Lung cancer is the most common cancer in men and the fourth most common cancer in women worldwide. Until today no effective method permits the early detection of lung cancer. Consequently, lung cancer is often diagnosed owing to symptoms of advanced disease. To address this problem, detection methods with an improved sensitivity and specificity are urgently needed.

Over the past decade, accumulating evidence shows that the metabolism of cancer cells differs from that of normal cells. More specifically, the entire metabolism of cancer cells is reorganized or reprogrammed to increase anabolic reactions that induce cell growth and survival. Metabolic reprogramming during the development of cancer is driven by aberrant signaling pathways due to the activation of oncogenes and the loss of tumor suppressor genes. Furthermore, the microenvironment of the tumor plays a role in metabolic reprogramming. The altered cancer metabolism is characterized by an increased glycolysis, the production of lactate and the biosynthesis of macromolecules, such as proteins, lipids and nucleotides. Cancer cells have a high glycolytic rate and eliminate most of the glucose-derived carbon as lactate rather than oxidizing it completely via oxidative phosphorylation, a phenomenon known as the Warburg effect. The breakdown of glucose and other nutrients leads to a high energy production and provides the Krebs cycle with intermediates, which consequently are allocated to metabolic pathways that support biosynthesis. Metabolites are the end products of cellular metabolism and are therefore closely related to the observed phenotype. Disturbances in biochemical pathways which occur during the development of cancer consequently provoke changes in the metabolic phenotype. As a result, low-molecular weight metabolites are very attractive biomarkers for different cancer types. Nuclear magnetic resonance (NMR) spectroscopy enables the identification and quantitative analysis of complex mixtures of metabolites, as in plasma and serum, without an extended sample preparation.

The present study aims to determine the metabolic phenotype of lung cancer by means of proton (1H)-NMR spectroscopy. Once the phenotype determined (training cohort), this has to be validated by an independent cohort.

Subjects Subjects with lung cancer detected by a computed tomography (CT)-scan and referred to a positron emission tomography (PET)/CT-scan are included. The diagnosis of lung cancer is confirmed by means of an pathological biopsy or by a medical doctor specialized in oncology with respect to radiological or clinical data. The control group consists of subjects who were referred to the department Nuclear Medicine for an examination of the heart. This control group represents the average population, consists of healthy subjects and patients with non-cancer diseases and did not undergo a PET/CT-scan. Exclusion criteria are as follows: (1) not fasted for at least 6 hours, (2) poorly controlled diabetes (fasting plasma glucose concentration ≥ 200 mg/dl) in cancer patients, (3) medication intake at the day of blood sampling and (4) treatment or history of cancer in the preceding 5 years.

The training cohort consists of 80 subject with lung cancer and 80 controls. The validation cohort consist of 250 subject with lung cancer and 250 controls.

Blood sampling and processing Fasting venous blood samples (BD Vacutainer® LH 17 I.U. 10 ml tube) are collected and stored at 4°C within 5 to 10 minutes. Around 8 hours after blood collection, blood samples are transported on crushed ice to the central laboratory and centrifuged at room temperature (swinging bucket centrifuge, 1600 g, 15 minutes). Subsequently, 4 plasma aliquots of 500 µl are transferred into sterile cryovials and stored at -80°C until examination within 6 months. When subjects give permission to store their biological material, 3 aliquots are stored at the University Biobank Limburg (UBiLim) for biomedical research purposes.

Prior to NMR analysis, plasma aliquots are thawed and homogenized using a vortex mixer. After centrifugation at 13000 g for 4 minutes at 4°C (fixed rotor Eppendorf centrifuge 5415 R, Hamburg, Germany), plasma aliquots are diluted in deuterium oxide (D2O, 99.9%, Cambridge Isotope Laboratories Inc, Andover, USA) containing 180 µg/µl trimethylsilyl-2,2,3,3-tetradeuteropropionic acid (TSP, 98%, Cambridge Isotope Laboratories Inc, Andover, USA) as a chemical shift reference. Finally, the prepared plasma samples are transferred into a 5 mm NMR tube and analyzed.

1H-NMR analyses and assignment of present resonances The 1H-NMR spectra are recorded on a 400 Megahertz (MHz) NMR spectrometer (Varian/Agilent, Nuclear Magnetic Resonance Instruments, Palo Alto, California, USA) with a magnetic field strength of 9.4 Tesla at 294 K. Slightly T2- weighted spectra are acquired using a Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence to attenuate signals of macromolecules, such as proteins and polysaccharides. Additionally, water suppression is performed in order to allow optimal detection and quantification of low-molecular weight metabolites. The 1H-NMR spectra are phased manually, baseline corrected and referenced to the TSP resonance at 0.015 parts per million (ppm). The assignment of the present 1H-NMR resonances occurs by means of spiking experiments. A reference plasma sample is alternately spiked with 34 known metabolites with a concentration of 1 mg compound per 100 µl plasma. The obtained chemical shifts are double checked with Chenomx NMR suite software (Version 7.5, Chenomx Inc., Edmonton, Alberta, Canada). Finally, 1H-NMR spectra are divided in 112 spectral regions, which are integrated and normalized relative to the total integrated area of all spectral regions, irrespective of the remaining water, TSP, fructose and glucose resonances. The end result corresponds to 110 normalized integration regions (all integration regions except those of water and TSP).

Statistical analysis At first, the integration values of all 110 spectral regions are analyzed by means of a student t-test with correction for multiple testing by Benjamini-Hochberg to identify those which differ significantly between lung cancer patients and controls (IBM SPSS Version 20.0, Chicago, Illinois, USA). Secondly, multivariate statistical analyses are performed using SIMCA-P+ (Version 12.0, Umetrics, Umea, Sweden) to investigate whether the metabolic composition of blood plasma allows to discriminate between lung cancer patients and controls. An unsupervised principal component analysis (PCA) was performed to identify intrinsic clusters and outliers within the dataset. After the removal of outliers (detected by a Hotelling's T2 range plot), an orthogonal partial least squares discriminant analysis (OPLS-DA), an extension of partial least squares discriminant analysis (PLS-DA) with an integrated orthogonal signal correction filter, is performed to remove variability not relevant to class separation. The predicted classification is expressed as specificity (the percentage of controls that are actually classified as controls) and sensitivity (the percentage of lung cancer patients that are actually classified as lung cancer patients). The outcome of using the integration values of the significantly different spectral regions, obtained by the student t-test with correction for multiple testing by Benjamini-Hochberg, is compared to the outcome in which the integration values of all 110 spectral regions were used.

Inclusion Criteria:

• Diagnosis of a new lesion in the lung

Exclusion Criteria:

• a prior diagnosis of cancer in the past
• Not fasted for at least 6 hours
• Plasma glucose concentration ≥ 200 mg/dl
• Intake of medication at the day of investigation
• History/treatment of cancer in the previous 5 years

Study population: patients with a new diagnosis of lung cancer Control population: matched subjects with no diagnosis of cancer