1
H NMR Metabonomic Analysis in Renal Cell Carcinoma: a Possible Diagnostic Tool
Athina N. Zira,†,‡ Stamatios E. Theocharis,† Dionisios Mitropoulos,§ Vasilios Migdalis,§ and Emmanuel Mikros*,‡ Department of Forensic Medicine and Toxicology, Medical School, National and Kapodistrian University of Athens, Athens, Greece, Department of Pharmaceutical Chemistry, School of Pharmacy, National and Kapodistrian University of Athens, Athens, Greece, and First Department of Urology, Laikon General Hospital, Medical School, National and Kapodistrian University of Athens, Athens, Greece Received March 12, 2010 1 H NMR based metabonomic approach was applied in order to monitor the alterations of plasma metabolic profile in Renal Cell Carcinoma (RCC) patients and controls. 1H NMR spectra of plasma samples from 32 RCC patients and 13 controls (patients exhibiting benign urologic disease) were recorded and analyzed using multivariate statistical techniques. Alterations in the levels of LDL/VLDL, NAC, lactate, and choline were observed between RCC patients and controls discriminating these groups in Principal Component Analysis (PCA) plots. Post OSC PLS-DA presented a satisfactory clustering between T1 with T3 RCC patients. Decrease in plasma lipid concentrations in RCC patients was verified using conventional clinical chemistry analysis. The results suggest that combination of 1H NMR spectroscopy with PCA has potential in cancer diagnosis; however, a limitation of the method to monitor RCC is that major biomarkers revealed (lipoproteins and choline) in this metabolic profile are not unique to RCC but may be the result of the presence of any malignancy.
Keywords: RCC • Metabonomics • NMR • PCA • PLS-DA • OSC
Introduction Renal cell carcinoma (RCC) is a relatively uncommon tumor with an annual incidence of approximately 10 cases per 100 000 population in USA. It is an heterogeneous disease classified into various subtypes based on morphological and genetic features. Additionally, the tumor is commonly large at presentation and symptoms may not occur until relatively late in the disease progression. The prognosis for RCC patient is also variable and patients with metastatic or recurrent RCC usually present poor prognosis with rare long-term survival.1-3 The identification and validation of RCC specific biomarkers are urgently needed for disease early detection, prognosis, and patients’ management. Specific markers are also needed for the assessment of the response to nonsurgical novel therapeutic strategies and the early detection of possible recurrences. Various studies have been undertaken to search for candidate protein markers in RCC that might be implemented for early diagnosis, staging, detection of metastasis, prediction of patients outcome as well as for the monitoring of the response to treatment.4,5 * CORRESPONDING AUTHOR FOOTNOTE Mikros Emmanuel, tel +30 2107274813, fax +30 2107274747
[email protected]. † Department of Forensic Medicine and Toxicology, National and Kapodistrian University of Athens. ‡ Department of Pharmaceutical Chemistry, National and Kapodistrian University of Athens. § First Department of Urology, National and Kapodistrian University of Athens.
4038 Journal of Proteome Research 2010, 9, 4038–4044 Published on Web 06/09/2010
Nuclear magnetic resonance spectroscopy (NMR) of biofluids allows the simultaneous measurement of endogenous metabolites, reflecting the biochemical fingerprint of the organism. The NMR-based metabonomics approach6 combines data obtained from the spectroscopic and multivariate statistical analyses to cluster the samples according to their common spectroscopic/ metabolic characteristics. Initially applied in experimental animal models of toxicity,7 the study of endogenous metabolites alterations in human biofluids (e.g., plasma, serum, urine), intact tissues, or tissue extracts has expanded during the past decade, in order to develop models for diagnosis of major diseases like coronary heart disease8 and cancer.9-12 NMRbased metabonomics presents a holistic picture of the metabolites’ alterations in the organism providing disease biomarkers that could revolutionize cancer treatment if early stage disease can be detected. RCC diagnosis has been approached using metabolic profiling of urine samples on the basis of Mass Spectrometry measurements,13-17 while NMR based studies remain limited.16,17 In the present study, the metabonomic approach has been applied to monitor the alterations in the metabolic profile in plasma samples of patients with RCC. To this end, 1H NMR spectra of plasma samples from RCC patients were recorded and analyzed using multivariate analysis techniques such as Principal Component Analysis (PCA) and PLS-DA and compared to those obtained from patients exhibiting benign urological diseases. Differences in metabolic profiles are discussed on the basis of variation of diverse metabolites. 10.1021/pr100226m
2010 American Chemical Society
1
H NMR Metabonomic Analysis in Renal Cell Carcinoma
research articles
Figure 1. Representative of 400 MHz 1H NMR CPMG spectrum (δ 4.5-0.5) of blood plasma from (A) control and (B) T3 samples.
Materials and Methods Collections and Preparation of Specimens. The sample set consists of blood plasma collected from 32 RCC patients and 13 patients with benign urological problems (controls). The mean age of the RCC patients cohort was 63.33 ( 12.15 years (range, 45-80 years) and of controls cohort was 64.1 ( 7.74 years (range, 56-74 years). RCC tumor stage and classification were established according to the 2002 TNM Staging System.18 Histopathologically, tumors were classified as clear cell (N ) 22), chromophobe (N ) 4), papillary (N ) 5), and mucinus tubular spindle cell (N ) 1) type. According to their size, tumors were characterized as T1 (N ) 12), T2 (N ) 8), and T3 (N ) 12). Tumors were further characterized according to Furhman as grade I (N ) 1), grade II (N ) 10), grade III (N ) 18), and grade IV (N ) 3).19 Controls were classified as suffering from bladder outlet obstruction (N ) 13). None of the patients had received antitumor therapy. The study protocol was approved by the Hospital Ethics Committee, and all subjects studied gave their informed consent. Fasting blood samples (10 mL) were drawn under sterile conditions from an antecubital vein of all patients and transferred immediately to EDTA coated tubes the morning before surgery. Plasma was separated from whole blood by centrifugation at 1000g for 10 min at room temperature, and then the samples were frozen at -80 °C until assayed. 1 H NMR Spectroscopic Analysis of Blood Plasma. Spectra were acquired using a Bruker DRX-400 Avance Spectrometer. Blood plasma samples were thawed immediately before use and 400 µL of each was diluted by 99.98% D2O (100 µL) in 5 mm NMR tubes. 1H chemical shifts were measured relative to the internal standard 3-(trimethylsilyl)-propionic acid-d4, sodium salt (TSP, 50 µL) at δ 0.00. The CPMG (Carr-PurcellMeiboom-Gill) spin-echo pulse sequence was used with presaturation of the water signal and a fixed relaxation delay (2nτ) of 87.8 ms. Typically, 128 scans were collected into 64K data points over a spectral width of 5593 Hz with a relaxation delay of 2 s and an acquisition time of 5.8 s. A line broadening of 1.0 Hz was applied to all 1H NMR spectra prior to Fourier transform.
Sampling and Analysis of Blood Plasma Lipids. Plasma analyses for cholesterol, triglycerides, and HDL cholesterol were performed using the Siemens Advia 1650 Clinical Chemistry System (Siemens Healthcare Diagnostics, Erlangen, Germany). LDL cholesterol values were determined by calculation, known as “Friedewald’s formula”, using measured values for total cholesterol, HDL cholesterol, and triglycerides. VLDL cholesterol values were determined by dividing five times the triglycerides values. The intra-assay precision coefficient of variation for cholesterol, triglycerides, and HDL cholesterol was less than 1.4%, 1.6%, and 1.2%, while the interassay one was less than 2.1%, 2.4%, and 2.5%, respectively. Internal quality control of cholesterol, triglycerides, and HDL cholesterol was carried out according to the laboratory protocol. Data Reduction of NMR Spectra and Statistical Analysis. Following phase and baseline correction, each spectrum was segmented into chemical shift regions of 0.04 ppm width using the software package AMIX (Analysis of Mixtures, version 2.7, Bruker Analytische Messtechnik, Karlsruhe, Germany). The region δ 4.50-5.1, which included the water resonance, was removed from all spectra prior to statistical analysis. The regions δ 2.50-2.74, 3.04-3.30, and 3.50-3.66 containing the free EDTA and EDTA metal complexes were also excluded.20 The remaining spectral regions were integrated and scaled to the total intensity of the spectrum. Data were further subjected to PCA and Partial Least Squares-Discriminant Analysis (PLS-DA)21 using SIMCA-P 10.5 software package (Umetrics, Umeå, Sweden). Prior to PCA, data were centered and Pareto scaled. PCA is a multivariate projection method useful in classifying samples according to their common spectral characteristics. A plot of the first two principal components (scores plot) provides the most efficient 2D representation of the information contained in the data set. In addition, a corresponding “loading plot or coefficient histograms” provides information on the variables that influence any observed clustering of samples. The overall quality of the models was judged by the cumulative R2, and the predictive ability by cumulative Q2 extracted according to the internal cross-validation default method of SIMCA-P software. PLS-DA Journal of Proteome Research • Vol. 9, No. 8, 2010 4039
research articles
Zira et al.
is a supervised method used when clusters are not distinctly separated in the scores plot and groups overlap;21 PLS attempts to derive latent variables, analogous to PCs, which maximize the covariation between the measured data (X) and the response variable (Y) regressed against. PLS-DA was applied to the groups of samples using NMR spectral data as X matrix and group membership as the response matrix Y. PLS-DA also provides a plot of variance importance, which presents the most important variables of the separation. For further optimization of the classification of samples, orthogonal signal correction (OSC) was applied to the data prior to reanalysis by PCA or PLS-DA algorithms.21,22 OSC is a data filtration method that uses information based on class membership to eliminate the contribution of systematic variation arising from features in the data that are unrelated to the class. The statistical significance of the differences observed between the average of each group concerning bucket integrals as well as lipoprotein concentrations was performed using the t test as implemented in STATISTICA (7.0) software package and significance was determined at P < 0.05.
Results 1
H NMR Spectra of Blood Plasma. A representative 400 MHz H NMR CPMG spectrum of blood plasma from controls is depicted in Figure 1. Resonances assigned to metabolites such as LDL/VLDL, lactate, alanine, pyruvate, acetate, acetoacetate, 3-hydroxybutyrate, creatine, choline, glucose, N-acetylglycoproteins (NAC), glutamine, glutamate, formate, tyrosine, phenylalanine, and histidine are visible.23 1
PCA of 1H NMR Spectra of Plasma from Cancer Patients and Controls. In a first step, a tentative PCA has been performed for all groups. Only control samples were discriminated, forming a separate group in score plot (Figure 2A). PLS-DA scores plots ameliorates the separation of controls from RCC patients, but still no separation between cases of different tumor size (T) is apparent (Figure 2B). When comparing separately controls and patients with each one of the RCC groups (controls vs T1, controls vs T2, and controls vs T3), PCA scores plots differentiated the groups and the separation was even better when PLS-DA was applied (Figure 3). Comparing controls and T1 RCC patients (Figure 3A) using PLS-DA revealed a satisfactory discrimination between the two groups (R2X ) 0.878, R2Y ) 0.964, and Q2 ) 0.905). The plots of variable coefficients revealed that the resonances responsible for the discrimination were LDL/VLDL, valine/isoleucine, lactate, alanine, NAC, glutamine, the amino acid resonances mixed with glucose (δ 3.3-3.9), and CHdCH of lipids (Figure 3B). Similar PLS-DA plots were obtained when analysis was performed for T2 and T3 was compared to control (Figure 3, panels C and E, respectively). Quality factors for those models were R2X ) 0.862, R2Y ) 0.979, and Q2 ) 0.696 for T2 and R2X ) 0.874, R2Y ) 0.913, and Q2 ) 0.845 for T3. The plot of variable coefficients attributed the discrimination to the same variables as mentioned above (Figure 3D,F). Neither PCA nor PLS-DA could cluster the RCC samples T1, T2, and T3 according to tumor size (data not shown); thus, the OSC filter was applied in order to improve the clustering, and to identify metabolites discriminating these groups. When comparing samples of T1 and T3 RCC patients, post OSC PLSDA (Figure 4A) clearly discriminated them (R2X ) 0.941, R2Y ) 0.977, and Q2 ) 0.603) and the plot of variables importance 4040
Journal of Proteome Research • Vol. 9, No. 8, 2010
Figure 2. PCA (A) and PLS-DA (B) analysis of 400 MHz 1H NMR spectra of all patients (controls vs RCC patients). Key: Controls (9), T1 (b), T2 ((), and T3 (2).
attributed the separation mainly to the resonances of LDL/ VLDL, NAC, and amino acid resonances at δ 3.3-3.9 (Figure 4B). When comparing controls with RCC patients categorized according to different stage and histological subtype, no clustering among the classes was achieved (data not shown). Table 1 summarizes the variation of the integrals of the normalized spectral regions (buckets) accounting for different plasma metabolites and lists the results from the statistical analysis (P < 0.05) for comparison. The resonances assigned to the methyl and methylene groups of LDL/VLDL were statistically significantly dropped in all RCC patients compared to controls. Lactate levels were not altered in T1 and T2 patients; however, they were decreased in T3 patients compared to controls. N-acetyl-glycoproteins (NAC) levels were decreased in T1 and T2 patients, but they were increased in T3 patients compared to controls. Choline levels (as measured from signal at δ 4.05) were significantly increased in all RCC patients compared to controls. Levels of valine, isoleucine, and glutamate were increased in RCC patients compared to controls, while alanine levels did not present significant alterations. On the other hand, glutamine levels were decreased in all RCC patients compared to controls. Glucose levels were not altered in the RCC patients compared to controls. Decreased levels of unsaturated lipids resonances (δ 5.30) were noted in RCC patients compared to controls along with a statistically significant depletion in T1 RCC patients. In the aromatic region, free amino acids (phenylalanine, tyrosine, and histidine) were not altered in different RCC stages compared to controls. However, raised levels of formic acid were noticed in T1 RCC patients
1
H NMR Metabonomic Analysis in Renal Cell Carcinoma
research articles
Figure 3. PLS-DA analysis of 400 MHz 1H NMR spectra of (A) controls compared with T1 RCC patients and (B) the corresponding variable coefficients; (C) controls compared with T2 RCC patients and (D) the corresponding variable coefficients; (E) controls compared with T3 RCC patients and (F) the corresponding variable coefficients. Key: Controls (9), T1 (b), T2 ((), T3 (2).
compared to controls, while patients with T2 and T3 RCC patients presented decreased formic acid compared to controls. To further verify the observed lipoprotein alterations, standard biochemical methods were applied. A decrease in the lipid levels was observed in agreement with the NMR analysis and the resulting concentrations are summarized in Table 2. A statistically significant decrease in the levels of HDL in T1, T2, and T3 RCC patients compared to controls was noted (p < 0.05). Furthermore, LDL cholesterol levels were also declined in T2 and T3 RCC patients compared to controls (p < 0.05). No statistically significant differences were observed when comparing patients with benign RCC and controls.
Discussion In most malignancies, early diagnosis has the potential to markedly improve patient’s survival. RCC is the most common
type of renal malignancy with patients often having advanced incurable disease at the time of diagnosis. Early diagnosis, prior to metastatic spread, can improve survival odds from less than 10% to greater than 90%; however, no biofluid screening tests exist for RCC. The development of new diagnostic methods, in easily accessible patient materials (such as blood and urine) should, therefore, be an important priority. In this study, we have applied 1H NMR based metabolic profiling combined with pattern recognition techniques of blood plasma to discriminate RCC patients from controls, demonstrating the feasibility of this method for disease diagnosis. Proteomics have been already utilized to study tissue samples from RCC patients indicating that proteins involved in glycolysis, urea cycle and propanoate, pyruvate, and arginine/proline metabolism were differentially expressed in RCC suggesting their activation in the disease.13 In the same study, urine Journal of Proteome Research • Vol. 9, No. 8, 2010 4041
research articles
Zira et al. 25
Figure 4. Post-OSC PLS-DA analysis of 400 MHz 1H NMR spectra of T1 compared with T3 RCC patients and the corresponding variable coefficients. Key: T1 (b), T3 (2). Table 1. Variation of the Integrals of the Normalized Spectral Regions (Buckets) Accounting for Different Plasma Metabolitesa chemical shift
metabolites
0.78-0.94 (m) 0.92 -1.06 (m) 1.20-1.31 1.33 (d) 4.11 (q) 1.47 (d) 1.92 (s) 2.10 (s) 2.27 (s) 2.31-2.8 (m) 2.41-2.48 (m) 4.05 5.23 (d) 5.30 6.93 7.22 (d) 7.09 7.79 (s) 7.34-7.46 (m) 8.46 (s)
LDL/VLDL CH3(CH2)n Valine/Leucine/Isoleusine (γ,δ CH3) VLDL/LDL (CH2)n Lactate Alanine Acetate NAC Acetoacetate Glutamate Glutamine Cho Glucose CHdCH (lipids) Tyrosine Histidine Phenylalanine Formic acid
T1 T2 T3
V* v V V V V* v* V v V* v
V* V V* V V* v V v V V
V* v V V V V V* v* V* v V V
a Asterik (*) represents statistically significant differences compared to control as determined by univariate statistical analysis (P < 0.05).
metabolic profiling showed that sorbitol is significantly elevated signifying that the sorbitol pathway of glucose metabolism is active in the RCC specimens. Protein profiling of urine samples has identified potential biomarkers,24 however, in a study investigating the clinical utility of Surface Enhanced Laser Desorption Ionization (SELDI) profiling of urine samples in conjunction with neural-network analysis to detect renal cancer and to identify proteins of potential use as prognostic markers, 4042
Journal of Proteome Research • Vol. 9, No. 8, 2010
but the results were not reproducible. Initial metabolic profiling analysis of urine based on a combination of three different LC and GC-MS platforms using a multivariate partial least-squares (PLS) approach for data analysis has shown the suitability of these methods for detection of RCC.14 However, a recent follow-up using a larger and more diverse patient cohort has shown that there exist substantial sources of variation of the urinary metabolomic profile, although group variation is sufficient to yield viable biomarkers.15 In this aspect it should be noted that in a recent study the use of urine metabolome has been contested because of the large intraindividual variability in both normal and animal control populations looking at specific metabolites;26 however, the possibility of eliminating the daily noise by multiple sample urine collection could give new perspective in urine based metabonomic studies.27 The use of 1H NMR spectroscopy to characterize biochemical changes in various toxic and disease states has been extensively reported. Specifically in cancer diagnosis, 1H and 31P NMR spectra of tissue extracts and 31P NMR spectra obtained in vivo have been used in distinguishing various types of tumors.28,29 More recently, the development of High Resolution MAS techniques allowed the study of whole tissue samples by means of very high quality NMR spectra.16,30 Cancer diagnosis based on NMR spectra of biological fluids remains controversial. Initial studies on serum lipid NMR signals failed as a tumor marker suitable for cancer screening due to low specificity ( ref 31 and references therein). Nonetheless, the use of updated NMR techniques and metabonomic statistical methods gave the possibility to reduce the complexity of data and to use successfully data obtained from patients’ blood in order to detect biomarkers of disease. Thus, metabonomic phenotyping models using human serum and 1H NMR have been used to differentiate metabolic profiles of persons with normal coronary arteries from those with coronary disease.8 The method could have great potential in cancer research and application of NMR metabonomics to the analysis of serum for markers indicative of cancer has been explored for early breast cancer detection by Odunsi et al.9 They illustrated that PCA of NMR spectral data obtained from serum of patients could not only detect epithelial ovarian cancer (EOC) but could distinguish EOC patient serum from that obtained from patients with benign ovarian cysts as well. In another study, the applicability of NMR-based metabolomics in the diagnosis of disease using serum samples of oral cancer patients has been demonstrated. Patients with oral squamous cell carcinoma showed a distinct signature of altered energy metabolism in blood serum, which includes altered lipolysis (an accumulation of ketone bodies), a distorted Krebs cycle, and amino acid catabolism.12 Our results demonstrate that the application of PCA and PLSDA on the plasma samples 1H NMR data can successfully distinguish between control and RCC patients suggesting alterations in the metabolic profile. In a further step, the application of post OSC PLS-DA discriminated T1 and T3 RCC patients. Plots of variable importance and variable coefficients along with univariate analysis revealed decreased levels of lipids (LDL/VLDL, NAC, and unsaturated lipoproteins), lactate, glutamine, and acetoacetate in RCC patients compared to controls, while choline and isoleucine levels were increased. Confirmation of the spectroscopic data depicting decrease in lipid levels was obtained by conventional lipoprotein measurement dem-
research articles
1
H NMR Metabonomic Analysis in Renal Cell Carcinoma
Table 2. Clinical Chemistry Data Concerning Blood Plasma Lipid Profile of the Different Groups
Cholesterol (mg/dL) Triglycerides (mg/dL) HDL-C (mg/dL) VLDL-C (mg/dL) LDL-C (mg/dL) a
controls
T1
T2
T3
211.9 ( 29.6 131.4 ( 47.7 55.8 ( 8.2 26.3 ( 9.5 129.9 ( 22.4
193.9 ( 31.3 109.9 ( 27.6 46.3 ( 9.6a 22.0 ( 5.5 125.7 ( 27.8
169.6 ( 35.0 126.9 ( 56.6 43.9 ( 7.1a 25.4 ( 11.3 100.3 ( 26.4a a
167.7 ( 30.8a 122.7 ( 27.4 33.7 ( 6.5a 24.5 ( 5.5 109.5 ( 27.3a
Statistically significant compared to control (P < 0.05).
onstrating the power and credibility of 1H NMR based metabonomics. The decrease of LDL/VLDL signals can be associated with hypolipidemia observed in various cancer types32-34 and the well described increase of lipid content of kidney specimens with RCC.35 Abnormal cholesterol metabolism and accumulation in solid tumors has been known for about a century,36 being associated with the cellular accelerated growth and division in tumor cells, requiring elevated levels of cholesterol and of cholesterol precursors. Much of the cholesterol residing in cell membranes originates from the uptake of circulating lipoproteins. Excessive uptake of lipoprotein cholesterol from serum might be the source of cholesteryl esters in RCC.37 High lipid levels in RCC kidney cortical biopsy samples have been also demonstrated by previous studies based on 1H-HRMAS and 13C NMR spectroscopy.16,38 The concomitant decrease of ketone bodies (acetate and acetoacetate), which are the final products of lipid metabolism, can be also correlated to a lesser catabolism of lipids because of the increased need for their utilization in cell proliferation. The statistically significantly decreased lipoprotein levels found in RCC patients compared to controls are in agreement with the findings of Abiaka et al. who studied incidences of gastric, colorectal, and breast cancer,39 as well as the decrease of phospholipids observed by Su ¨ lentrop et al.40 in blood plasma of RCC patients using 31P NMR spectroscopy. It has been hypothesized that tumors arising in a particular organ site impose a characteristic plasma free amino acid (PFAA) profile reflecting cancer-induced protein metabolism in tumors, skeletal muscle, and the liver in patients. Redistribution or translocation of PFAAs to support tumor protein synthesis represents an essential feature in cancer patients and reduced availability at the late state of cancer patients is related to malnutrition. The PFAA profile can be affected by the type of cancer being also different between early and late stages of cancer; therefore, it is postulated that a detailed analysis of the PFAA profile may serve as one of the biological markers for cancer patients.41 However, a number of existing studies reviewed recently showed that no consistent cancer-specific amino acid has emerged.42 In our study, RCC patients exhibited relatively decreased glutamine and increased glutamate, valine, and isoleucine, while aromatic amino acids and alanine remained unaffected. Glutamine is the most abundant amino acid in the blood, comprising 50% of the whole body pool of free amino acids. It is a necessary substrate for nucleotide synthesis in most dividing cells and an important respiratory fuel. The observed decrease of glutamine and increase of glutamate is consistent with enhanced glutaminase activity and is in agreement with previously reported plasma glutamine reduction in breast, gastrointestinal, and head and neck cancers.41 Additionally, it has been postulated that increasing muscle protein breakdown, which provides substrates for enhanced Branched Chain Amino
Acids oxidation in muscle such as valine and isoleucine, is the primary mechanism accounting for higher BCAA levels during the early stages of cancer.42 Recently, a study similar to ours has been conducted by Gao et al., using serum samples obtained from patients with different stage of RCC and patients before and after nephrectomy.17 PCA revealed a clear discrimination between RCC and healthy serum samples. However, in contrast to our results, they concluded an increase in lipoprotein NMR signals in patients samples which was not further verified. This controversy could be attributed to differences in metabolic phenotypes between Asian and Caucasian populations as demonstrated in recent metabonomic studies,43 and make obvious the need of further investigation. In the current study, the 1H NMR based metabonomic approach was successfully applied in the discrimination of plasma samples from RCC patients and patients exhibiting benign urological diseases. The comparison between RCC samples and control revealed that adequate classification in PCA, as well as discrimination between T1 and T3, is possible using the supervised OSC PLSDA method. On the basis of these findings, the method could be used for the screening of persons presenting predisposition for the disease. When comparing RCC patients based on the TNM, no clustering among the classes was achieved. Metabolic profiling is still at the early phase of discovery regarding application to cancer risk assessment, diagnosis, and treatment. In this study, we show that 1H NMR of plasma samples in combination with PCA might be a promising tool in RCC diagnostics. It is possible the observed blood metabolic alterations noted in RCC may be related to any malignancy, given the known differences in lipid content in cancer states; it is, however, important to notice that comparison of the profile observed for RCC patients with the recently reported for oral cancer exhibiting distinct metabolic signature, suggests the feasibility of the development of cancer type specific diagnosis. Variations in the metabolic profile of different populations could be a further difficulty in the application of NMR based diagnostic techniques. It is evident that further studies are necessary requiring larger cohorts of patients and intercomparisons between populations as well as implication of other high-throughput methods combining different “omics” technologies in oncology.
References (1) Furniss, D.; Harnden, P.; Ali, N.; Royston, P.; Eisen, T.; Oliver, R. T.; Hancock, B. W. On behalf of the National Cancer Research Institute Renal Clinical Studies Group. Cancer Treat. Rev. 2008, 34, 407–426. (2) Mancini, V.; Battaglia, M.; Ditonno, P.; Palazzo, S.; Lastilla, G.; Montironi, R.; Bettocchi, C.; Cavalcanti, E.; Ranieri, E.; Selvaggi, F. P. Current insights in renal cancer pathology. Urol. Oncol.: Semin. Orig. Invest. 2008, 26, 225–238.
Journal of Proteome Research • Vol. 9, No. 8, 2010 4043
research articles (3) Figlin, R. A. Renal cell carcinoma: management of advanced disease. J. Urol. 1999, 161, 381–387. (4) Nogueira, M.; Kim, H. L. Molecular markers for predicting prognosis in renal cell carcinoma. Urol. Oncol.: Semin. Orig. Invest. 2008, 26, 113–124. (5) Crispen, P. L.; Boorjian, S. A.; Loshe, C. M.; Leibovich, B. C.; Kwon, E. D. Predicting disease progression after nephrectomy for localized renal cell carcinoma: the utility of prognostic models and molecular biomarkers. Cancer 2008, 113, 450–460. (6) Nicholson, J. K.; Lindon, J. C.; 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– 1189. (7) Lindon, J. C.; Holmes, E.; Nicholson, J. K. Toxicological applications of magnetic resonance. Prog. Nucl. Magn. Reson. Spectrosc. 2004, 45, 109–143. (8) Brindle, J. T.; Antti, H.; Holmes, E.; Tranter, G.; Nicholson, J. K.; Bethell, H. W.; Clarke, S.; Schofield, P. M.; McKilligin, E.; Mosedale, D. E.; Grainger, D. J. Rapid and noninvasive diagnosis of the presence and severity of coronary heart disease using 1H-NMRbased metabonomics. Nat. Med. 2002, 8, 1439–1444. (9) Odunsi, K.; Wollman, R. M.; Ambrosone, C. B.; Hutson, A.; McCann, S. E.; Tammela, J.; Geisler, J. P.; Miller, G.; Sellers, T.; Cliby, W.; Qian, F.; Keitz, B.; Intengan, M.; Lele, S.; Alderfer, J. L. Detection of epithelial ovarian cancer using 1H-NMR-based metabonomics. Int. J. Cancer. 2005, 113, 782–8. (10) Beckonert, O.; Monnerjahn, J.; Bonk, U.; Leibfritz, D. Visualizing metabolic changes in breast-cancer tissue using 1H-NMR spectroscopy and self-organizing maps. NMR Biomed. 2003, 16, 1–11. (11) Beger, R. D.; Schnackenberg, L. K.; Holland, R. D.; Li, D.; Dragan, Y. Metabonomic models of human pancreatic cancer using 1D proton NMR spectra of lipids in plasma. Metabolomics 2006, 2, 125–134. (12) Tiziani, S.; Lopes, V.; Gu ¨ nther, U. L. Early stage diagnosis of oral cancer using 1H NMR-based metabolomics. Neoplasia 2009, 11, 269–276. (13) Kind, T.; Tolstikov, V.; Fiehn, O.; Weiss, R. H. A comprehensive urinary metabolomic approach for identifying kidney cancer. Anal. Biochem. 2007, 363, 185–195. (14) Perroud, B.; Lee, J.; Valkova, N.; Dhirapong, A.; Lin, P.-Y.; Fiehn, O.; Ku ¨ ltz, D.; Weiss, R. H. Pathway analysis of kidney cancer using proteomics and metabolic profiling. Mol. Cancer 2006, 5:64, 1– 17. (15) Kim, S. K.; Aronov, P.; Zakharkin, S. O.; Anderson, D.; Perroud, B.; Thompson, I. M.; Weiss, R. H. Urine metabolomics analysis for kidney cancer detection and biomarker discovery. Mol. Cell. Proteomics 2009, 8, 558–570. (16) Tate, A. R.; Foxall, P. J. D.; Holmes, E.; Moka, D.; Spraul, M.; Nicholson, J. K.; Lindon, J. C. Distinction between normal and renal cell carcinoma kidney cortical biopsy samples using pattern recognition of 1H magic angle spinning (MAS) NMR spectra. NMR Biomed. 2000, 13, 64–71. (17) Gao, H.; Dong, B.; Liu, X.; Xuan, H.; Huang, Y.; Lin, D. Metabonomic profiling of renal cell carcinoma: high-resolution proton nuclear magnetic resonance spectroscopy of human serum with multivariate data analysis. Anal. Chim. Acta 2008, 624, 269–277. (18) Sobin, L. H.; Wittekind, C. TNM classifications: kidney. In TNM Classifications of Malignant Tumours, 6th ed.; Wiley: New York, 2003; pp 193-195. (19) Fuhrman, S. A.; Lasky, L. C.; Limas, C. Prognostic significance of morphologic parameters in renal cell carcinoma. Am. J. Surg. Pathol. 1982, 6, 655–663. (20) Barton, R. H.; WatermanD.; BonnerF. W.; HolmesE.; ClarkeR.; the PROCARDIS Consortium;Nicholson, J. K.; Lindon, J. C. The influence of EDTA and citrate anticoagulant addition to human plasma on information recovery from NMR-based metabolic profiling studies. Mol. BioSyst. 2010, 6, 215–224. (21) Lindon, J. C.; Holmes, E.; Nicholson, J. K. Pattern recognition methods and applications in biomedical magnetic resonance Toxicological applications of magnetic resonance. Prog. Nucl. Magn. Reson. Spectrosc. 2001, 39, 1–40. (22) Wold, S.; Antti, H.; Lindgren, F.; Ohman, I. Orthogonal signal correction of near-infrared spectra. Chemom. Intell. Lab. Syst. 1998, 44, 175–185.
4044
Journal of Proteome Research • Vol. 9, No. 8, 2010
Zira et al. (23) Nicholson, J. K.; Foxall, P. J. D.; Spraul, M.; Farrant, R. D.; Lindon, J. C. 750 MHz 1H and 1H-l3C NMR spectroscopy of human blood plasma. Anal. Chem. 1995, 67, 793–811. (24) Han, W. K.; Alinani, A.; Wu, C. L.; Michaelson, D.; Loda, M.; McGovern, F. J.; Thadhani, R.; Bonventre, J. V. Human kidney injury molecule-1 is a tissue and urinary tumor marker of renal cell carcinoma. J. Am. Soc, Nephrol 2005, 16, 1126–1134. (25) Rogers, M. A.; Clarke, P.; Noble, J.; Munro, N. P.; Paul, A.; Selby, P. J.; Banks, R. E. Proteomic profiling of urinary proteins in renal cancer by surface enhanced laser desorption ionization and neuralnetwork analysis: identification of key issues affecting potential clinical utility. Cancer Res. 2003, 63, 6971–6983. (26) Saude, E. J.; Adamko, D.; Rowe, B. H.; Marrie, T.; Sykes, B. D. Variation of metabolites in normal human urine. Metabolomics 2007, 3, 439–451. (27) Assfalg, M.; Bertini, I.; Colangiuli, D.; Luchinat, C.; Schaefer, H.; Schutz, B.; Spraul, M. Evidence of different metabolic phenotypes in humans. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 1420–1424. (28) Whitehead, T. L.; Kieber-Emmons, T. Applying in vitro NMR spectroscopy and 1H NMR metabonomics to breast cancer characterization and detection. Prog. Nucl. Magn. Reson. Spectrosc. 2005, 47, 165–174. (29) Griffin, J. L.; Kauppinen, R. A. A metabolomics perspective of human brain tumours. FEBS J. 2007, 274, 1132–1139. (30) Sitter, B.; Lundgren, S.; Bathen, T. F.; Halgunset, J.; Fjosne, H. E.; Gribbestad, I. S. Comparison of HR MAS MR spectroscopic profiles of breast cancer tissue with clinical parameters. NMR Biomed. 2006, 19, 30–40. (31) Lindon, J. C.; Nicholson, J. K.; Holmes, E.; Everett, J. R. Metabonomics: metabolic processes studied by NMR spectroscopy of biofluids. Concepts Magn. Reson. 2000, 12, 289–320. (32) Fuhrman, B.; Wolfovitz, E.; Cogan, U.; Brook, G. J. Hypocholesterolemia and cancer: tumor cells induce enhanced LDL uptake by non-tumor cells and stimulate oxidative modification of LDL. Pathophysiology 1999, 6, 205–210. (33) Vitols, S.; Angelin, B.; Ericsson, S.; Gahrton, G.; Juliusson, G.; Masquelier, M.; Paul, C.; Peterson, C.; Rudling, M.; Soderberg-Reid, K.; Tidefelt, U. Uptake of low density lipoproteins by human leukemic cells in vivo: Relation to plasma lipoprotein levels and possible relevance formselective chemotherapy. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 2598–2602. (34) Fiorenza, A. M.; Branchi, A.; Sommariva, D. Serum lipoprotein profile in patients with cancer. A comparison with non-cancer subjects. Int. J. Clin. Lab. Res. 2000, 30, 141–145. (35) Gebhard, R. L.; Clayman, R. V.; Prigge, W. F.; Figenshau, R.; Staley, N. A.; Reesey, C.; Beart, A. Abnormal cholesterol metabolism in renal clear cell carcinoma. J. Lipid Res. 1987, 28, 1177–1184. (36) White, R. M. On the occurrence of crystals in tumours. J. Pathol. Bacteriol. 1909, 13, 3–10. (37) Rudling, M.; Collins, V. P. Low density lipoprotein receptor and 3-hydroxy-3-methylglutaryl coenzyme A reductase mRNA levels are coordinately reduced in human renal cell carcinoma. Biochim. Biophys. Acta 1996, 1299, 75–79. (38) Tosia, M. R.; Tugnoli, V. Cholesteryl esters in malignancy. Clin. Chim. Acta 2005, 359, 27–45. (39) Abiaka, C.; Al-Awadi, F.; Al-Sayer, H.; Gulshan, S.; Behbehani, A.; Farghally, M.; Simbeye, A. Serum antioxidant and cholesterol levels in patients with different types of cancer. J. Clin. Lab. Anal. 2001, 15, 324–330. (40) Su ¨ llentrop, F.; Moka, D.; Neunauer, S.; Haupt, G.; Engelmann, U.; Hahn, J.; Schicha, H. 31P NMR spectroscopy of blood plasma: determination and quantification of phospholipid classes in patients with renal cell carcinoma. NMR Biomed. 2002, 15, 60–68. (41) Kubota, A.; Meguid, M. M.; Hitch, D. C. Amino-acid profiles correlate diagnostically with organ site in 3 kinds of malignanttumors. Cancer 1992, 69, 2343–2348. (42) Lai, H. S.; Lee, J. C.; Lee, P. H.; Wang, S. T.; Chen, W. J. Plasma free amino acid profile in cancer patients. Semin. Cancer Biol. 2005, 15, 267–276. (43) Holmes, E.; Leng, L. R.; Stamler, J.; Bictash, M.; Yap, I. K. S.; Chan, Q.; Ebbels, T.; De Iorio, M.; Brown, I. J.; Veselkov, K. A.; Daviglus, M. L.; Kesteloot, H.; Ueshima, H.; Zhao, L.; Nicholson, J. K.; Elliott, P. Human metabolic phenotype diversity and its association with diet and blood pressure. Nature 2008, 453, 396–400.
PR100226M
