparison of metabolites reflecting tubular cells function showed a less pronounced effect of Tac and, particularly, the absence of modification of urinary glucose (13). The authors concluded that Tac affects the kidneys to a lesser extent than CsA. The P-glycoprotein expression in endothelial and renal tubules cells was differentially affected by cyclosporine and Tac in culture cells (22), and one may hypothesize that other transporters are also affected. It has also been suggested that CsA and Tac have different effects on mineral and water transport (23, 24). CsA inhibits the activity of the Na+/K+-ATPase and the Na+/K+/2Cl- cotransporter, whereas Tac stimulates the Na+/K+/2Cl- cotransporter (25). However, ions cannot be analyzed using GC-MS, and therefore, it is not possible to assess the implications of such changes on our own findings. The fact that no significant difference was found between urines of patients treated with either Tac or CsA also indicate that neither these drugs nor their metabolites have interfered on the results. Obviously, the presence of any drug-related metabolite would have created an artificial separation of the metabolic profiles, but this was not the case. Elimination of CsA is primarily biliary with only 6% of the dose excreted in the urine and 0.1% as unchanged drug. The same is true for tacrolimus with less than 1% of the dose administered excreted unchanged in the urine, fecal elimination being the major pathway (http://www.drugbank.ca). Any analysis of potential transplantation-related confounding factors that may have cause changes in metabolomic profiles over time (urinary tract infections, viral infections, or acute rejection) was hampered by the weak number of such events surrounding the urine collection times. Similarly, a number of patients were off-prednisone at M3 and M12 in each group of treatment (Table 1), and some had rapamycin
Transplantation
& Volume 98, Number 2, July 27, 2014
added to Tac (4 patients at M3 and M12). However, although this could have contributed, there were no means to explore the specific impact of changes in the immunosuppressive therapies on urinary metabolites because of the low number of patients. We also showed that in the Tac-group, the metabolic profile differed between patients with IGF as compared with patients with SGF/DGF, indicating that it may be possible to define parameters for the assessment of early lesions in this setting. Inositol and hippurate in urine are such potential candidates. The fact that there was no difference seen between patients with IGF or DGF/SGF in the CsA group could come from the small number of patients (only 9 patients). However, a confounding factor could be that most of them did not receive thymoglobulin as induction therapy, whereas thymoglobulin and basiliximab were equally represented in Tac-treated patients (10 and 13 patients, respectively). However, when comparing patients with IGF and DGF/SGF in the Tac group, we found that a similar proportion of them had received thymoglobulin. Thus, the different metabolite profiles found in patients with IGF or DGF/SGF (Tac group) is a fact not explained by the drug used for induction. After the early posttransplant period, monitoring urinary metabolites representative of tubular injury may confer critical information for patient follow-up. Our study identified lactate, citrate, hippurate, inositol, glucose, and other sugars as metabolites varying over time but, probably in part because of the small number of patients in each CNI group, they did not allow discriminating patients according to eGFR at M3 or M12. Overall, our results confirm that metabolomics is a relevant tool for biomarker development in renal transplantation. GC-MS and 1H-NMR are complementary analytical tools that should probably be used in combination to better characterize the metabolic profiles and explore further the underlying pathophysiologic mechanisms involved at each stage of transplantation.
MATERIALS AND METHODS Patient Selection All consecutive patients receiving a kidney allograft in our department over a 6-month period (July 2009 to February 2010) were included. They received induction therapy with antithymocyte globulin (Thymoglobulin; Amgen, Lyon, France) or interleukin-2 receptor antagonist (basiliximab, Simulect; Novartis, Rueil-Malmaison France) and maintenance immunosuppression with oral mycophenolate mofetil, a calcineurin inhibitor (CNI, either cyclosporine [CsA] or tacrolimus [Tac]), and prednisone. Prednisone at 1 mg/kg per day for the first 2 weeks was then progressively decreased and finally withdrawn within the first year after transplantation in patients with low immunologic risk. Dosage modifications of tacrolimus and cyclosporine were guided by therapeutic drug monitoring.
Collected Parameters We focused analysis on biological and clinical data at day 7 (D7), month 3 (M3), and M12. Graft function on the seventh day after transplantation (D7) was characterized based on serum creatinine value and the need for dialysis. Immediate graft function (IGF) corresponded to a patient whose creatininemia was less than 250 Kmol/L at D7. Patients requiring dialysis within the first week after transplantation were assigned to the delayed graft function (DGF) group, and those whose creatininemia was greater than 250 Kmol/L but who were not dialyzed were assigned to the slow graft function (SGF) group. Renal function at M3 and M12 was estimated (eGFR) using the MDRD formula.
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Dieme et al.
* 2014 Lippincott Williams & Wilkins
Morning urine specimens were systematically collected in fasting state at D7, M3, and M12 after transplantation for routine measurement of proteinuria. The remaining volume was used for the metabolomic study. Urine samples were stored at j20-C until analysis, in compliance with current recommendations to ensure correct conservation (26).
5. 6. 7.
Data Processing The urines were analyzed using a GC-MS method submitted to a series of validation tests to ensure robustness of the results (see SDC, http://links.lww.com/TP/A940). The chromatograms obtained were processed using the freely available XCMS software (27) and normalized according to standard methods in metabolomics (28) (see SDC, http://links.lww.com/TP/A940).
8. 9. 10.
Statistic Analysis Several multivariate models were fitted to explore particular issues. The first ones aimed at comparing the pattern of metabolites between patients of the CsA group and those of the Tac group, at each single time point (D7, M3, and M12). The followings evaluated the variation of urinary composition over time by comparing the metabolite patterns at the three different sampling times (D7, M3, and M12). They were applied separately on urines from patients belonging to each treatment group (CsA or Tac). Finally, to study if particular metabolic signatures may reflect renal function, patients were segregated based on graft function at day 7 (D7) and on eGFR at M3 and M12 after transplantation. These analyses were conducted separately in patients of the CsA or the Tac group. SGF and DGF patients were grouped to compensate for the limited number of patients in each treatment group. The comparison at D7 thus referred to IGF versus SGF or DGF. At M3 and M12, the segregation was made according to the cutoff value of 60 mL/min/1.73 m2, thus defining ‘‘good renal function’’ (Q60 mL/min/1.73 m2) and ‘‘poor renal function’’ (G60 mL/min/1.73 m2) groups. Principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were applied to the data using SIMCA-P software version 13.0 (Umetrics, Umea, Sweden) and R (www.r-project.org) (see SDC for further description of multivariate analysis, http://links.lww.com/TP/A940). Univariate analysis was performed with JMP statistical software version 7.0.2 (SAS Institute, Cary, NC).
Identification of Metabolites Contributing to Discrimination Between Groups
11. 12. 13. 14.
15. 16. 17. 18. 19.
We annotated metabolites from chromatographic peaks based on their MS/MS fragment patterns and retention times. We used the NIST05 (The National Institute of Standards and Technology) library to identify their possible chemical constitutions and ionic structures. The software proposes a list of molecules along with their percentage of matching with standard compounds, indicating the most probable chemical structure. It should be acknowledged that metabolite identification is not always straightforward because of the complexity of biological samples, structural similarities among metabolites, and attenuated spectra resulting from coeluting metabolites.
23.
ACKNOWLEDGMENTS The authors thank Bernadette Pilorge who has contributed to this work with drive and dedication.
24.
REFERENCES
25.
1. 2. 3. 4.
Perkins D, Verma M, Park KJ. Advances of genomic science and systems biology in renal transplantation: a review. Semin Immunopathol 2011; 33: 211. Christians U, Klawitter J, Klawitter J, et al. Biomarkers of immunosuppressant organ toxicity after transplantation: status, concepts and misconceptions. Expert Opin Drug Metab Toxicol 2011; 7: 175. Anglicheau D, Muthukumar T, Hummel A, et al. Discovery and validation of a molecular signature for the noninvasive diagnosis of human renal allograft fibrosis. Transplantation 2012; 93: 1136. Sigdel TK, Klassen RB, Sarwal MM. Interpreting the proteome and peptidome in transplantation. Adv Clin Chem 2009; 47: 139.
20. 21. 22.
26. 27. 28.
201
Hu JD, Tang HQ, Zhang Q, et al. Prediction of gastric cancer metastasis through urinary metabolomic investigation using GC/MS. World J Gastroenterol 2011; 17: 727. Blasco H, Corcia P, Moreau C, et al. 1H-NMR-based metabolomic profiling of CSF in early amyotrophic lateral sclerosis. PLoS One 2010; 5: e13223. Graham SF, Chevallier OP, Roberts D, et al. Investigation of the human brain metabolome to identify potential markers for early diagnosis and therapeutic targets of Alzheimer’s disease. Anal Chem 2013; 85: 1803. Goek ON, Doring A, Gieger C, et al. Serum metabolite concentrations and decreased GFR in the general population. Am J Kidney Dis 2012; 60: 197. Portilla D, Li S, Nagothu KK, et al. Metabolomic study of cisplatininduced nephrotoxicity. Kidney Int 2006; 69: 2194. Serkova N, Fuller TF, Klawitter J, et al. H-NMR-based metabolic signatures of mild and severe ischemia/reperfusion injury in rat kidney transplants. Kidney Int 2005; 67: 1142. Schmitz V, Klawitter J, Bendrick-Peart J, et al. Metabolic profiles in urine reflect nephrotoxicity of sirolimus and cyclosporine following rat kidney transplantation. Nephron Exp Nephrol 2009; 111: e80. Bohra R, Schoning W, Klawitter J, et al. Everolimus and sirolimus in combination with cyclosporine have different effects on renal metabolism in the rat. PLoS One 2012; 7: e48063. Klawitter J, Bendrick-Peart J, Rudolph B, et al. Urine metabolites reflect time-dependent effects of cyclosporine and sirolimus on rat kidney function. Chem Res Toxicol 2009; 22: 118. Lenz EM, Bright J, Knight R, et al. Cyclosporin AYinduced changes in endogenous metabolites in rat urine: a metabonomic investigation using high field 1H NMR spectroscopy, HPLC-TOF/MS and chemometrics. J Pharm Biomed Anal 2004; 35: 599. Klawitter J, Haschke M, Kahle C, et al. Toxicodynamic effects of ciclosporin are reflected by metabolite profiles in the urine of healthy individuals after a single dose. Br J Clin Pharmacol 2010; 70: 241. Kim CD, Kim EY, Yoo H, et al. Metabonomic analysis of serum metabolites in kidney transplant recipients with cyclosporine AY or tacrolimus-based immunosuppression. Transplantation 2010; 90: 748. Wang J, Zhou Y, Xu M, et al. Urinary metabolomics in monitoring acute tubular injury of renal allografts: a preliminary report. Transplant Proc 2011; 43: 3738. Calderisi M, Vivi A, Mlynarz P, et al. Using metabolomics to monitor kidney transplantation patients by means of clustering to spot anomalous patient behavior. Transplant Proc 2013; 45: 1511. Hauet T, Baumert H, Gibelin H, et al. Noninvasive monitoring of citrate, acetate, lactate, and renal medullary osmolyte excretion in urine as biomarkers of exposure to ischemic reperfusion injury. Cryobiology 2000; 41: 280. Bohra R, Klepacki J, Klawitter J, et al. Proteomics and metabolomics in renal transplantation-quo vadis? Transplant International 2013; 26: 225. Naesens M, Kuypers DR, Sarwal M. Calcineurin inhibitor nephrotoxicity. Clin J Am Soc Nephrol 2009; 4: 481. Hauser IA, Koziolek M, Hopfer U, et al. Therapeutic concentrations of cyclosporine A, but not FK506, increase P-glycoprotein expression in endothelial and renal tubule cells. Kidney Int 1998; 54: 1139. Higgins R, Ramaiyan K, Dasgupta T, et al. Hyponatraemia and hyperkalaemia are more frequent in renal transplant recipients treated with tacrolimus than with cyclosporin. Further evidence for differences between cyclosporin and tacrolimus nephrotoxicities. Nephrol Dial Transplant 2004; 19: 444. Watanabe S, Tsuruoka S, Vijayakumar S, et al. Cyclosporin A produces distal renal tubular acidosis by blocking peptidyl prolyl cis-trans isomerase activity of cyclophilin. Am J Physiol Renal Physiol 2005; 288: F40. Aker S, Heering P, Kinne-Saffran E, et al. Different effects of cyclosporine a and FK506 on potassium transport systems in MDCK cells. Exp Nephrol 2001; 9: 332. Bando K, Kawahara R, Kunimatsu T, et al. Influences of biofluid sample collection and handling procedures on GC-MS based metabolomic studies. J Biosci Bioeng 2010; 110: 491. Smith CA, Want EJ, O’Maille G, et al. XCMS: processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Anal Chem 2006; 78: 779. Chan EC, Pasikanti KK, Nicholson JK. Global urinary metabolic profiling procedures using gas chromatography-mass spectrometry. Nat Protoc 2011; 6: 1483.
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
