WO2011128256A1

WO2011128256A1 - Metabolic markers for diagnosing of cancer

Info

Publication number: WO2011128256A1
Application number: PCT/EP2011/055492
Authority: WO
Inventors: Kurt Baeten; Peter Adriaensens; Pieter Stinissen
Original assignee: Universiteit Hasselt
Priority date: 2010-04-13
Filing date: 2011-04-08
Publication date: 2011-10-20

Abstract

The present invention relates to a panel of metabolites in a body fluid sample that can be used for the diagnosis of breast, lung, colon and prostate cancer. More specifically, the invention relates to the NMR analysis of a urine or blood sample, whereby the simultaneous determination of the concentration of a panel of metabolites allows the diagnosis of breast, lung, colon or prostate cancer with a high sensitivity and specificity. In particular, the panel of metabolites comprises methyl histidine and proline.

Description

Metabolic markers for diagnosing of cancer

The present invention relates to a panel of metabolites in a body fluid sample that can be used for the diagnosis of breast, lung, colon and prostate cancer. More specifically, the invention relates to the NMR analysis of a urine or blood sample, whereby the simultaneous determination of the concentration of a panel of metabolites allows the diagnosis of breast, lung, colon or prostate cancer with a high sensitivity and specificity.

Breast cancer is the most common death for women between the age of 35 and 55 years. It was estimated that in Europe 430.000 women suffered from breast cancer in 2006 with over 1 31 .000 of deaths (Ferlay et al . 2007). However, it was also noted that the death rate decreased by approximately 35% for women between the age of 50 and 69 years due to mammography screening programs (International agency for research on cancer 2002, Roder et al. 2008). However, most of the screening programs are focused to women between the age of 50 and 69 years old. This is problematic since approximately 25% of all breast cancers develop before the age of 50 years (Belgian Cancer Registry, 2006). However, the sensitivity of the mammography is only 62%-76% for women at this age (Vainio et al. 2002).

Lung cancer affects yearly ± 206.000 men and ± 74.000 women in Europe. This is the second most frequent cancer for men and the third most for women (Ferlay et al. 2007). In Belgium, each year ± 5200 men and ± 1700 women are diagnosed with lung cancer (Belgian Cancer Registry, 2010). The risk to be diagnosed for lung cancer before the age of 75 is 7.2% for men and 1 .9% for women. Even as important is the high mortality rate of this cancer. In Europe, ±182.000 men and ± 66.000 women die from lung cancer each year (Ferlay et al. 2007). Nonetheless, there are currently no screening tools present to detect this disease at an early stage (Pastorino et al. 2010). Consequently, lung cancer is often only diagnosed , when malignant cells have spread to secondary organs and physical complaints occur as a result of secondary tumors. New screening tools that allow the detection of lung cancer at an early stage would probably cause an important augmentation of the life expectancy of lung cancer patients and lead to a reduction of the mortality rate of this cancer (Whynes 2008).

Also for prostate cancer and colon cancer each year respectively 347.000 and 220.000 persons are affected (Ferlay, 2006). Currently the most used screening method for prostate cancer is the PSA test, however this test lacks specificity (Tosoian et al. 2010). With regard to colon cancer, the fecal blood test can be performed but also this latter lacks sensitivity and specificity (Fraser et al. 2006). However, also with regard to prostate and colon cancer early detection can prevent in many cases drastic therapy and surgery, and lead to a better disease outcome.

Recent studies indicate that a new enormous potential for the detection of many diseases arises from the monitoring of the metabolic signature. During disease unique alterations are expected to occur in this metabolic signature. N MR spectroscopy is the utmost excellent technique to analyse the metabolic signature. Positive results have been obtained in the field of prostate cancer, by using NMR of biopsy tissues (Van Asten et al., 2008, Tessem et al., 2008). Recently, using magic angle spinning MR spectroscopy on breast cancer biopsies, a method for determining prognostic factors has been developed (Giske0degard et al., 2010). However, although in this case metabolomics was used for the prognosis of the evolution of breast cancer, the use of a biopsy is rather confronting for the patient, and a reliable diagnosis method for cancer and on body fluid samples, such as blood, serum, or urine would be far more acceptable for the patients.

1

H NMR of body fluid, such as human serum has been used as prognostic and diagnostic tool of different diseases (Ala-Korpela, 2007). However, this author states that a metabolic profile for screening and detection of asymptomatic individuals at early stages of cancer is highly

1

unlikely via H NMR metabonomics of serum due to human biocomplexity and the top-of the

1

iceberg metabolic space reflected by H NMR spectroscopy. Odunsi et al. (2005) succeeded in classifying a set of epithelial ovarian cancer patients with 100% specificity and selectivity, but

1 most patients were in a late phase of the disease. Recently, Tizani et al (2009) used H NMR based metabolomics for early diagnosis of oral cancer. In both cases, the complete metabolome was used to make the diagnosis. US20090075284 describes the metabolic profiling of prostate cancer, and discloses a method for diagnosing cancer, comprising the detection of the presence or absence of one or more cancer specific metabolites selected from the group consisting of sarcosine, cysteine, glutamate, asparagines, glycine, leucine, proline, threonine, histidine, n-acetyl-aspartic acid, inosine, inositol, adenosine, taurine, creatine, uric acid, glutathione, uracil, kynurenine, glycerol-s-phosphate, glycocholic acid, suberic acid, glutamic acid, xanthosine, 4-acetamidobutyric acid and thymine. However, the panel is only validated for prostate cancer. Moreover, although the application indicates that NMR may be used for analysis, it is unclear whether the cited panel of metabolites could be analyzed in one NMR run. The set of metabolites cited is rather extensive, and there is no indication that a random, limited selection from the list would lead to a reliable diagnosis. Finally, although the list of metabolites cited in the application is extremely elaborated, it doesn't mention the analysis of methyl histidine, which in our hands is an important metabolite for a reliable diagnosis of cancer. Indeed, surprisingly, we found that a limited set of metabolites, comprising at least methyl histidine and proline, as measured in body fluids such as blood, serum, plasma as well as urine can be used to determine cancer, specifically breast cancer, lung cancer, colon cancer and prostate cancer and that the limited set gives a higher specificity and selectivity than the complete metabolome.

A first aspect of the invention is the use of the analysis of methyl histidine and proline, preferably the analysis of methyl histidine, proline and citrate in a body fluid sample of a subject for the diagnosis of cancer. A "body fluid sample" as used here includes, but is not limited to a sample of blood, serum, plasma or urine. "Proline" as used here, includes also proline modifications such as but not limited to oxoproline and hydroxyproline. "Acetate" as used here, includes also acetate modifications such as but not limited to acetoacetate. Preferably, said cancer is selected from the group consisting of breast cancer, colon cancer, prostate cancer or lung cancer, more preferably said cancer is selected from the group consisting of breast cancer, colon cancer, and lung cancer, more preferably said cancer is selected from the group consisting of breast cancer and lung cancer, most preferably said cancer is breast cancer. Preferably, the analysis is the determination of the concentration. The determination of the concentration, as used here, can be absolute, or relative, compared to a healthy control. A healthy control in the meaning of the invention is a subject that is not suffering from cancer, preferably not suffering from any other chronic disease. Preferably, the concentration of metabolite of the healthy control is not derived from one subject, but taken as an average of several healthy subjects. In another embodiment, the analysis may be the follow up of said concentration over a certain period of time. Preferably, said analysis is the analysis of a limited panel of metabolites. Preferably, said limited panel comprises less than 30 metabolites, even more preferably less than 20 metabolites, even more preferably less than 15 metabolites, even more preferably less than 14 metabolites, more preferably less than 13 metabolites, even more preferably less than 12 metabolite, even more preferably less than 1 1 metabolites, even more preferably less than 1 1 metabolites, even more preferably less than 10 metabolites, even more preferably less than 9 metabolites. Most preferably said limited panel consists of 8 metabolites or less, comprising methyl histidine and proline, in combination with one or more metabolite species selected from the group consisting of acetate, citrate, isoleucine, leucine, tryptophane and lipids, for the diagnosis of cancer. Even more preferably, said analysis is carried out by NMR. Most preferably, said analysis has a sensitivity and specificity of at least 80% for both parameters, more preferably at least 85%, more preferably at least 90%, more preferably at least 91 %, more preferably at least 92%, more preferably at least 93%, more preferably at least 94%, more preferably at least 95%, more preferably at least 96%, more preferably at least 97%, more preferably at least 98%, more preferably at least 99%, most preferably 100%.

Another aspect of the invention is the diagnosis of cancer comprising (1 ) NMR analysis of a bodyfluid sample; (2) identifying the spectral region of at least the metabolites acetate, citrate, methyl histidine, isoleucine, leucine, proline, tryptophane and lipids; and (3) comparing the peak intensity and or area of these regions with a healthy control. Preferably, said cancer is selected from the group consisting of breast cancer, colon cancer, prostate cancer and lung cancer more preferably said cancer is selected from the group consisting of breast cancer, colon cancer and lung cancer, more preferably said cancer is selected from the group consisting of breast cancer and lung cancer, most preferably said cancer is breast cancer. Preferably, analysis has a sensitivity and specificity of at least 80% for both parameters, preferably at least 85%, preferably at least 90%, more preferably at least 91 %, more preferably at least 92%, more preferably at least 93%, more preferably at least 94%, more preferably at least 95%, more preferably at least 96%, more preferably at least 97%, more preferably at least 98%, more preferably at least 99%, most preferably 100%. Preferably, the limited panel of metabolites leads to a higher specificity and/or sensitivity, even more preferably to a higher specificity and sensitivity than the whole metabolome, as measured by partial least square determinant analysis. The "whole metabolome" as used here means the use of the complete N MR spectrum from one analysis (and use of the spectrum to determine the metabolite concentration).

BRIEF DESCRIPTION OF THE FIGURES

Figure 1 : Comparison of patients with breast cancer and control persons (CP). The 400

1

MHz H NMR spectra of plasma samples from a typical breast cancer patient (a) and a control person (b). No systematically occurring changes could be detected by visual inspection.

Figure 2: Compounds identified to be statistically significantly different between the breast cancer patients (n=25) and control persons (n=25) by a student t-test. (a)Fold change of the significantly different components after normalization, (b) Significance of the different compounds indicated as standard (p < 0.05, grey) and strong (p < 0.01 , black).

Figure 3: Multivariate analyses on the entire NMR spectrum to discriminate between breast cancer patients and control persons, (a) PCA-score plot of breast cancer patients (V, n=25) and control persons (■, n=25) using all 96 integral segments, (b) PLS-DA score plot showing considerable separation between the breast cancer patients and the control persons. (c) The Y-predicted plot indicates a low misclassification ratio (23/25 breast cancer patients and 21/25 control persons were correctly classified), (d) The regression coefficients of the PLS-DA model shown in (b). Figure 4: Metabolite panel that allows discrimination between breast cancer patients and control persons using blood plasma samples, (a) Using the information from the student t-test and the regression coefficients from the PLS-DA, a panel of 8 metabolites was found that allows the separation of the breast cancer patients (V, n=25) and control persons (■, n=25) to a high degree, (b) The variable of importance plot indicates the weight factor of the different metabolites in the PLS-DA model shown in (a). Positive coefficients indicate a down regulation in the breast cancer patient persons compared to the control persons, negative coefficients indicate higher values, (c) The Y-predicted plot indicates a very low misclassification ratio (23/25 breast cancer patients and 22/25 control persons were correctly classified).

Figure 5: Methyl histidine and proline are most important factors for discriminating between breast cancer patients and control persons using blood plasma samples, (a) A

PLS-DA model was constructed using the data of the integral segments representing Methyl histidine and Proline of the 25 cancer patients (V) and the 25 control persons (■). (b) Y- predicted scatterplot with the a priori cut-off of 0.5 for class membership.

Figure 6: Metabolite panel that allowing discrimination between breast cancer patients and control persons using blood plasma samples in a second independent cohort.

Using the same panel of metabolites consisting of Tryptophan, Methylhistidine, Lipids (ie lipid signals from triglyceride/phospholipids), Citrate, Proline (ie L-proline as well as 5-oxoproline), Acetate, Isoleucine and Leucine a partial least square discriminate analyses (PLS-DA) was conducted on plasma samples of 28 breast cancer patients (V) and 17 control persons (■). The scatter plot shows a high degree of discrimination with a sensitivity of 89% an d a specificity of 88%.

Figure 7: Metabolite panel allowing discrimination between breast cancer patients in the expanded first cohort of blood plasma samples. Using an orthogonal partial least square discriminant (OPLS-DA) analysis based on the earlier described panel of metabolites with the addition of the metabolites threonine (3.53ppm - 3.51 ppm), lipids (ie lipid signals of fatty acyl chains (2.82ppm - 2.69ppm), glutamine (2.46ppm - 2.42ppm), lactate (1 .34ppm - 1.32ppm) and myoinositol (3.64ppm -3.61 ppm), the 66 breast cancer patients (V) and 63 controls (■) are discriminated with a sensitivity of 92% (61/66) and a specificity of 97% (61/63) (Figure 7a). Using an OPLS-DA analysis on the entire spectrum only lead to a sensitivity and a specificity of both 90% (Figure 7b).

Figure 8: Metabolite panel allowing discrimination between lung cancer patients and control persons using blood plasma samples. Using the panel of metabolites consisting of tryptophan (Trp), Methylhistidine (1 Me-His), Lipids (ie lipid signals from triglyceride/phospholipids), Citrate, Proline (ie L-proline as well as 5-oxoproline, Pro; 5 O-Pro), Acetate (Ace), Isoleucine (lie) and Leucine (Leu) a partial least square discriminate analyses (PLS-DA) was conducted on plasma samples of 18 lung cancer patients (V) and 35 control persons (■). The scatter plot shows a high degree of discrimination with a sensitivity of 94.5% and a specificity of 94.5% (a). Performing a PLS-DA analysis using only the integral segments of Methylhistidine and Proline, lead to a sensitivity of 84% and a specificity of 97% (b). The graph plot indicates the up and downregulated levels of the different metabolites (c).

Figure 9: Metabolite panel allowing discrimination between colon cancer patients and control persons using blood plasma. Using the panel of metabolites consisting of tryptophan (Trp), Methylhistidine (1 Me-His), Lipids (ie lipid signals from triglyceride/phospholipids), Citrate, Proline (ie L-proline as well as 5-oxoproline, Pro; 5 O-Pro), Acetate (Ace), Isoleucine (lie) and Leucine (Leu) a partial least square discriminate analyses (PLS-DA) was conducted on plasma samples of 5 colon cancer patients (V) and 15 control persons (■). The scatter plot shows a high level of discrimination with a sensitivity of 80% and a specificity of 100% (a). Performing a PLS-DA analysis using only the integral segments of Methylhistidine and Proline, lead to a sensitivity of 80% and a specificity of 100% (b). The graph plot indicates the up and downregulated levels of the different metabolites (c).

Figure 10: Metabolite panel allowing discrimination between breast cancer patients and control persons using urine. Using the panel of metabolites consisting of tryptophan (Trp) , Methylhistidine (1 Me-His), Citrate, Proline (ie L-proline as well as 5-oxoproline, Pro; 5 O-Pro), Acetate (Ace), Isoleucine (lie) and Leucine (Leu) a partial least square discriminate analyses (PLS-DA) was conducted on urine samples of 8 breast cancer patients (V) and 5 control persons (■). The scatter plot shows an absolute discrimination with a sensitivity of 100% and a specificity of 100% (a). Performing a PLS-DA analysis using only the integral segments of Methylhistidine and Proline, lead to a sensitivity of 87,5% and a specificity of 100% (b). The graph plot indicates the up and downregulated levels of the different metabolites (c).

Figure 11 : Metabolite panel allowing discrimination between lung cancer patients and control persons using urine samples. Using the panel of metabolites consisting of tryptophan (Trp), Methylhistidine (1 Me-His), Citrate, Proline (ie L-proline as well as 5- oxoproline, Pro; 5 O-Pro), Acetate (Ace), Isoleucine (lie) and Leucine (Leu) a partial least square discriminate analyses (PLS-DA) was conducted on urine samples of 4 lung cancer patients (Δ) and 10 control persons (■). The scatter plot shows an absolute discrimination with a sensitivity of 100% and a specificity of 100% (a). Performing a PLS-DA analysis using only the integral segments of Methylhistidine and Proline, lead to a sensitivity of 75% and a specificity of 80% (b). The graph plot indicates the up and downregulated levels of the different metabolites (c).

Figure 12: Metabolite panel allowing discrimination between prostate cancer patients and control persons using urine samples. Using the panel of metabolites consisting of tryptophan (Trp), Methylhistidine (1 Me-His), Citrate, Proline (ie L-proline as well as 5- oxoproline, Pro; 5 O-Pro), Acetate (Ace), Isoleucine (lie) and Leucine (Leu) a partial least square discriminate analyses (PLS-DA) was conducted on urine samples of 4 prostate cancer patients (Δ) and 5 control persons (■). The scatter plot shows an absolute discrimination with a sensitivity of 100% and a specificity of 100% (a). Performing a PLS-DA analysis using only the integral segments of Methylhistidine and Proline, lead to a sensitivity of 100% and a specificity of 100% (b). The graph plot indicates the up and downregulated levels of the different metabolites (c).

EXAMPLES

Materials and methods to the examples

Specimen collection and preparation

From December 2009 to March 201 1 , preoperative or pretreatment blood plasma and/or urine samples of patients diagnosed with breast cancer, lung cancer, prostate cancer or colon cancer were collected at the Ziekenhuis Oost-Limburg (ZOL) and Universitair Ziekenhuis Leuven under an approved IRB protocol. The tumor tissue pathology was verified by a pathologist. For controls, plasma and urine samples were obtained from a mixed population, i.e. healthy persons, persons with arthritis, elevated cholesterol, pregnant persons... . Within 30 minutes of blood collection by venipuncture, the plasma was separated by centrifugation. Alternative, blood was stored at 4°C and plasma was separated by centrifugation within 12hrs after the venipuncture. Aliquots were stored at -80°C until assayed. For urine collection, samples were frozen within 1 hr at -20°C and within 12hrs stored at -80°C until assayed.

1

H-NMR spectroscopic analyses

Thawed samples were centrifuged at 13000g for 4min. Subsequently, two hundred microliter bloodplasma was d i l uted i n 600 μ Ι deuteriu m oxide contai n ing tri methylsi lyl-2, 2,3,3- tetradeuteropropionic acid (TSP). Urine was centrifuged at 13000g for 4min to remove cell debris and 2000μΙ deuterium oxide was added to 4000μΙ urine and adjusted to a pH of 6.8. Eight hundred microliter was transferred into 5mm NMR tubes. 1 H NMR spectra were obtained using a 400 MHz spectrometer (Varian/Agilent, Nuclear Magnetic Resonance Instruments, Palo Alto, California, USA) operating at 9.4 Tesla with a liquid state probe. One-dimensional spectra were acquired using a presat NOESY -RD-90°-t-90°-tm-90°-ACQ and a presat Carr- PurcellMeiboom-Gill (CPMG) pulse sequence which has the form -RD-90°-(t-180°-t)2n-ACQ, with 907180° being a 907180° RF pulse, 2t.2n is the spin-echo delay, and 2n representing the number of spin echoes. The number of scans is 96, number of dummy scans = 16, number of time domain points = 36k and spectrum width = 6k. Irradiation is at the water peak during RD. Datasets were zero filled to 65k points and exponential line broadenings of 0.7 Hz were applied. Spectra were phased manually, baseline corrected and referenced to the TSP resonance (O.Oppm). Spectra were binned and each binned spectrum was normalized to the total spectra area except for the remnant water and TSP signal, as well as the urea, glucose and fructose signals, due to their disproportionate levels. Statistical analyses

The binned and normalized NMR spectral data were firstly analyzed using student t-tests. Subsequently data were imported into SIMCA-P+ (Version 12.0, Umetrics, AB, Umea, United Kingdom). The discrimination between breast cancer patients and control persons was investigated in an unsupervised mode using the principal component analysis (PCA), as well as using supervised partial least squares discriminant analysis (PLS-DA) and orthogonal projections to latent structures discriminant analysis (OPLS-DA) using all or a selected panel of metabolites. The predicted classification of the different subjects was expressed as sensitivity (the proportion of diseased persons that are classified as diseased) and specificity (the proportion of control persons that are classified as control persons). Example 1 : Metabolite identification and quantification in breast cancer

The 400-MHz CPMG 1 H-NMR-spectra of human plasma samples from patients with breast cancer and control persons were compared visually (Figure 1 ). No systematically occurring changes could be detected. To reduce the complexity of the NMR data and to facilitate statistical analyses the spectra were reduced to 96 integral segments. To ascertain the metabolic differences between the 25 breast cancer patients and 25 control persons, a student t-test was applied to all features in the normalized data set. Figure 2a shows the normalized changes from the normal state for all metabolites and resonances with p-values < 0.05. Fifteen integral segments were found to be significantly different with p-values < 0.05, of which twelve had a p-value between 0.05-0.01 and three segments differed with a p-value lower than 0.01 (Figure 2b).

Example 2: Multivariate analyses on blood plasma of breast cancer patients and controls

To determine whether it is possible to distinguish breast cancer patients from control persons, we performed multivariate analysis on a first cohort of 25 breast cancer patients (V) and 25 control persons (■) from Ziekenhuis Oost-Limburg and Hasselt University. Firstly, we conducted an unsupervised principal component analysis (PCA) using the data of the entire spectrum, i.e. all 96 integral segments (Figure 3a). In the supervised partial least squares- discriminant analyses (PLS-DA) the potential to detect breast cancer by applying NMR spectroscopy on blood plasma samples to monitor metabolic changes becomes more clear (Figure 3b). Using the information of the entire spectrum on all the metabolites detected in the bloodplasma, clustering of the breast cancer patients occurs. The Y-predicted plot indicate a low misclassification rate of the included samples, i.e. 23 out of 25 breast cancer patients were correctly classified as well as 21/25 control patients corresponsive with a sensitivity of 92% and a specificity of 84% (Figure 3c). The contribution of the different signal integral segments in this discrimination is shown by the regression coefficients (Figure 3d). It can be noted that the metabolites found to be significantly different by the student t-test have an important role in this discrimination. For example integral segment Var_8 (=methyl histidine), Var_61 (proline), Var_93 (leucine) show a high absolute coefficient value.

Subsequently, we investigated whether a select panel of metabolites allows discriminating breast cancer patients from control persons. Using the information from the student t-tests, the regression coefficients from the PLS-DA analyses and by inserting and deleting metabolites from the model, we determined that a panel of 8 different metabolites (Table 1 ), allows the separation of breast cancer patients and control persons to a high degree (Figure 4a). This panel consists of Tryptophan (Trp; 7.64ppm - 7.53ppm); Methylhistdine (1 Me-His; 7.08ppm - 6.98ppm), Lipids (ie lipid signals from triglyceride/phospholipids (4ppm - 3.99ppm, 3.99ppm - 3.98ppm)), Citrate (2.53ppm - 2.51 ppm), Proline (ie L-proline as well as 5-oxoproline) (Pro, 5 O-Pro; 2.4ppm -2.39ppm, 2.39ppm - 2.37ppm, 2.34ppm - 2.315ppm), Acetate (Ace; 1 .93ppm -1 .92ppm), Isoleucine (l ie; 1 .03ppm -1 .02ppm, 1 .02ppm - LOOppm) and Leucine (Leu; 0.98ppm - 0.96ppm, 1.72ppm - 1.70ppm). Figure 4b shows the importance of the different metabolites in the model, indicating an important role for proline and methylhistidine. Using this panel of metabolites breast cancer patients and control patients were discriminated with a sensitivity of 92% (23/25 breast cancer patients) and specificity of 88% (22/25 control persons) as visualized by the Y-predicted scatterplot with the a priori cut-off of 0.5 for class membership (Figure 4c). The discriminatory power of this panel is even higher as compared to using the entire metabolome as described above. Further investigation indicated a predominant role for methyl-histidine and proline in the separation of breast cancer patients and control persons. A PLS-DA model build on these 2 metabolites shows already a high degree of clustering of the breast cancer patients and the control persons (Figure 5a). Based on the Y-predicted scatterplot, it was shown that this already leads to a correct classification of 19/25 breast cancer patients and 21/25 control persons (Figure 5b).

These findings were confirmed in a second cohort of 17 controls collected in collaboration with the Leuvens University Centre of Cancer Prevention and 28 breast cancer patients of the Academic Hospital of Leuven. Conducting a partial least discriminate analysis (PLS-DA) using the integral segments of tryptophan, (Trp; 7.64ppm - 7.53ppm); Methylhistdine (1 Me-His; 7.08ppm - 6.98ppm), Lipids (ie lipid signals from triglyceride/phospholipids (4ppm - 3.99ppm, 3.99ppm - 3.98ppm)), Citrate (2.53ppm - 2.51 ppm), Proline (ie L-proline as well as 5- oxoproline) (Pro, 5 O-Pro; 2.4ppm -2.39ppm, 2.39ppm - 2.37ppm, 2.34ppm - 2.315ppm), Acetate (Ace; 1 .93ppm -1 .92ppm), Isoleucine (lie; 1 .03ppm -1 .02ppm, 1.02ppm - LOOppm) and Leucine (Leu; 0.98ppm - 0.96ppm, 1 .72ppm - 1 .70ppm). A high level of discrimination is achieved with a sensitivity of 88% and a specificity of 90% (Figure 6).

Subsequently the first cohort was expanded to 63 control persons and 66 breast cancer patients. Using an orthogonal partial least square discriminant (OPLS-DA) analysis based on the earlier described panel of metabolites added with the metabolites threonine (3.53ppm - 3.51 ppm), lipids signal of fatty acyl chains (2.82ppm - 2.69ppm), glutamine (2.46ppm - 2.42ppm), lactate (1 .34ppm - 1 .32ppm) and myoinositol (3.64ppm -3.61 ppm) breast cancer patients (V) and controls (■) are discriminated with a sensitivity of 92% (61 /66) and a specificity of 97% (61 /63) (Figure 7a). Using an OPLS-DA analysis on the entire spectrum lead only to a sensitivity and a specificity of both 90% (Figure 7b) indicating that the described limited panel of metabolites leads to a higher sensitivity and specificity than the whole metabolome..

In conclusion, using NMR spectroscopy we could identify a panel of metabolites that allows the discrimination of breast cancer patients and control persons using a blood sample.

Table 1: Metabolite panel for the detection of breast cancer

Example 3: Multivariate analyses on blood plasma of lung cancer patients and controls To determine whether it is possible to also distinguish lung cancer patients from control persons using the above described panel of metabolites, we performed multivariate analysis on blood plasma samples of 18 lung cancer patients and 36 control persons. Firstly, we conducted a partial least discriminate analysis (PLSDA) using the integral segments of tryptophan, (Trp; 7.64ppm - 7.53ppm); Methylhistidine (1 Me-His; 7.08ppm - 6.98ppm), Lipid signals of triglyceride/phospholipids (4ppm - 3.99ppm, 3.99ppm - 3.98ppm), Citrate (2.53ppm

- 2.51 ppm), Proline (ie L-proline as well as 5-oxoproline) (Pro, 5 O-Pro; 2.4ppm -2.39ppm, 2.39ppm - 2.37ppm, 2.34ppm - 2.315ppm), Acetate (Ace; 1 .93ppm -1 .92ppm), Isoleucine (lie; 1 .03ppm -1 .02ppm, 0.952ppm - 0.943ppm) and Leucine (Leu; 0.98ppm - 0.96ppm, 0.96ppm

- 0.95ppm; 1 ,72ppm - 1 .70ppm). As shown in Figure 8a a high level of discrimination is achieved with a sensitivity of 94.5% and a specificity of 94.5%. Performing a PLSDA analysis using only the integral segments of Methylhistidine and Proline, lead to a sensitivity of 79% and a specificity of 97% (Figure 8b), indicating the importance of these two metabolites in the discrimination between lung cancer patients and controls. Analysing the changes in the different metabolites included in the panel indicated a downregulation of tryptophan, 1 - Methylhistidine, Lipids, Citrate, Acetate, Leucine and Isoleucine and an upregulation of proline and oxoproline (Figure 8c).

Example 4: Multivariate analyses on plasma of colon cancer patients and controls

To determine whether it is possible to distinguish colon cancer patients from control persons using the above described panel of metabolites, we performed multivariate analysis on blood plasma samples of 5 colon cancer (V) patients and 10 control persons (■). Hereto, we conducted a partial least discriminate analysis (PLS-DA) using the integral segments of tryptophan, (Trp; 7.64ppm - 7.53ppm); Methylhistidine (1 Me-His; 7.08ppm - 6.98ppm), Lipids (ie lipid signals of from triglyceride/phospholipids (4ppm - 3.99ppm, 3.99ppm - 3.98ppm)), Citrate (2.53ppm - 2.51 ppm), Proline (ie L-proline as well as 5-oxoproline) (Pro, 5 O-Pro; 2.4ppm -2.39ppm, 2.39ppm - 2.37ppm, 2.34ppm - 2.315ppm), Acetate (Ace; 1 .93ppm - 1 .92ppm), Isoleucine (lie; 1 .03ppm -1 .02ppm, 0.952ppm - 0.943ppm) and Leucine (Leu; 0.98ppm - 0.96ppm, 0.96ppm - 0.95ppm; 1 ,72ppm - 1.70ppm). As shown in Figure 9a a discrimination is achieved with a sensitivity of 80 % and a specificity of 90%. Performing a PLS-DA analysis using only the integral segments of Methylhistidine and Proline, lead to a sensitivity of 80% and a specificity of 100% (Figure 9b). Analysing the changes in the different metabolites included in the panel indicated a downregulation of tryptophan, 1 -Methylhistidine and an upregulation of lipids, Citrate, Acetate, Leucine and Isoleucine, Proline and 5- Oxoproline (Figure 9c). Example 5: Multivariate analyses on urine of breast cancer patients and controls

To determine whether it is possible to distinguish breast cancer patients from control persons by measurement of the above described panel of metabolites in urine samples with NMR spectroscopy, we performed multivariate analysis on urine samples of 8 breast cancer patients (V) and 5 control persons (■). Firstly, we conducted a partial least discriminate analysis (PLS- DA) using the integral segments of tryptophan, (Trp; 7.64ppm - 7.53ppm); Methylhistidine (1 Me-His; 7.77ppm -7.75ppm, 7.02ppm - 7.00ppm), Citrate (2.765ppm - 2.72ppm, 2.72ppm - 2.67ppm, 2.61 ppm - 2.57ppm, 2.57ppm - 2.53ppm), Proline (ie L-proline as well as 5- oxoproline) (Pro, 5 O-Pro; 4.15ppm - 4.1 1 ppm, 3.44ppm - 3.41 ppm, 3.33ppm - 3.32ppm 2.41 ppm -2.37ppm, 2.02ppm - 1 .98ppm), Acetate (Ace; 1 .91 ppm -1 .89ppm), Isoleucine (lie; 1 .03ppm -1 .01 ppm) and Leucine (Leu; 1 .71 ppm - 1 .60ppm). The concentration of the triglycerides/phospholipids could not be determined in the urine due to their absence and/or overlap with creatinine and hippurate signals. As shown in Figu re 10a an absolute discrimination is achieved with a sensitivity of 100 % and a specificity of 100%. Performing aft PLS-DA analysis using only the integral segments of Methylhistidine and Proline, lead to a sensitivity of 87.5% and a specificity of 100% (Figure 10b). Analysing the changes in the different metabolites included in the panel indicated an upregulation of tryptophan, 1 - Methylhistidine, Citrate, Acetate, Leucine and Isoleucine, proline and oxoproline (Figure 10c). Example 6: Multivariate analyses on urine samples of lung cancer patients and controls

To determine whether it is possible to also distinguish lung cancer patients from control persons using the above described panel of metabolites measured in urine, we performed multivariate analysis on urine samples of 4 lung cancer patients and 10 control persons. Firstly, we conducted a partial least discriminate analysis (PLS-DA) using the integral segments of tryptophan, (Trp; 7.64ppm - 7.53ppm); Methylhistdine (1 Me-His; 7.77ppm - 7.75ppm, 7.02ppm - 7.00ppm), Citrate (2.765ppm - 2.72ppm, 2.72ppm - 2.67ppm, 2.61 ppm - 2.57ppm, 2.57ppm - 2.53ppm), Proline (ie L-proline as well as 5-oxoproline) (Pro, 5 O-Pro; 4.15ppm - 4.1 1 ppm, 3.44ppm - 3.41 ppm, 3.33ppm - 3.32ppm 2.41 ppm -2.37ppm, 2.02ppm - 1 .98ppm), Acetate (Ace; 1 .91 ppm -1 .89ppm), Isoleucine (lie; 1 .03ppm -1 .01 ppm) and Leucine (Leu; 1 .71 ppm - 1 .60ppm). As shown in Figure 1 1 a an absolute discrimination is achieved with a sensitivity of 100 % and a specificity of 100%. Performing a PLS-DA analysis using only the integral segments of Methylhistidine and Proline, lead to a sensitivity of 75% and a specificity of 80% (Figure 1 1 b). Analysing the changes in the different metabolites included in the panel indicated a downregulation of Citrate and Acetate and an upregulation of Tryptophan, 1 -Methylhistidine, Leucine and Isoleucine, Proline and oxoproline (Figure 1 1 c).

Example 7: Multivariate analyses on urine of prostate cancer patients and controls

To determine whether it is possible to distinguish prostate cancer patients from control persons using the above described panel of metabolites measured in urine , we performed multivariate analysis on urine samples of 4 prostate cancer patients and 5 control persons. Firstly, we conducted a PLS-DA analysis using the integral segments of tryptophan, (Trp; 7.64ppm - 7.53ppm); Methylhistidine (1 Me-His; 7.77ppm -7.75ppm, 7.02ppm - 7.00ppm), Citrate (2.765ppm - 2.72ppm, 2.72ppm - 2.67ppm, 2.61 ppm - 2.57ppm, 2.57ppm - 2.53ppm), Proline (ie L-proline as well as 5-oxoproline) (Pro, 5 O-Pro; 4.15ppm - 4.1 1 ppm, 3.44ppm - 3.41 ppm, 3.33ppm - 3.32ppm 2.41 ppm -2.37ppm, 2.02ppm - 1.98ppm), Acetate (Ace; 1 .91 ppm -1 .89ppm), Isoleucine (lie; 1 .03ppm -1 .01 ppm) and Leucine (Leu ; 1 .71 ppm - 1 .60ppm). As shown in Figure 12a an absolute discrimination is achieved with a sensitivity of 1 00 % and a specificity of 100%. Performing a PLS-DA analysis using only the integral segments of Methylhistidine and Proline, lead to a sensitivity of 100% and a specificity of 100% (Figure 12b). Analysing the changes in the different metabolites included in the panel indicated a downregulation of tryptophan, 1 -Methylhistidine, Proline and acetate, and an upregulation of Isoleucine and Leucine, while the levels of citrate and oxoproline were unchanged (Figure 12c).

REFERENCES

Ala-Korpela, A. (2007). Potential role of body fluid ¹H NMR metabonomics as prognostic and diagnostic tool. Expert Rev. Mol. Diagn. 7, 761 -773.

Cancer incidence in Belgium 2004-2005, Belgian Cancer Registry 2008, Brussels. - Ferlay, J Ferlay, P Autier, M Boniol, M Heanue, M Colombet and P Boyle (2007). Estimates of the cancer incidence and mortality in Europe in 2006. Ann. One. 18, 581 -592.

- Fraser, C.G., Matthew CM., Mowat N.A., Wilson J.A., Carey F.A., Steele R.J. (2006) Immunochemical testing of individuals positive for guaiac faecal occult blood test in a screening programme for colorectal cancer: an observational study. Lancet Oncol. 7, 127- 131 .

Giske0degard, G.F., Grinde, M.T., Sitter, B., Axelson, D.E., Lundgren, S., Fj0sne, H.E., Dahl, S., Gribbestad, I.S. and Bathen, T.F. (2010). Multivariate medeling and prediction of breast cancer prognostic factors using MR metabolomics. J. Proteome Res 9, 972-979.

International Agency for Research on Cancer (2002) IARC Handbooks of Cancer Prevention. Vol 7. Breast Cancer Screening (IARC, Lyon, France).

Odunsi,K.; Wollman,R.M.; Ambrosone,C.B.; Hutson,A.; McCann,S.E.; Tammela,J.; Geisler,J.P.; Miller,G.; Sellers,!^".; Cliby,W.; Qian,F.; Keitz,B.; lntengan,M.; Lele,S.; Alderfer,J.L. (2005). Detection of epithelial ovarian cancer using 1 H-NMR-based metabonomics. Int. J. Cancer 1 13, 782-788. - Pastorino, U (2010). Lung cancer screening Br. J. Cancer 102, 1681 -1686

Roder D, Houssami N, Farshid G, et al. (2008) Population screening and intensity of screening are associated with reduced breast cancer mortality: evidence of efficacy of mammography screening in Australia. Breast Cancer Res Treat 108:409-416.

Tessem, M.B., Swanson, M.G., Keshari, K.R., Albers, M.J., Joun, D., Tabatabai, Z.L., Si m ko, J . P . , Sh i noha ra , K. , N elson , S .J . , Vigneron , D . B . , Gri bbestad , I .S . and

Kurhanewicz, J. (2008). Evanluation of lactate and alanine as metabolic biomarkers of prostate cancer using 1 H NMR-MAS spectroscopy of biopsy tissues. Magn. Reson. Med. 60,510-516. - Tiziani,S.; EmwasAH.; LodiA; Ludwig.C; Bunce,C.M.; Viant,M.R.; Gunther,U.L. (2008) Optimized metabolite extraction from blood serum for 1 H nuclear magnetic resonance spectroscopy. Anal. Biochem. 377,16-23

- Tosoian, J; Loeb, S. (2010) PSA and beyond: the past, present, and future of investigative biomarkers for prostate cancer. ScientificWorldJournal 10, 1919- 1931 .

- Van Asten, J.J.A., Cuijpers, V., Hulsbergen-van de Kaa, C, Soede-Huijbregts, C, Withes, J.A., Verhofstad, A. and Heerschap, A. (2008). High resolution magic angle spinning NMR spectroscopy for metabolic assessment of cancer presence and gleason score in human prostate needle biopsies. Magn. Reson. Mater. Phy. 21 , 435-442.

Whynes, D.K. (2008). Could CT screening for lung cancer ever be cost effective United Kingdom? Cost. Eff. Resour. Alloc. 6, 5-22

Claims

Use of the analysis of methyl histidine and proline in a body fluid sample for the diagnosis of cancer.

Use of the analysis of a panel of metabolites, present in a body fluid sample, comprising methyl histidine and proline, according to claim 1 , in combination with one or more metabolite species selected from the group consisting of acetate, citrate, isoleucine, leucine, tryptophane and lipids, for the diagnosis of cancer.

The use of the analysis of at least methyl histidine and proline, according to claim 1 or 2, wherein said cancer is selected from the group consisting of breast cancer, colon cancer, prostate cancer and lung cancer.

The use of the analysis of at least methyl histidine and proline, according to any of the claims 1 -3, whereby said panel is analyzed by NMR.

The use of the analysis of at least methyl histidine and proline, according to any of the preceding claims, wherein said body fluid is selected from the group consisting of blood, serum, plasma and urine.

A method for diagnosis of cancer, comprising (1 ) NMR analysis of a body fluid sample; (2) identifying the spectral region of at least the metabolites acetate, citrate, methyl histidine, isoleucine, leucine, proline, tryptophane and lipids; and (3) comparing the peak intensity and or area of these regions with a healthy control.

The method according to claim 6, wherein said cancer is selected from the group consisting of breast cancer, colon cancer, prostate cancer and lung cancer.

The method according to claim 6 or 7, wherein said body fluid is selected from the group consisting of blood, plasma, serum and urine.