The intrinsic capacity of a donor liver to regain its original function after transplantation is largely determined by the integrity of the adenine nucleotide metabolism of the organ (1, 2). The adenine nucleotide composition can be assessed for viability testing with high-performance liquid chromatography (HPLC*) (1-4). HPLC is very sensitive, but the results are not readily available to the clinician because the technique is time-consuming. Another important drawback of HPLC for routine clinical use, is the necessity to take a liver biopsy, which carries the risks of liver damage and bacterial contamination. A method for rapid, noninvasive assessment of the adenine nucleotide composition of the donor liver would be helpful.
With phosphorus-31 magnetic resonance spectroscopy (31P-MRS), the adenine nucleotide composition can be assessed noninvasively in living cells. A potential role for 31P-MRS in viability research for transplantation has already been demonstrated in several studies on isolated animal livers (5-12) as well as in human kidneys (13). Preliminary experience on isolated human donor livers also suggested the potential clinical value of 31P-MRS as a rapid viability test (14). However, so far, 31P-MRS findings from isolated human donor livers have not correlated with posttransplant organ function. Therefore, the clinical value of 31P-MRS as a viability test has not been validated.
The purpose of this study was to investigate whether the composition of the high-energy phosphates (of which the adenine nucleotides adenosine mono-, di-, and triphosphate [AMP, ADP, and ATP, respectively] are part) of the isolated donor liver as assessed with 31P-MRS could serve as a viability indicator with prognostic value for transplantation outcome. Specifically, we tested whether the relevant spectral peak areas of the isolated donor liver correlated with the amount of hepatocellular graft damage and liver metabolic function shortly after implantation.
MATERIALS AND METHODS
Donor livers. Forty human livers were harvested according to standardized procedures (15). This group of livers was also subject of a tissue pH study (published previously [16]). During in situ cold flush, approximately 4 L of University of Wisconsin solution UW (DuPont Critical Care, Waukegan, IL) were flushed through the organ, 1.5 L were flushed through the portal vein, and 2.5 L were flushed through the aorta. At the same time, the abdominal cavity was filled with cold saline to provide adequate surface cooling of the liver. On the back table, an additional amount of UW was flushed through the hepatic artery and portal vein until the outflow from the caval vein was clear. The biliary system was also flushed with UW. Subsequently, the graft was packed in sterile plastic bags and stored in UW solution on melting ice in Styrofoam containers.
Spectroscopy. A standard spectroscopic measurement was performed 1 hr before the start of the recipient operation and took about 30 min. 31P-MRS was performed as described previously (14, 16). In brief, cold-stored livers were examined in their original package with a 1.5 Tesla clinical whole-body magnetic resonance system (Gyroscan S15-HP, Philips Medical Systems, Best, The Netherlands). A conventional 15-cm-diameter receive-only coil, tuned to the 1H and 31P resonance frequencies (63.5 and 25.6 Mhz, respectively), was used in all measurements. The body coil was used for the excitation pulses. To reduce excessive signal from the inorganic phosphate of the UW, the liver was positioned as close as possible to the coil surface. The spatial relation of the liver and the coil was verified by proton magnetic resonance imaging. Signal strength was optimized by adjusting the static magnetic field, guided by the proton signal. Adiabatic excitation pulses with a duration of 170 MSEC and a flip angle of 90° were applied. All spectra were processed out of 128 signal acquisitions over a spectral width of 3000 Hz with a 3-sec repetition time. The signal was Fourier-transformed to 4096 points in the frequency domain. Peak areas were calculated in a time domain fitting procedure in which a sum of exponentially damped sinusoids is fitted to the time signal. The same custom-designed fitting program was used for all spectra. To determine time-dependent alterations in the high-energy phosphate composition, sequential measurements of the same liver were performed when allowed by the logistics of the transplantation procedure. This was the case in two cases of a local liver donation and in three cases of a reduced size liver transplantation in which the remaining part of the liver was repeatedly examined.
Liver transplantation. Orthotopic liver transplantation was performed according to standardized procedures (15).
Spectral analysis. As was known from our previous experiments (14), if phosphocreatine (PCr) and β-ATP resonances were present, they were very small, which made them unsuitable for quantification by means of a peak area ratio. For that reason, we determined whether the presence of PCr and/or β-ATP was correlated with the amount of hepatocellular damage or liver metabolic function.
The remaining spectral resonances of phosphomonoesters (PME), inorganic phosphate (Pi), and phosphodiesters (PDE) were quantitated relative to the area of the nicotine adenine dinucleotide (NAD(H)) peak. This peak is present in the spectra of all livers, and its position on the horizontal scale is pH-independent (9, 12, 17). The NAD(H) peak area was corrected for the contribution of α-ATP and α-ADP by subtracting peak areas equal to β-ATP plus (γ-ATP + β-ADP) minus β-ATP). Peak areas were calculated in a time domain fitting procedure in which a sum of exponentially damped sinusoids is fitted to the time signal. Peak areas were corrected for saturation effects by determining T1 values of the respective metabolites (18).
Data analysis. To determine the relation between the high-energy phosphate composition of the isolated liver and the amount of hepatocellular graft damage after transplantation, simple regression analysis of the peak area ratios was used on serum concentrations of aspartate aminotransferase (AST), alanine aminotransferase (ALT), and γ-glutamyl transpeptidase at days 1, 2, 3, and 7 after transplantation. To determine the relation between the high-energy phosphate status of the isolated liver and the posttransplant liver metabolic function, regression analysis of the calculated peak area ratios was used on serum bilirubin, daily bile production, plasma fibrinogen, prothrombin time, and antithrombin III (ATIII) levels at days 1, 2, 3, and 7 after transplantation. To determine the effect of cold ischemia time (CIT) on the high-energy phosphate composition, CIT was also included in the regression analysis. The significance of the presence of PCr and β-ATP was assessed nonparametrically using the Wilcoxon rank-sum test. P-values lower than 0.05 were considered significant. In all analyses, missing clinical or spectroscopical data led to exclusion of the specific case from that specific data analysis.
RESULTS
Spectra. In all cases, spectra with excellent signal to noise ratios could be obtained. A representative spectrum is shown in Figure 1. In five livers, the Pi peak area could not be determined due to major overlap with the Pi signal from the UW inorganic phosphate. Similarly, in six cases, the PDE peak area could not be determined. The distribution of the peak area ratios of PME to NAD(H), Pi to NAD(H), and PDE to NAD(H) is given in Table 1. PCr could be detected in 34 of the 40 livers. β-ATP was detectable in 11 of the 40 livers. Sequential spectra were obtained in five livers (a representative case is shown in Fig. 2). It is clearly visible that with lengthening of the CIT, ATP disappeared, PME decreased, and Pi levels increased. The NAD(H) peak was consistently present in all livers, and its peak area hardly changed over time (Fig. 2).
Posttransplantation liver tests. The liver test results from the first postoperative week in the 40 recipients are listed in Table 2. The transaminase levels were highest at day 2. Bilirubin was still rising at day 7; however, protein synthesis, as measured by ATIII and prothrombin time, was improving at the end of the first week.
Correlations. Simple regression analysis showed a positive correlation between the PME to NAD(H) ratio and fibrinogen concentrations on days 1, 2, and 3 (P<0.001, P<0.01, and P<0.01, respectively). AST, ALT, γ-glutamyl transpeptidase, bilirubin, ATIII concentrations, daily bile production, and prothrombin time were not correlated with any of the peak area ratios. Livers exhibiting the PCr peak showed significantly (P<0.02) elevated ALT on day 7 (median value, 349.5%; 95% confidence interval, 208-439) compared with livers without detectable PCr (median value, 132; 95% confidence interval, 125-196). The presence of β-ATP resonance was significantly and positively correlated with several indicators of the extent of hepatocellular graft damage and liver metabolic function, the most striking of which was the persistent correlation with postoperative bilirubin levels and prothrombin time (Table 3). A short CIT was associated with the presence of the β-ATP resonance but not with the presence of the PCr resonance.
Follow-up until almost 3 years after the last spectroscopy session showed that 12 of the 40 recipients had died (30%). None of the patients died from acute liver failure. Causes of death are shown in Table 4.
DISCUSSION
The role of adenine nucleotides in liver viability has been studied extensively with 31P-MRS in isolated perfused animal livers, either with or without subsequent transplantation (5-12, 19, 20). From these experiments, it appeared that assessment of the adenine nucleotide status with 31P-MRS could offer a viability test with prognostic capacity for transplantation outcome.
In this clinical study using a commercial whole-body 1.5 Tesla magnet, a number of peak resonances could be evaluated for their capacity to predict postoperative metabolic function and hepatocellular damage. The significance of the spectral peaks and their relationships with transplantation outcome are discussed below.
The PME peak is a multicomponent peak that is considered to represent glycolytic intermediates and precursors of membrane biosynthesis (21, 22). The lower PME signal, which was observed with longer CIT, is due to a decrease in the AMP fraction of the PME peak as cold ischemia induces dephosphorylation of AMP. This effect is well known and has also been observed in human donor kidneys (13). In the individual donor liver, the PME resonance may therefore be an indicator for the length of the CIT. The clinical importance of such a single parameter, however, appears to be limited. Although a significant positive correlation between PME and postoperative fibrinogen was found, it appears that PME in human livers is not the prognostic indicator for postoperative organ function, as has been shown in human kidneys (13). We could not confirm a significant relationship between PME and CIT among different livers. This also suggests that CIT is not the only factor influencing PME resonance, which is in agreement with evidence that PME may also depend on the donor's nutritional status (23).
The increase of the free inorganic phosphate fraction during CIT can be attributed to cold ischemia-induced dephosphorylation of ATP, ADP, and AMP and is clearly visible in the sequential spectral plots (Fig. 2). An explantation for the absence of a correlation between hepatic inorganic phosphate and postoperative hepatocellular damage and metabolic function may be that experiments at 1.5 Tesla do not discriminate between cytosolic Pi and intramitochondrial Pi. Because the action of Pi in these two cellular compartments on the enzymes they contain (the glycolytic enzymes in the cytosol and the terminal respiratory chain in the mitochondria) may be completely different (19, 20), the Pi peak, with its heterogeneous contents observed in our measurements, is not suited for clinical viability research.
The PDE peak is a broad peak consisting mainly of glycerophosphocholine and glycerophosphoethanolamine resonances, in addition to resonances from bilayer and nonbilayer phospholipids (24). Glycerophosphocholine and glycerophosphoethanolamine have been reported to be elevated in membrane breakdown processes (21, 22, 24). Furthermore, the PDE peak has been reported to be dependent n liver glycogen content and the nutritional status of the donor (6, 23). Thus, the PDE peak appeared to be a useful tool for viability testing. Moreover, previous experiments on isolated human donor livers indeed indicated a potential role for the PDE peak as an indicator for cumulative organ damage (14). However, as the present study did not show any correlation with liver metabolic function or hepatocellular damage, the role of PDE for viability testing appears to be limited, and its significance as an indicator for cumulative organ damage remains unclear. As with PME, the PDE peak area may also dependen on nonviability-related factors, yet unknown.
PCr has been reported to be spectroscopically undetectable in the human liver (8, 21). Therefore, it was surprising to find that in the majority of cases of our study, a small but distinct resonance at the PCr site was present (Fig. 1). To our knowledge, this may be the first 31P-NMR spectroscopic demonstration of PCr in the human liver, although its possible detection with magnetic resonance spectroscopy has been speculated upon (25). PCr resonance most likely originates from the sinusoidal endothelium, since these cells contain creatine kinase-BB, the enzyme that catalyzes the reaction: creatine + ATP lrarr PCr + ADP + H+. A small part of PCr resonance might originate from the CKm fraction in the hepatocyte mitochondria (26). Because damaged sinusoidal cells release the enzyme into the perfusate, creatine kinase has gained particular attention as a marker of damage to the sinusoidal epithelium in liver transplantation (25, 27, 28). Others have questioned the value of such a test for clinical liver transplantation (29, 30). Although we could not demonstrate creatine kinase itself spectroscopically, the presence of its reaction product, PCr, correlated with a significantly lower ALT on day 7. However, in the absence of similar changes of other markers for parenchymal damage, this finding is of doubtful clinical significance. It has been suggested that PCr probably originates from smooth muscle cells of the large hepatic blood vessels. As smooth muscle cells in the liver are present only in the walls of the hepatic arteries, which form only a fraction of the liver volume, a minor contribution of these structures to the overall PCr signal cannot be excluded. On the other hand, one would expect that all livers would have shown PCr resonance, which was not the case. The true meaning of the presence of PCr resonance is not yet known. Further studies may provide more insight into the metabolic role of PCr during cold ischemia and its potential to act as an ATP buffer during ischemia (11).
The NAD(H) peak is a broad multicomponent peak consisting of NAD+/NAD(H) and NADP+/NADP(H) as well as contributions from α-ADP and α-ATP. The NAD(H) peak was consistently present in all livers, and its peak area hardly changed over time (Fig. 2). Since oxidized NAD+ cannot be distinguished in the magnetic resonance spectrum from reduced NAD(H) or from NADP and NADP(H), it is not completely clear whether the NAD(H) peak indicates the presence of intact mitochondria. Nevertheless, in the time span of safe clinical preservation (less than 20 hr), this peak remains stable, probably because the terminal respiratory chain is blocked by the absence of oxygen to act as H+ acceptor. After more prolonged cold ischemia, NAD(H) decreases, which may reflect decomposition of mitochondria (10). Both effects are visible in the sequential spectra (Fig. 2). The stability of the peak during preservation makes it useful as a reference peak against which to measure other signal intensities (9, 12, 17). At the same time, because of its stability, it is not suited as an indicator of viability. In contrast to human donor kidneys, where the presence of NAD(H) correlated with good posttransplant organ function (13), this correlation was not found in human donor livers.
ATP, when present, could be demonstrated in the spectra in only very small amounts, as compared with in vivo spectra (31). This is not surprising since (both warm and cold) ischemia during organ procurement is known to cause a rapid decrease of ATP due to dephosphorylation (2, 6). During cold storage, the remaining portion of the ATP also decreases, albeit in a much slower way. In one of the sequential measurements on the remaining part of a reduced size liver, ATP could no longer be detected after 17 hr of cold ischemia (Fig. 2). In livers examined after a very short CIT, a fair amount of ATP could be detected (Fig. 1). The ATP presence was of prognostic significance for prothrombin time, bilirubin, fibrinogen, and ATIII levels after transplantation (Table 3). Therefore, it appears that the presence of ATP during cold storage predicts a better liver metabolic capacity after transplantation. This is in agreement with the findings in human donor kidneys (13). However, ATP absence cannot be regarded as a specific predictor of early graft dysfunction, since none of the 29 livers without detectable ATP demonstrated primary nonfunction.
The presence of ATP was faintly linked to the degree of hepatocellular damage as indicated by the AST levels at day 3. However, this was not paralleled by ALT levels, which makes this correlation of minor clinical significance. Therefore, the degree of hepatocellular damage is not determined by pretransplant ATP presence, which suggests that hepatocellular damage is also determined by other, probably reperfusion-related, factors.
The importance of ATP for liver viability has been addressed by HPLC studies on biopsied human donor livers at the end of cold ischemia. Kamiike and others found that transplantation outcome is independent of the ATP level itself (1, 4), whereas Lanir et al. demonstrated a positive correlation between high ATP content and good posttransplant outcome (3). Our noninvasive magnetic resonance study on whole organs demonstrates that it was the presence of ATP that appeared to be important. If the presence of ATP signifies intact mitochondria capable of maintaining low ATP levels, it may mean that the presence of ATP also indicates the regeneration capacity for ATP after transplantation. Since it has been well established that the regeneration capacity for ATP determines liver viability (1, 3, 32), this could be an explanation for the prognostic significance of ATP for liver metabolic function found in this study. Nevertheless, as ATP levels also correlated significantly with CIT, the deterioration of both liver viability and ATP levels may be unrelated.
A potential source of error in this study was the clinical setting, which is essentially different from in vitro animal experiments under fully controlled laboratory conditions. Recipient conditions were not taken into account. Furthermore, the recipient data not only reflect the degree of (pre-) preservation injury, they also include additional effects of reperfusion injury as described by Clavien et al. (33). Such variables cannot be graded exactly, yet they may have interfered with the outcome of our study. Also, although our recipient population contained several cases with initial poor graft function, there was no proven case of primary nonfunction. Therefore, this study was unable to identify a spectrum that was prognostic for primary nonfunction. Other potential sources of error are the curve-fitting procedures used for spectral quantification. Finally, the nondetectability of PCr and ATP may be due to a certain threshold detection level of these compounds with the equipment we used. We estimate the sensitivity of our clinical magnetic resonance setup to detect ATP to be about 0.5 mmol/L, although we did not biopsy the livers to obtain biochemical proof.
In conclusion, the results of the study indicate that the individual PME, Pi, PDE, and NAD(H) peaks are not prognostic for postoperative hepatocellular damage or liver metabolic capacity. The presence of ATP, however, predicts a significantly better metabolic capacity to eliminate bilirubin, to synthesize fibrinogen and ATIII, and to maintain a better prothrombin time after transplantation. An exact causal explantation for this finding remains to be determined. Furthermore, this study may be the first 31P-MRS demonstration in the human liver of PCr, which presumably originates from the sinusoidal lining cells. Although the presence of PCr was not significantly associated with liver viability, its role in liver metabolism during cold storage is worth further study. In the clinical setting described in this article, metabolic assessment using 31P-MRS did not result in a reliable noninvasive test to predict primary graft dysfunction.
Footnotes
Abbreviations: ADP, adenine diphosphate; ALT, alanine aminotransferase; AMP, adenine monophosphate; AST, aspartate aminotransferase; ATIII, antithrombin III; ATP, adenine triphosphate; CIT, cold ischemia time; HPLC, high-performance liquid chromatography; NAD(H), nicotine adenine dinucleotide; PCr, phosphocreatine; PDE, phosphodiesters; Pi, inorganic phosphate; PME, phosphomonoesters; 31P-MRS, phosphorus-31 magnetic resonance spectroscopy; UW, University of Wisconsin solution.
REFERENCES
1. Kamiike W, Burdeiski M, Steinhoff G, Ringe B, Lauchart W, Pichlmayr R. Adenine nucleotide metabolism and its relation to organ viability in human liver transplantation. Transplantation 1988; 45: 138.
2. Harvey PRC, Iu S, McKeown CMB, Petrunka CN, Ilson RG, Strasberg SM. Adenine nucleotide tissue concentrations and liver allograft viability after cold preservation and warm ischemia. Transplantation 1988; 45: 1016.
3. Lanir A, Jenkins RL, Caldwell C, Lee RGL, Khettry U, Clouse ME. Hepatic transplantation survival: correlation with adenine nucleotide level in donor liver. Hepatology 1988; 8: 471.
4. Hamamoto I, Takaya S, Todo S, et al. Can adenine nucleotides predict primary nonfunction of the human liver homograft? Transplant Proc 1993; 25: 3036.
5. Clouse ME, Lanir A, Jones DA, Khettry U, Lee RGL. 31-P nuclear magnetic resonance evaluation of intermittent perfusion as a method of liver preservation. Transplantation 1988; 45: 1137.
6. Lanir A, Clouse ME, Lee RGL. Liver preservation for transplant, evaluation of hepatic energy metabolism by 31 P NMR. Transplantation 1987; 43: 786.
7. Fuller BJ, Busza AL, Proctor E. Possible resuscitation of liver function by hypothermic reperfusion in vitro after prolonged (24-hour) cold preservation: a 31 P NMR study. Transplantation 1990; 50: 511.
8. Busza AL, Proctor E, Myles M, Gardian DG, Hobbs KEF. Control of pH during hypothermic liver storage. Transplantation 1987; 45: 239.
9. Busza AL, Proctor E, Fuller BJ. Biochemical consequences of reflushing hypothermically stored liver with fresh cold perfusate. NMR Biomed 1989; 2: 115.
10. Matsunami H, Hirose H, Onitsuka A, et al. 31 P-magnetic resonance spectroscopy in evaluating hepatic function: a possible application in donor assessment prior to liver transplantation. Transplant Proc 1990; 22: 2146.
11. Miller K, Halow J, Koretsky AP. Phosphocreatine protects transgenic mouse liver expressing creatinine kinase from hypoxia and ischemia. Am J Physiol 1993; 265: C1544.
12. Busza AL, Fuller BJ, Proctor E. Evaluation of cold reperfusion as an indicator of viability in stored organs: a 31 P NMR study in rat liver. Cryobiology 1994; 31: 26.
13. Bretan PN, Baldwin N, Novicck AC, et al. Pretransplant assessment of renal viability by phosphorus-31 magnetic resonance spectroscopy. Transplantation 1989; 48: 48.
14. Wolf RFE, Kamman RL, Mooyaart EL, Haagsma EB, Bleichrodt RP, Slooff MJH. 31 P magnetic resonance spectroscopy of the isolated human donor liver: feasibility in routine clinical practice and preliminary findings. Transplantation 1993; 55: 949.
15. Calne R. Orthotopic liver transplantation. In: Blumgarth LH, ed. Surgery of the liver and biliary tract. Edinburgh: Churchill & Livingstone, 1994: 1811.
16. Wolf RFE, van der Hoeven JAB, Kamman RL, et al. Tissue pH in cold-stored human donor livers preserved in University of Wisconsin solution. Transplantation 1996; 61: 66.
17. Busza AL, Fuller BJ, Lockett CJ, Proctor E. Maintenance of liver adenine nucleotides during cold ischaemia: the value of a high-pH, high-pk flush. Transplantation 1992; 54: 562.
18. Wolf RFE, Slooff MJH, Go KG, Kamman RL. Changes in
31P-relaxation times during organ preservation: observations on cold stored human donor livers. Magn Reson Imaging (in press).
19. Tanaka A, Chance B, Quistorff B. A possible role of inorganic phosphate as a regulator of oxidative phosphorylation in combined area synthesis and gluconeogenesis in perfused rat liver. J Biol Chem 1989; 264: 10034.
20. Vanstapel F, Waebens M, Vanhecke P, Decanniere C, Staimans W. The cytosolic concentration of phosphate determines the maximal rate of glycogenolysis in perfused rat liver. Biochem J 1990; 266: 207.
21. Oberhaensli R, Rajagopalan B, Galloway GJ, Taylor DJ, Radda GK. Study of human liver disease with P31 magnetic resonance spectroscopy. Gut 1990; 31: 463.
22. Iles RA, Cox IJ, Bell JD, Dubowitz LMS, Cowan F, Bryant DJ. 31 P magnetic resonance spectroscopy of the human paediatric liver. NMR Biomed 1990; 3: 90.
23. Schilling VA, Gewiese B, Stiller D, Romer T, Wolf KJ. Influence of nutritional status on the 31 P-NMR spectrum of the healthy liver. Fortschr Rontgenstr 1990; 153: 369.
24. Bell JD, Cox IJ, Sargentoni J, et al. A 31P and 1H-NMR investigation in vitro of normal and abnormal human liver. Biochim Biophys Acta 1993; 1225: 71.
25. Malninck SDH, Bass DD, Kaye AM. Creatinine kinase BB: a response marker in liver and other organs. Hepatology 1994; 19: 261.
26. Rauen U, Erhard J, Kuhnhenrich P, et al. Non-parenchymal cell and hepatocellular injury to human liver grafts assessed by enzyme-released into the perfusate. Langenbecks Arch Chir 1994; 379: 241.
27. Vaubourdolle M, Chazouilleres O, Poupon R. Creatinine kinase-BB: a marker of liver sinusoidal damage in ischemia-reperfusion. Hepatology 1993; 17: 423.
28. Chazouilleres O, Vaubourdolle M, Fourel V, et al. Creatinine kinase-BB and soluble thrombomodulin: new markers for sinusoidal damage in human liver transplantation. Ann NY Acad Sci 1994; 723: 466.
29. Karayalcin K, Harrison JD, Attard A. Can effluent hyaluronic acid or creatinine kinase predict sinusoidal injury severity after cold ischemia? Transplantation 1993; 56: 1336.
30. Devlin J, Dunne JB, Sherwood RA, et al. Relationship between early graft viability and enzyme activities in effluent preservation solution. Transplantation 1995; 60: 627.
31. Buchli R, Meier D, Martin E, Boesiger P. Assessment of absolute metabolise concentrations in human tissue by 31 P MRS in vivo. Part 2: muscle, liver, kidney. Magn Reson Med 1994; 32: 453.
32. Bowers JL, Teramoto K, Khettry U, Clouse ME. 31 P-NMR assessment of orthotopic rat liver transplant viability. Transplantation 1992; 54: 604.
33. Clavien PA, Harvey PRC, Strasberg SM. Preservation and reperfusion injuries in liver allografts. Transplantation 1992; 53: 957.
