Anal. Chem. 2005, 77, 5570-5578
Metabolic Assessment of Human Liver Transplants from Biopsy Samples at the Donor and Recipient Stages Using High-Resolution Magic Angle Spinning 1H NMR Spectroscopy Iola F. Duarte,† Elizabeth G. Stanley,‡ Elaine Holmes,‡ John C. Lindon,‡ Ana M. Gil,† Huiru Tang,‡ Roxanne Ferdinand,‡ Claire Gavaghan McKee,‡ Jeremy K. Nicholson,‡ Hector Vilca-Melendez,§ Nigel Heaton,§ and Gerard M. Murphy*,‡
Department of Chemistry, CICECO, University of Aveiro, 3810-193 Aveiro, Portugal, Biological Chemistry, Biomedical Sciences Division, Imperial College London, Sir Alexander Fleming Building, South Kensington, London SW7 2AZ U.K., and Liver Transplant Unit, Institute of Liver Studies, King’s College Hospital, Denmark Hill, London SE5 9RS U.K.
This work presents the first application of high-resolution magic angle spinning (HR-MAS) 1H NMR spectroscopy to human liver biopsy samples, allowing a determination of their metabolic profiles before removal from donors, during cold perfusion, and after implantation into recipients. The assignment of peaks observed in the 1H HRMAS NMR spectra was aided by the use of two-dimensional J-resolved, TOCSY and 1H-13C HMQC spectra. The spectra were dominated by resonances from triglycerides, phospholipids, and glycogen and from a variety of small molecules including glycerophosphocholine (GPC), glucose, lactate, creatine, acetate, amino acids, and nucleoside-related compounds such as uridine and adenosine. In agreement with histological data obtained on the same biopsies, two of the six livers were found to contain high amounts of triglycerides by NMR spectroscopy, which also indicated that these tissues contained a higher degree of unsaturated lipids and a lower proportion of phospholipids and low molecular weight compounds. Additionally, proton T2 relaxation times indicated two populations of lipids, a higher mobility triglyceride fraction and a lower mobility phospholipid fraction, the proportions of which changed according to the degree of fat content. GPC was found to decrease from the pretransplant to the posttransplant biopsy of all livers except for one with a histologically confirmed high lipid content, and this might represent a biomarker of liver function posttransplantation. NMR signals produced by the liver preservation solution were clearly detected in the cold perfusion stage biopsies of all livers but remained in the posttransplant spectra of only the two livers with a high lipid content and were prominent mainly in the graft that later developed primary graft dysfunction. This study has shown biochemical differences between livers used for transplants that can be related to the degree and type of * Corresponding author. E-mail:
[email protected]. † University of Aveiro. ‡ Imperial College London. § King’s College Hospital.
5570 Analytical Chemistry, Vol. 77, No. 17, September 1, 2005
lipid composition. This technology might therefore provide a novel screening approach for donor organ quality and a means to assess function in the recipient after transplantation. Liver transplantation is now firmly established as a viable treatment for patients with acute and chronic end-stage liver disease, and its success starts with proper organ retrieval and adequate donor assessment. Although the assessment of donor livers prior to transplantation has been the subject of much research, the transplant surgeon still has to rely on a subjective interpretation of donor clinical data and visual inspection of the liver to decide whether to use the graft. Furthermore, due to the widening of indications for treatment and the current shortage of organ donors, experienced transplant centers are increasingly using suboptimal or “marginal” grafts, such as those with a visually assessed high level of fat, which have higher risks of dysfunction/ nonfunction. This has led to a renewed interest in finding reliable methods for assessing donor livers and predicting graft dysfunction in order to reduce clinical complications and costs after transplantation. Monitoring of liver function is difficult, and although many tests have been evaluated to reliably assess pretransplant graft function, none has yet found a place in clinical practice.1 The simultaneous detection of many metabolites in biofluids, tissue extracts, and tissues and the evaluation of their changed levels in a pathological situation has been termed metabonomics.2 The main technologies underpinning this approach are 1H NMR spectroscopy and LC coupled to mass spectrometry.3 The principal approach has been to analyze biofluids such as urine or bile,4 and indeed, high-resolution 1H NMR of hepatic bile from liver donors and recipients enabled the detection of compositional differences (1) Vilca Melendez, H.; Rela, M.; Murphy, G.; Heaton N. Transplantation 2000, 70, 560-565. (2) Nicholson, J. K.; Lindon, J. C.; Holmes, E. Xenobiotica 1999, 29, 11811189. (3) Lindon, J. C.; Holmes, E.; Bollard, M. E.; Stanley, E. G.; Nicholson, J. K. Biomarkers 2004, 9, 1-31. (4) Lindon, J. C.; Nicholson, J. K.; Holmes, E.; Everett, J. R. Concepts Magn. Reson. 2000, 12, 289-320. 10.1021/ac050455c CCC: $30.25
© 2005 American Chemical Society Published on Web 07/26/2005
between normal and fatty grafts that may potentially help to predict the development of primary graft dysfunction (PGD),5 and an attempt was made to use 1H NMR spectroscopy of serum and urine from one liver transplant recipient to monitor glutamine and urea levels as possible indicators of graft function.6 However, in this study, the graft showed necrosis from hepatic artery thrombosis, and assessment of graft function posttransplant in such a damaged liver was not feasible. In addition, in vivo and in vitro 31P NMR spectroscopy has been used to provide information on the hepatic energetics and viability of cold preserved liver using animal models,7,8 the rejection of human allografts after transplantation,9 and the metabolic abnormalities of human cirrhotic liver.10,11 While biofluid NMR spectroscopy provides an integrated view of the whole organism, to achieve a more direct correlation with tissue histology, the direct biochemical analysis of whole tissue is now achievable using high-resolution magic angle spinning (HRMAS) NMR spectroscopy of small biopsy samples. This technique has been used to characterize the composition of normal12,13 and toxin-treated rat liver14,15 but has not been previously applied to human liver studies. The potential of this technique for studying human biological tissues has been largely demonstrated by recent work on kidney,16 prostate,17 breast tumors,18 and brain tumors,19 proving it to be a rapid, noninvasive, and powerful method for characterizing metabolic composition and biochemical alterations in such systems. In this work, we report the first application of HR-MAS 1H NMR spectroscopy to human liver tissues evaluating the metabolic composition and molecular dynamics with the aim of deducing the type and quality of analytical information that can be obtained on such human biopsy samples collected at surgery. The differ(5) Melendez, H. V.; Ahmadi, D.; Parkes, H. G.; Rela, M.; Murphy, G.; Heaton, N. Transplantation 2001, 72, 855-860. (6) Singh, H. K.; Yachha, S. K.; Saxena, R.; Gupta, A.; Gowda, G. A. N.; Bhandari, M.; Khetrapal, C. L. NMR Biomed. 2003, 16, 185-188. (7) Changani, K. K.; Fuller, B. J.; Bell, J. D.; Bryant, D. J.; Moore, D. P.; TaylorRobinson, S. D.; Davidson, B. R. Transplantation 1996, 62, 787-793. (8) Changani, K. K.; Fuller, B. J.; Bryant, D. J.; Bell, J. D.; Ala-Korpela, M.; Taylor-Robinson, S. D.; Moore, D. P.; Davidson, B. R. J. Hepatol. 1997, 26, 336-342. (9) Taylor-Robinson, S. D.; Sargentoni, J.; Bell, J. D.; Thomas, E. L.; Marcus, C. D., Changani, K. K.; Saeed, N.; Hodgson, H. J. F.; Davidson, B. R.; Burroughs, A. K.; Rolles, K.; Foster, C. S.; Cox, I. J. Gut 1998, 42, 735743. (10) Taylor-Robinson, S. D.; Thomas, E. L.; Sargentoni, J.; Marcus, C. D.; Davidson, B. R.; Bell, J. D. Biochim. Biophys. Acta 1995, 1272, 113-118. (11) Taylor-Robinson, S. D.; Sargentoni, J.; Bell, J. D.; Saeed, N.; Changani, K. K.; Davidson, B. R.; Rolles, K.; Burroughs, A. K.; Hodgson, H. J. F.; Foster, C. S.; Cox, I. J. Liver 1997, 17, 198-209. (12) Bollard, M. E.; Garrod, S.; Holmes, E.; Lindon, J. C.; Humpfer, E.; Spraul, M.;, Nicholson, J. K. Magn. Reson. Med. 2000, 44, 201-207. (13) Waters, N. J.; Garrod, S.; Farrant, R. D.; Haselden, J. N.; Connor, S. C.; Connelly, J.; Lindon, J. C.; Holmes, E.; Nicholson, J. K. Anal. Biochem. 2000, 282, 16-23. (14) Waters, N. J.; Holmes, E.; Waterfield, C. J.; Farrant, R. D.; Nicholson, J. K. Biochem. Pharmacol. 2002, 64, 67-77. (15) Garrod, S.; Humpfer, E.; Connor, S. C.; Connelly, J. C.; Spraul, M.; Nicholson, J. K.; Holmes, E. Magn. Reson. Med. 2001, 45, 781-790. (16) Moka, D.; Vorreuther, R.; Schicha, H.; Spraul, M.; Humpfer, E.; Lipinski, M.; Foxall, P. J. D.; Nicholson, J. K.; Lindon, J. C. J. Pharm. Biomed. Anal. 1998, 17, 125-132. (17) Taylor, J. L.; Wu, C. L.; Cory, D.; Gonzalez, R. G.; Bielecki, A.; Cheng, L. L. Magn. Reson. Med. 2003, 50, 627-632. (18) Sitter, B.; Sonnewald, U.; Spraul, M.; Fjosne, H. E.; Gribbestad, I. S. NMR Biomed. 2002, 15, 327-337. (19) Barton, S. J.; Howe, F. A.; Tomlins, A. M.; Cudlip, S. A.; Nicholson, J. K. MAGMA 1999, 8, 121-128.
ences between donor livers was also assessed and related to the liver quality. Here, biopsies from six donor human livers have been characterized at three stages of the transplantation process, before organ retrieval, during cold storage, and after implantation in the recipient with a view to exploring the metabolic signatures reflecting graft success and to assess the timing of biopsies with respect to prediction of graft survival. Assignment of resonance peaks in 1D and 2D NMR spectra was carried out in order to identify the main metabolites in human liver and the compositional differences between donor organs. NMR relaxation time differences have been used to assess metabolite molecular dynamics and hence compartmentation. In addition, the biochemical changes of each liver during the various phases of the transplantation process have been followed. EXPERIMENTAL SECTION Patients. Six randomly selected liver donors and corresponding transplant recipients were included in this study, having obtained Hospital Ethics Committee permission. The King’s College Hospital surgical team retrieved all liver grafts and transplanted all recipients included in this study. Standard clinical and biochemical data were collected during organ retrieval and orthotopic liver transplantation. The development of PGD was monitored in all recipients using the parameters proposed by Clavien and coauthors,20 which include first day aspartate aminotransferase (AST) g 2000 IU/L, a transient increase in AST levels of g1000 IU/L, or a persistent elevated prothrombin time of >20 s (or the equivalent international normalized ratio [INR] > 1.4) for at least 3 days. Liver graft characteristics and other relevant clinical and operative data of donors (labeled A-F) and recipients are summarized in Table 1. Samples. The liver tissue samples were obtained by Tru-cut biopsy from the left lobe of the donor livers at three stages during organ transplantation: before liver retrieval (biopsy 1), at the end of the cold storage period (biopsy 2), and after the liver was implanted into the recipient and the transplantation was completed (biopsy 3), producing a total of 18 samples. Thus, the three biopsies were taken from the same graft at different stages of the procedure, with the aim of determining how each graft changed during the transplant process and assessing its quality. All grafts (marginal or normal) went through the same stages of the procedure, and although strict time intervals between biopsies are not practicable because the times of retrieval, cold storage, and surgery are inevitably different, their quality may be determined by the recovery of their metabolic function. Half of each biopsy was immediately snap-frozen with liquid nitrogen and stored at -70 °C prior to NMR analysis, and the other half was placed in formalin prior to histological examination. For NMR spectroscopic analysis, samples (19.5 ( 1.0 mg) were washed with D2O saline (0.9%) and packed into 4-mm-diameter zirconia MAS rotors with top inserts to give a final volume of 15 µL. These provide a symmetrical distribution of the tissue sample, therefore improving spectral resolution and sensitivity.13 NMR Measurements. HR-MAS 1H NMR data were acquired on a Bruker AV-400 spectrometer, operating at 400.13 MHz for 1H observation, at a temperature of 288 K and a spinning rate of (20) Clavien, P.; Camargo, C. J.; Croxford, R.; Langer, B.; Levy, G.; Greig, P. Ann. Surg. 1994, 220, 109.
Analytical Chemistry, Vol. 77, No. 17, September 1, 2005
5571
Table 1. Demographic and Clinical Data for Donor Livers
donor
age
sex
diagnosisa
A B C D E F
67 55 64 63 44 56
F F F M M F
SAH ICH RTA SAH CI SAH
a
wt (kg)
days in ITU
liver histology (fat infiltration)
liver wt (g)
cold ischemia (h)
PGD
recipient status
49 70 55 81 80 70
1 1 1 1 1 3
< 5% (normal) < 5% (normal) < 5% (normal) < 5% (normal) 20% (mild) 40% (moderate)
1246 1327 1135 1537 1905 1840
13:15 11:33 9:25 11:18 11:25 12:21
no no no no yes no
alive alive alive alive dead alive
SAH, subarachnoid hemorrhage; ICH, intracerebral hemorrhage; RTA, road traffic accident; CI, cerebral infarct; PGD, postgraft dysfunction.
4 kHz. Typically, 128 transients were collected into 32K data points using a standard 1D pulse sequence (relaxation delay-90°-t190°-tm-90°-acquire FID), in which the water signal is irradiated during the relaxation delay (2 s) and the mixing period (tm ) 100 ms), with t1 being a short delay of 3 µs. A spectral width of 6000 Hz and an acquisition time of 2.73 s were used. Additionally, the Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence21 with simple presaturation of the water peak and a total spin-spin relaxation delay (2nτ) of 120 ms was used to measure spin-echo 1H spectra. Typically, 256 transients were collected into 32K data points. All 1D spectra were processed with a line broadening of 0.3 Hz and a zero filling factor of 2. To aid spectral assignment, 2D 1H-1H total correlation (TOCSY), J-resolved (JRES), and 1H13C heteronuclear multiple quantum coherence (HMQC) spectra were recorded on selected samples. The TOCSY spectra were acquired in the phase-sensitive mode using time proportional phase incrementation and the MLEV17 pulse sequence for the spin lock.22 For each spectrum, 4K data points with 64 transients per increment and 128 increments were acquired with a spectral width of 4195 Hz in both dimensions. The relaxation delay between successive pulse cycles was 2 s, and the mixing time of the MLEV spin lock was 80 ms. These data were zero-filled in the f1 dimension to 256 data points, and a sine-bell apodization function was applied prior to FT. 1H-13C HMQC spectra were recorded with 13C decoupling during acquisition.23 The 2K data points with 48 scans per increment and 128 increments were acquired with spectral widths of 4195 and 20 121 Hz in the 1H and 13C dimensions, respectively. The FIDs were weighted using a sine-bell-squared function in both dimensions and were zerofilled in the f1 dimension to 512 data points. 2D homonuclear J-resolved spectra24 with water presaturation were measured by acquiring 8K data points with 64 transients per each of 64 increments, using spectral widths of 6000 Hz in the f2 dimension and 60 Hz in the f1 (J-coupling) dimension. Prior to FT, the FIDs were weighted in both dimensions by a sine-bell function and zerofilled in the f1 dimension to 256 data points. The spectra were tilted by 45° to provide orthogonality of the chemical shift and coupling constant axes and subsequently symmetrized about the f1 axis. In addition to 1D and 2D spectra, relaxation time measurements were performed under MAS conditions (rotation rate 4 kHz). 1H spin-lattice relaxation times (T1) were measured by using the inversion-recovery sequence (180°-tr-90°-FID), (21) Meiboom, S.; Gill, D. Rev. Sci. Instrum. 1958, 29, 688-691. (22) Bax, A.; Davis, D. G. J. Magn. Reson. 1985, 65, 355-360. (23) Bax, A.; Griffey, R. H.; Hawkins, B. L. J. Magn. Reson. 1983, 55, 301-315. (24) Aue, W. P.; Karhan, J.; Ernst, R. R. J. Chem. Phys. 1976, 64, 4226-4227.
5572
Analytical Chemistry, Vol. 77, No. 17, September 1, 2005
where the recovery time (tr), was varied in the range 5 ms-10 s, recording 18 experimental points. The 1H T1 values were calculated by monoexponential fitting of the inversion recovery curves. 1H T relaxation times were measured by a CPMG sequence (90°2 ts-180°-ts-FID), by varying the spin-echo time (ts) in the 5 ms-5 s range (18 experimental points). The plots of intensity versus echo-time clearly did not fit a single-exponential decay and were thus fitted by a biexponential function, from which two transverse relaxation time constants T2a and T2b and two intensities were obtained, respectively, for a slower relaxing (more mobile) and a faster relaxing (less mobile) proton fraction. Histological Fat Assessment. Fat infiltration in the graft was a qualitative assessment made by an experienced histopathologist. The biopsy sample was first fixed in formalin and then stained with haematoxylin/eosin (which is not specific for fat). Under the microscope, the pathologist then determined the percentage of macrovesicular fat present in relation to the adjacent tissue and reported as normal (less than 10% fat infiltration), mild (10-30% fat infiltration), moderate (30-60% fat infiltration), or severe (more than 60% fat infiltration). RESULTS AND DISCUSSION Clinical Results. The six liver donors (four female and two male) had an average age of 59.5 years (range 44-67 years), average intensive therapy unit (ITU) stay of 1 day (range 1-3), and normal liver function tests. Of these donors, one provided a normal graft and five provided “mildly fatty” livers as visually assessed by the surgeon during organ retrieval. However, the histological assessment of these livers (Table 1) showed that four grafts could be considered “normal” (fat infiltration below 5%), one mildly fatty (20% fat infiltration and focal parenchyma necrosis), and one moderately fatty (40% fat infiltration). The liver weight (Table 1) also correlated with the degree of fat infiltration with an average of 1.3 kg for the normal grafts (n ) 4) and 1.9 kg for the “fatty” grafts (n ) 2). All donor livers were transplanted as whole grafts in six recipients, all male with an average age of 48 years (range 25-54 years). The average surgical time was 7 h (range 5-13 h), and average cold ischemia time was 12 h (range 9-13 h). Five transplant recipients are alive with a median follow-up of 18.5 months. The remaining patient was retransplanted for hepatitis C recurrence with the graft retrieved from donor E (20% fat infiltration). The patient developed PGD posttransplantation (first day AST of 2599 IU/L), and his recovery was complicated by a hemorrhagic episode that rendered his liver graft ischemic. He underwent a third liver transplant but died of multiorgan failure.
Figure 1. 400-MHz 1H HR-MAS NMR spectra of a human liver biopsy sample (rotation rate 4 kHz): (a) standard 1D spectrum; (b) spin-echo (CPMG) spectrum; (c) JRES f2 projection. Assignment: L1, lipid CH3; L2, lipid (CH2)n; L3, lipid CH2CH2CO; L4, lipid CH2CHdCH; L5, lipid CH2CO; L6, lipid CHdCHsCH2sCHdCH; L7, lipid CHdCH; Cho, choline; PC, phosphocholine; GPC, glycerophosphocholine; TMAO, trimethylamine-N-oxide; Bet, betaine; Glc, glucose; Val, valine; Leu, leucine; Ala, alanine; Gln, glutamine; Gly, glycine; Tyr, tyrosine; Urd, uridine; Ado, adenosine.
NMR Spectral Assignments for Human Liver Tissue. The conventional 1H NMR spectra of the human liver tissue samples (Figure 1a), recorded under HR-MAS conditions at a 4-kHz spinning rate, showed principally broad resonances arising from fatty acyl chains in lipid molecules, identified as mainly triglycerides because of the characteristic glyceryl moiety peaks. These were overlapped with sharper signals from lower molecular weight compounds. The spin-echo spectrum (Figure 1b) showed a lower intensity for the lipid peaks, which are attenuated in the CPMG experiment due to their shorter T2 relaxation times, enabling a clearer inspection of the signals arising from small molecules. This was supported by further reduction of the broad lipid resonances and removal of J-coupling effects achieved by JRES spectroscopy, as shown by the f2 projection in Figure 1c. To aid spectral interpretation, the information provided by 1D and 2D TOCSY, HMQC, and JRES spectra was combined and resonances were assigned by comparison of 1H and 13C chemical shifts and spin-spin coupling constants with literature values of metabolites commonly found in biological samples.12,15,25,26 A complete list of identified compounds is given in Table 2, with corresponding 1H and 13C NMR chemical shifts. In the low-frequency region of the spectrum (δ 0-3), the fatty acyl groups from various types of lipid give rise to broad, intense (25) Fan, T. W. M. Prog. NMR Spectrosc.1996, 28, 161-219. (26) Nicholson, J. K.; Foxall, P. J. D.; Spraul, M.; Farrant, R. D.; Lindon, J. C. Anal. Chem. 1995, 67, 793-811.
resonances for which spin-spin connectivities and 1H-13C correlations are easily identified in the TOCSY (Figure 2a) and HMQC (Figure 2b) spectra, respectively. In addition, several amino acids, e.g., valine, leucine, alanine, lysine, and proline, and some organic acids, namely acetate and lactate, could also be detected. In the case of lactate, its assignment was enabled by TOCSY and JRES spectra, since in the 1D NMR spectra its methyl signal was overlapped by the broad and prominent lipid resonance at δ 1.30 and its methine peak by the glyceryl resonance from triglycerides at δ 4.10. In the midfrequency region (δ 3-6), one of the major contributions arises from glucose, clearly identified by its anomeric proton peaks at δ 4.65 and δ 5.23. Glycogen was also found to be present, although its detection is hindered both by the broad nature of its signals, resulting from the short T2 relaxation times, and by overlap with the signals of unsaturated lipids. However, the fact that glycogen signals are detected at all, given its high molecular weight, is probably due to the high flexibility of the carbohydrate chains.12 Other important resonances in this region arise from the glyceryl moiety protons (δ 4.10, 4.31, 5.24), the N+(CH3)3 group of choline, phosphocholine, glycerophosphocholine (GPC), and betaine, and the N(CH3)3 group of trimethylamine-N-oxide (TMAO), as well as the NCH2 group of taurine, which together give rise to a cluster of singlets in the range δ 3.1-3.3, with the taurine band as a triplet. The singlets of glycine and creatine are also detected. In the high-frequency region (>δ 6), there are a number of low-intensity signals, which have been assigned to uridine, uracil, adenine, adenosine, tyrosine, and phenylalanine (insets in Figure 1). Comparison of Biopsies from Different Donor Livers. The 1H HR-MAS NMR spectra of liver biopsies collected from the six donors (A-F) before organ retrieval are shown in Figure 3. Although the spectra show similar qualitative profiles, there are clear differences in the ratios of some spectral components. Assuming that the MAS technique provides comparable visibility of the NMR peaks for all major mobile metabolites,27 and given that all of the biopsy samples were of similar mass (18.5-20.5 mg), the overall changes in peak area relative to the spectral noise level provide an indication of how metabolite concentration levels are different between the different donor livers. In the lowfrequency region, the signals of the fatty acyl chains showed lower intensities in samples A-C, indicating that these samples contained smaller amounts of triglycerides, compared to samples D-F. By dividing the sum of the areas of fatty acyl signals by the total spectral area, it was found that donors E and F contain 1.51.8 times more triglycerides than donors A-C, whereas donor D has an intermediate amount of fat (Figure 4a). Histological examination of biopsies collected from the same grafts also classified samples E and F as the fattest, estimating 20 and 40% of fat infiltration, respectively, whereas all other samples were considered to have δ 6, not shown), a few minor metabolites are detected in samples A-C (e.g., tyrosine, uridine, and adenosine), whereas the spectra of samples D-F show much lower signal-to-noise ratio thus hindering further assignments. Based on the NMR results presented above, the six donor livers analyzed show three distinct types. Livers E and F are characterized by the highest amount of triglycerides (in agreement with histological data), a higher degree of lipid unsaturation, and a markedly smaller proportion of phospholipids and of low
molecular weight compounds such as glucose and amino acids. Liver D contains an intermediate amount of fat. Livers A-C have the smallest amount of triglycerides and the highest proportion of phospholipids and low molecular weight compounds. At this stage, given the low number of samples analyzed, it is not possible to establish definitive objective correlations between these compositional differences and the performance of grafts in the recipients. Liver Metabolite Time Course Changes. To investigate variations in the liver composition throughout the transplantation process, the spectra of the three biopsies, collected from each organ before retrieval from the donor, at the end of the cold storage period, and after graft implantation in the recipient, have been compared. These results are shown in Figure 5 for liver A. Most livers showed comparable effects apart from slight variations in the intensities of acetate and some amino acids and specific differences described below. In the low-frequency region, there are no significant differences observed between the three sampling stages and in the midfrequency region; the most prominent difference arises from the strong contribution of the University of Wisconsin (UW) solution in the spectrum of the cold perfusion period biopsy samples. UW solution is sugar-based and is used to perfuse the graft during its retrieval from the donor in order to preserve the tissue during Analytical Chemistry, Vol. 77, No. 17, September 1, 2005
5575
Figure 4. Area ratios of some NMR signals measured in the spectra of donor livers A-F: (a) triglycerides (fatty acid chain peaks, δ 0.90, 1.30, 1.60, 2.04, 2.26, 2.77, 5.33) relative to total spectral area; (b) unsaturated fatty acid signal (δ 5.33) relative to the total area of triglycerides signals; (c) GPC (δ 3.23) and PC (δ 3.22) relative to total spectral area. Coefficients of variation, obtained from the areas measured in three replicate spectra of each sample, are 2πω0-1, where τc is the rotational correlation time and 2πω0 is the Larmor frequency, 4 × 108 s-1. The values increased as a function of temperature, and from an Arrhenius analysis, this could be used in principle to derive activation energies for the molecular motion as carried out recently for C18 chromatographic phases.33 In addition, the biopsies of livers E and F are characterized by higher T1 values (0.400-0.421 ms), compared to those found for the remaining samples (0.2950.367 ms). This is consistent with previous studies on acyl chains, and the longer T1 value can be associated with a more mobile lipid fragment.33 Thus, it can be concluded that the higher T1 found for livers E and F do reflect higher mobility of their lipid fraction. Regarding the comparison of biopsies collected from each graft at different stages of the transplantation process, no significant differences are found in 1H T1 and T2 relaxation time constants. (33) Coen, M.; Wilson, I. D.; Nicholson, J. K.; Tang, H.; Lindon, J. C. Anal. Chem. 2004, 76, 3023-3028.
Analytical Chemistry, Vol. 77, No. 17, September 1, 2005
5577
CONCLUSIONS The potential of HR-MAS 1H NMR spectroscopy for characterizing the metabolic profile and biochemistry of human liver grafts has been shown for the first time, as a preliminary exploratory attempt to assess the quality and accuracy of metabolic information from human biopsy samples, to evaluate graft quality for transplantation, and to seek to discover biomarkers of graft function. Histological studies, the currently used approach to assess liver fat levels, is limited to a visualization of the morphology, whereas the NMR spectroscopic approach can yield information on different types of lipid, their relative concentrations, and their variations in molecular dynamics. By histology, two of the grafts were found to contain considerably higher amounts of fat compared to the remaining samples, and this was confirmed by the NMR analyses, which also characterized the lipids as mainly triglycerides. In addition, NMR spectroscopy showed that these samples are determined by a higher degree of lipid unsaturation and a smaller proportion of phospholipids and low molecular weight compounds such as amino acids, glucose, and nucleotiderelated compounds. The dynamics of the lipids, as assessed by proton T1 and T2 relaxation measurements, were also found to be different in these two samples, which showed a higher proportion of mobile triglyceride lipids typical of cytosolic-based droplets. Regarding the comparison of biopsies collected at different stages of the transplantation process, NMR spectroscopy detected consistent variations in most donor livers. First, GPC (a phospholipid degradation product) was seen to decrease for all livers but one, suggesting increased cell turnover. Interestingly, in the graft that developed PGD, GPC did not vary, probably reflecting a lower degree of cellular activity, and this substance might
5578
Analytical Chemistry, Vol. 77, No. 17, September 1, 2005
therefore represent a new biomarker for liver function. Second, the clearance of UW preservation solution was seen to be efficient only for nonfatty livers. Other variations included changes in nucleotide-related compounds and the increase of betaine/TMAO during the transplantation process, but the significance of these changes remains to be determined. These results show that HR-MAS 1H NMR spectroscopy is a useful tool for assessing intact liver tissue during the whole process of transplantation. Ongoing work aims at integrating the observations presented here with 1H NMR studies of biofluids (serum, urine, bile) collected from the same subjects. The establishment of objective relationships between the NMRdetected compositional variations and postoperative liver function, as well as the investigation of definitive biochemistry using UPLC-MS, is the scope of future work involving, necessarily, an expanded patient cohort. ACKNOWLEDGMENT The authors thank the South Thames Transplant Coordination Service for their help with the consenting process, the Roche Organ Transplantation Research Foundation for financial support, and Dr. Raffaelle Girlanda and Mrs. Carol Keenan-Pillai for their assistance with sample collection during organ retrieval. I.F.D. thanks the Foundation for Science and Technology, Portugal, for funding support through the Grant SFRH/BPD/11516/2002 within the III Community framework. Received for review March 16, 2005. Accepted June 24, 2005. AC050455C
