Journal Logo

Original Clinical Science

Metabolomic Analysis of Perfusate During Hypothermic Machine Perfusion of Human Cadaveric Kidneys

Guy, Alison J.1; Nath, Jay1; Cobbold, Mark2; Ludwig, Christian3; Tennant, Daniel A.4; Inston, Nicholas G.1; Ready, Andrew R.1

Author Information
doi: 10.1097/TP.0000000000000398

Hypothermic machine perfusion (HMP) is increasingly used to preserve cadaveric kidneys in the period before transplantation.1 Studies have demonstrated reduced rates of delayed graft function (DGF) and improved graft survival in machine-perfused kidneys compared to those preserved in traditional static cold storage (SCS).2-5 Hypothermic machine perfusion also provides a unique opportunity for assessment of cadaveric kidneys during storage with easy and safe access to perfusate samples, kidney tissue and information regarding flow dynamics.

Accurate evaluation of allograft quality is essential to prevent unnecessary kidney discard, allow maximized donor-recipient matching and to plan appropriate perioperative care. Viability assessment is of increasing importance in an era where marginal kidneys, with a greater risk of poor graft function, are increasingly being used in an attempt to address the organ shortage.6-9 Despite considerable interest in various indicators and markers of graft quality, few are used routinely in clinical practice, and no single biomarker or parameter has proven to be definitive.

Although the clinical benefits of HMP are now well documented, the exact processes by which this occurs remain unclear. Although initially the mechanism of action was thought to be the maintenance of vascular bed patency, there is increasing evidence that substantial metabolic activity persists, and this may also have a role.10 However, the metabolic activity in the ex vivo, hypoxic, hypothermic environment provided by HMP is poorly understood.

Proton-nuclear magnetic resonance (1H-NMR) spectroscopy, or metabolomics, can be used to analyze fluids or tissues and can simultaneously detect, identify, and quantify hundreds of metabolites—some of which may have prognostic or therapeutic value.11 Although several publications have demonstrated the feasibility of NMR for evaluating different clinical scenarios in transplantation, no previous studies, to our knowledge, have examined the perfusate of human cadaveric kidneys during HMP using NMR.11-14

The aim of this study was to use NMR spectroscopy to examine the metabolic profile of the perfusate of human cadaveric kidneys for transplantation and to identify possible discriminators between the profiles of kidneys with delayed and immediate graft function (IGF).


Twenty-nine kidneys were included in the study. Three kidneys were rejected for implantation after HMP. Reasons included severe atherosclerosis of the renal artery with an adverse donor history, a hypoplastic cystic kidney, and inadequate initial flush of the kidney at recovery. No kidneys were rejected on HMP parameters. Transplantation proceeded in the remaining 26 kidneys. Donor and recipient characteristics, along with HMP parameters, are shown in SDC1, There were no statistically significant differences in characteristics or parameters between the IGF and DGF kidneys.

Twenty-eight metabolites were detected in the perfusate of these kidneys at both 45 min and 4 hr of HMP. Of these metabolites, six were recognized constituents of the KPS-1 fluid (SDC2, adenine, gluconate, glucose, glutathione (reduced form), mannitol, and ribose. Median concentrations of these metabolites measured by 1H-NMR at each timepoint, and the P values for change over time are shown in Table 1. There was a significant change in concentration of glucose and glutathione between the two timepoints.

Metabolite concentrations (mM) for constituents of KPS-1 identified in kidney perfusate

The remaining 22 metabolites are not listed as constituents of KPS-1. Median concentrations of these metabolites at each timepoint, and P values for change over time are shown in Table 2. The majority of these metabolites changed significantly.

Metabolite concentrations (mM) measured in kidney perfusate not listed as constituents of KPS-1

Of the 26 transplanted kidneys after HMP, 19 (73%) kidneys displayed IGF posttransplant, and seven (27%) suffered from DGF. There were differences between the metabolomic profile of these groups—metabolites that were significantly different at one or both timepoints were glucose, inosine, leucine, and gluconate.

Glucose concentrations were significantly lower in DGF kidneys compared to those with IGF at both 45 min (7.772 vs. 9.459 mM, P = 0.006) and 4 hr (8.202 vs. 10.235 mM, P = 0.003) (Figure 1A). Receiver operating characteristic (ROC) curves assessing the predictive accuracy of glucose for DGF yielded an area under the ROC (AUROC) curve of 0.842 (standard error [SE], 0.080) at 45 min and 0.895 (SE, 0.069) at 4 hr (Figure 2).

Box and Whisker plots representing the significantly different metabolites between the kidney perfusate of DGF and IGF kidneys. DGF, delayed graft function; IGF, immediate graft function.
Receiver-operator Characteristic (ROC) Curves and Areas Under the ROC Curves (AUROCs) for Significant Metabolites.

Concentrations of inosine and leucine were significantly different between DGF and IGF kidneys at 45 min (0.002 vs. 0.013 mM, P = 0.009 and 0.011 vs. 0.006 mM, P = 0.036) (Figure 1B and C) but not at 4 hr. The AUROC for inosine at 45 min was 0.833 (SE, 0.082) and for leucine at 45 min was 0.732 (SE, 0.135) (Figure 2).

Gluconate levels were also significantly different between DGF and IGF kidneys at 4 hr (49.099 vs. 59.513 mM, P = 0.009) (Figure 1D) but not at 45 min. The AUROC for gluconate at 4 hr was 0.851 (SE, 0.089) (Figure 2). Cutoff points for ROC curves are shown in SDC3 (


To our knowledge, this is the first report to detail the metabolomic profile of perfusate during HMP of human cadaveric kidneys. Changes in the perfusate composition during this time may represent substances being removed by the kidney to supply ongoing cell processes or products of metabolism or degradation being released from the kidney. This study demonstrates that after only 45 min of machine perfusion, the perfusate is markedly different from the original preservation solution. Furthermore, this study has identified differences in the metabolomic profile of IGF and DGF kidney perfusate that may allow prediction of functional graft outcome after transplantation.

Accurate assessment of graft quality is increasingly important to achieve the highest levels of success in transplantation. Although donor information, such as patient age, comorbidity, or terminal serum creatinine, along with kidney biopsy data are important, they still have a limited capacity for accurate prediction of graft outcomes.15,16 Machine perfusion parameters, such as resistance, are also accepted as good indicators of graft quality, but several studies have warned of the dangers of using resistance values to determine kidney discard.17-19

Theoretically, biomarkers measured in urine and perfusate have an advantage over biopsy data of being noninvasive and can be measured frequently and objectively. A recent review highlighted the biomarkers that have been assessed in regards to graft outcome.20 Few studies were recent, and even fewer deemed to be of good quality. Levels of lactate dehydrogenase, glutathione-S-transferase and aspartate transaminase were significantly associated with DGF in the greatest number of studies, but further validation was recommended.

The NMR-based metabolomics is a novel approach for rapidly identifying the changes in global metabolite profiles of biologic samples and is widely used in other disease entities.21 Several studies have demonstrated possibilities for the use of NMR-based metabolomics within transplantation; it has shown promise as a tool to predict long-term graft outcome based on the energy state of the kidney in vivo, and it may be able to assess markers of kidney injury ex vivo.22-24 More recently, NMR has been used to examine HMP perfusate in a preclinical donation after cardiac death porcine model.14 This study concluded that analysis of biomarkers during HMP using NMR could be an interesting tool to assess graft quality and was compatible with clinical application.

In this study, the main constituents of KPS-1 were identified by 1H-NMR, except for HEPES and hydroxyethyl starch which the software cannot recognize. Of the identified metabolites, median levels of gluconate, mannitol, adenine, and ribose did not change significantly over HMP time for all perfused kidneys. However, gluconate levels were significantly lower in the perfusate of DGF kidneys at 4 hr with a good AUROC for prediction of DGF. Gluconate (like mannitol) is present to provide osmotic stability so a significant change in concentration is unexpected. It is possible that cellular damage in the DGF kidneys allows influx of gluconate into the cell that would not normally occur within healthy tissue.

Glutathione is included in many preservation solutions, including KPS-1, and acts as a free radical scavenger to attenuate ischemia-reperfusion injury. Concentrations of reduced glutathione decreased over time during HMP in both DGF and IGF kidney perfusate. Glutathione in its oxidized form was not detected.

Glucose levels in the perfusate of DGF kidneys were significantly lower at both timepoints than those found in the perfusate of IGF kidneys, with good AUROCs for prediction of DGF. Although the reason for this is not known, it is possible that more glucose is used for repair by damaged DGF kidneys or that DGF kidneys are not effectively suppressed by hypothermia and require more glucose for metabolism.

Of the metabolites discovered that are not part of the preservation solution two of these, leucine and inosine, had significantly different concentrations in the perfusate of DGF kidneys compared to IGF kidney perfusate at 45 min. Leucine, a branched chain essential amino acid, was detected in significantly higher levels in the perfusate of DGF kidneys at 45 min. It is known that larger proteins are released into the perfusate during HMP as a sign of cellular damage.25 Raised concentrations of amino acids could indicate increased cellular breakdown in the more ischemically damaged DGF kidneys. Other amino acids identified in the perfusate were alanine, glycine, glutamate, isoleucine, tyrosine, and valine. The concentrations of all of these increased significantly over time in the perfusate of all kidneys but were not significantly different between the DGF and IGF groups.

In contrast, concentrations of inosine were significantly lower in DGF kidney perfusate at 45 min compared to IGF perfusate. Inosine is formed from the breakdown of adenosine nucleotides, such as adenosine monophosphate, adenosine diphosphate, and adenosine triphosphate. The lower levels of inosine detected in the more ischemically damaged DGF kidneys would suggest that this is not being released as a product of cellular degradation but is a product of cellular metabolism. Although the pathway involved is not clear, it is unlikely that the detected inosine was formed from the adenine in the preservation fluid because adenine levels did not decrease over time to correspond with this. The AUROCs for leucine and inosine at 45 min were good at 0.833 and 0.732.

Other metabolites of interest identified in HMP perfusate include the ketone body 3-hydroxybutyrate, increasing concentrations of which may indicate ongoing fatty acid metabolism which is the main source of energy in the renal cortex during hypothermia.26 Levels of lactate increased in the perfusate of HMP over time, as might be expected, because of the production of lactate by glycolysis in anaerobic conditions. Also detected to be increasing over time were citrate and glutamate which are both intermediates of the tricarboxylic acid cycle.

To further elucidate the processes occurring during HMP, examination of additional timepoints would be useful. Two early timepoints were chosen for this study to examine the potential for indicators of graft function that would be applicable in a clinically useful timeframe and to ensure that samples at comparable times were available for all kidneys. Furthermore, HMP parameters changed most markedly within the first hour of perfusion which might have been reflected in the metabolic profile of the perfusate. A combined analysis of HMP parameters and metabolomic data might be of interest but has not been performed in this study.

All cadaveric kidneys arrive at our unit in SCS. In this study, as in normal departmental practice, the decision to transfer to HMP was based on donor-recipient readiness and theatre availability. If the predicted time to theatre was within elective hours (8:00 A.M. to 8:00 P.M.), the kidney was transplanted from SCS. If the predicted time to theatre was outside of these hours (8:00 P.M. to 8:00 A.M.), the kidney was transferred to HMP and then transplanted at the earliest opportunity during elective hours. This is to ensure that as many renal transplants as possible are performed in optimal conditions in the dedicated renal transplant theatre.

At present, HMP is not used as a standard preservation technique at organ recovery in the United Kingdom. In this study, kidneys were placed on HMP at the accepting unit and remained at the center for transplantation. If HMP were used from recovery, sample collection for studies such as this would be more complicated. However, some evidence suggests that using HMP from recovery might be more beneficial than using a combination of SCS and HMP.2,27

The NMR has revealed many metabolites that may help to elucidate the underlying metabolic processes occurring during HMP. However, not surprisingly, there are limitations. Not every signal produced on the spectra can, as yet, be identified, and complex molecules can produce spectral patterns that overlap each other. This can sometimes make identification and quantification difficult. Furthermore, it is unclear how accurately levels of perfusate metabolites reflect intracellular activity. This study has used the technique to screen perfusate and has identified specific metabolites (glucose, inosine, leucine, gluconate) that might be predicative of graft function. Studies have now commenced to determine whether routine biochemical assays of these metabolites would be helpful in providing real time data to support clinical practice.

In this study, it has been possible to identify differences in the metabolomic profiles of perfusate from kidneys with IGF and DGF. These differing metabolites may prove to have a useful predictive role in viability assessment. With a better understanding of the underlying metabolic processes occurring in damaged kidneys, it may be possible to modify harmful metabolic processes, support cell function, and possibly extend storage periods before transplantation.


Adult cadaveric kidneys accepted for transplantation and undergoing HMP at the Queen Elizabeth Hospital, Birmingham between July 2012 and August 2013 were included, subject to consent and resource availability. Ethical approval was obtained (REC reference number: 12/WS/0166). All kidneys arrived at the unit in SCS. The decision to transfer a kidney to HMP was made according to departmental guidelines. Demographic and clinical data were collected prospectively. Delayed graft function was defined as the requirement for dialysis within the first postoperative week. Immediate graft function kidneys were those not requiring dialysis support postoperatively.

Kidneys were cold stored in the period after retrieval and transferred to the LifePort Kidney Transporter 1.0 (Organ Recovery Systems, Chicago, IL) at the host center under aseptic conditions. The decision to perfuse kidneys was determined by clinical protocol taking into account donor-recipient issues and theatre availability. Perfusion pressure was set at 30 mm Hg and not altered during perfusion time. All kidneys were perfused with 1 L of KPS-1 at 4 °C. No additional oxygen was supplied.

Two milliliters of perfusate was sampled at 45 min and 4 hr for each HMP kidney. Perfusate was transferred to a cryogenic vial and stored at −20°C until thawed at room temperature, prepared, and processed.

The NMR samples were prepared by mixing 150 μL of a 400 mM pH 7.0 phosphate buffer solution containing 2 mM (3-trimethylsilyl)propionic-(2,2,3,3-d4)-acid sodium salt (TSP) with 390 μL of each perfusate sample and 60 μL of deuterium oxide to reach a final phosphate buffer concentration of 100 mM and a final TSP concentration of 500 μM. Deuterium oxide provides a field-frequency lock, whereas TSP is used as a chemical shift as well as a concentration reference. After mixing, the 600-μL samples were pipetted into NMR tubes and centrifuged to remove any air bubbles.

The 1H-NMR spectra were acquired using a Bruker AVII 500 MHz spectrometer equipped with a 5 mm inverse Cryoprobe (Bruker Corporation, Billerica, MA). The sample temperature was set to 300 K, excitation sculpting was used to suppress the water resonance.28 One-dimensional spectra were acquired using a 6-kHz spectral width, 32,768 data points, 4 s relaxation delay and 128 transients. Matching was manual before acquisition of first sample, and each sample was automatically shimmed (1D-TopShim) to a TSP line width of less than 1 Hz before acquisition. Samples with a TSP line width greater than 1 Hz were acquired again after manual shimming where the TSP half height line width was shimmed below 1 Hz. Total experimental time was approximately 15 min per sample.

All data sets were processed using the MATLAB based MetaboLab software.29 Data sets were zero filled to 65,536 data points. An exponential line broadening of 0.3 Hz was applied before Fourier transformation. The chemical shift axis was calibrated by referencing the TMSP signal to 0 ppm. Spectra were manually phase corrected and baseline correction using a spline before segmental alignment of all resonances using Icoshift.30 Spectra were then exported into Bruker format.

Resultant spectra were examined using Chenomx 7.0 (Chenomx Inc., Edmonton, AB, Canada) profiling to identify metabolites and their concentrations. Chemical shift assignments are shown in SDC4, ( Example spectra are shown in SDC5, Each signal annotation and quantification was checked manually.


Data were analyzed using GraphPad Prism version 6.0c for Mac (GraphPad Software, La Jolla, CA, and IBM SPSS 19 (IBM Corp. Armonk, NY). Metabolite averages stated as median values because of nonparametric data distribution. Change in metabolite concentration over time was analyzed by Wilcoxon signed rank test. Metabolite concentrations were compared using Mann-Whitney U test. P less than 0.05 was considered to be indicative of statistical significance.


The authors thank James Hodson (University Hospitals Birmingham) for his assistance with the statistical analysis and to Peter DeMuylder & Gunther Vanwezer at Organ Recovery Systems for their help and support.


1. Henry SD, Guarrera JV. Protective effects of hypothermic ex vivo perfusion on ischemia/reperfusion injury and transplant outcomes. Transplant Rev (Orlando). 2012; 26: 163.
2. Moers C, Smits JM, Maathuis M-HJ, et al. Machine perfusion or cold storage in deceased-donor kidney transplantation. N Engl J Med. 2009; 360: 7.
3. Machine Preservation Trial Group. Machine Perfusion or cold storage in deceased-donor transplantation. N Engl J Med. 2012; 366: 770.
4. Lodhi SA, Lamb KE, Uddin I, et al. Pulsatile pump decreases risk of delayed graft function in kidneys donated after cardiac death. Am J Transplant. 2012; 12: 2774.
5. Cannon RM, Brock GN, Garrison RN, et al. To pump or not to pump: a comparison of machine perfusion vs cold storage for deceased donor kidney transplantation. J Am Coll Surg. 2013; 216: 625; discussion 633–4.
6. Audard V, Matignon M, Dahan K, et al. Renal transplantation from extended criteria cadaveric donors: problems and perspectives overview. Transpl Int. 2008; 21: 11.
7. Metzger RA, Delmonico FL, Feng S, et al. Expanded criteria donors for kidney transplantation. Am J Transplant. 2003; 3 (Suppl 4): 114.
8. Ojo AO, Hanson JA, Meier-Kriesche H, et al. Survival in recipients of marginal cadaveric donor kidneys compared with other recipients and wait-listed transplant candidates. J Am Soc Nephrol. 2001; 12: 589.
9. Port FK, Bragg-Gresham JL, Metzger RA, et al. Donor characteristics associated with reduced graft survival: an approach to expanding the pool of kidney donors. Transplantation. 2002; 74: 1281.
10. Fuller BJ, Lee CY. Hypothermic perfusion preservation: the future of organ preservation revisited? Cryobiology. 2007; 54: 129.
11. 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.
12. Hauet T, Gibelin H, Godart C, et al. Kidney retrieval conditions influence damage to renal medulla: evaluation by proton nuclear magnetic resonance (NMR) pectroscopy. Clin Chem Lab Med. 2000; 38: 1085.
13. Gibelin H, Eugene M, Hebrard W, et al. Comparison of Euro-Collins and University of Wisconsin solutions in the isolated perfused rat liver model: evaluation by proton nuclear magnetic resonance spectroscopy. Transplant Proc. 2000; 32: 2792.
14. Bon D, Claire B, Thuillier R, et al. Analysis of perfusates during hypothermic machine perfusion by NMR spectroscopy: a potential tool for predicting kidney graft outcome. Transplantation. 2014; 97: 810.
15. Jochmans I, Pirenne J. Graft quality assessment in kidney transplantation: not an exact science yet!. Curr Opin Organ Transplant. 2011; 16: 174.
16. Louvar DW, Li N, Snyder J, et al. “Nature versus nurture” study of deceased-donor pairs in kidney transplantation. J Am Soc Nephrol. 2009; 20: 1351.
17. Jochmans I, Moers C, Smits JM, et al. The prognostic value of renal resistance during hypothermic machine perfusion of deceased donor kidneys. Am J Transplant. 2011; 11: 2214.
18. de Vries EE, Hoogland ER, Winkens B, et al. Renovascular resistance of machine-perfused DCD kidneys is associated with primary nonfunction. Am J Transplant. 2011; 11: 2685.
19. Sonnenday CJ, Cooper M, Kraus E, et al. The hazards of basing acceptance of cadaveric renal allografts on pulsatile perfusion parameters alone. Transplantation. 2003; 75: 2029.
20. Bhangoo RS, Hall IE, Reese PP, et al. Deceased-donor kidney perfusate and urine biomarkers for kidney allograft outcomes: a systematic review. Nephrol Dial Transplant. 2012; 27: 3305.
21. Nicholson JK, Lindon JC, Holmes E. ‘Metabonomics’: understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica. 1999; 29: 1181.
22. Seto K, Ikehira H, Obata T, et al. Long-term assessment of posttransplant renal prognosis with 31 P magnetic resonance spectroscopy. Transplantation. 2001; 72: 627.
23. Hauet T, Gibelin H, Richer JP, et al. Influence of retrieval conditions on renal medulla injury: evaluation by proton NMR spectroscopy in an isolated perfused pig kidney model. J Surg Res. 2000; 93: 1.
24. Richer JP, Baumer TH, Gibelin H, et al. Evaluation of renal medulla injury after cold preservation and transplantation: noninvasive determination of medullar damage by proton nuclear magnetic resonance spectroscopy of urine and plasma. Transplant Proc. 2000; 32: 47.
25. Hoogland ERP, de Vries EE, Christiaans MH, et al. The value of machine perfusion biomarker concentration in DCD kidney transplantations. Transplantation. 2013; 95: 603.
26. Ayala-García M, Hernández M, Ramírez-Barba É, et al. Preservation of Renal Allografts for Transplantation. In: Long L, ed. Renal Transplantation—Updates and Advances. Rijeka, Croatia: Intech Europe; 2012. ISBN: 978-953-51-0173-4. Available from:
27. Watson CJ, Wells AC, Roberts RJ, et al. Cold machine perfusion versus static cold storage of kidneys donated after cardiac death: a UK multicenter randomized controlled trial. Am J Transplant. 2010; 10: 1991.
28. Hwang T-L, Shaka AJ. Water suppression that works. Excitation sculpting using arbitrary waveforms and pulsed field gradients. J Magn Reson. 1995; 112: 275.
29. Ludwig C, Günther UL. MetaboLab—advanced NMR data processing and analysis for metabolomics. BMC Bioinformatics. 2011; 12: 366.
30. Savorani F, Tomasi G, Engelsen SB. Icoshift: a versatile tool for the rapid alignment of 1D NMR spectra. J Magn Reson. 2010; 202: 190.

Supplemental Digital Content

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.