Molecular Medicine

, Volume 17, Issue 9–10, pp 980–989 | Cite as

Prognostic Potential and Tumor Growth-Inhibiting Effect of Plasma Advanced Glycation End Products in Non-Small Cell Lung Carcinoma

  • Babett BartlingEmail author
  • Hans-Stefan Hofmann
  • Antonia Sohst
  • Yvonne Hatzky
  • Veronika Somoza
  • Rolf-Edgar Silber
  • Andreas Simm
Open Access
Research Article


The plasma fluorescence related to the standard fluorescence of advanced glycation end products (AGEs) is a simple measurable blood parameter for distinct diseases but its importance in human cancer, including non-small cell lung carcinoma (NSCLC), is unknown. Plasma samples of 70 NSCLC patients who underwent resection surgery of the tumor were analyzed for the distinct AGE-related fluorescence at 370 nm excitation/440 nm emission. In a retrospective study, we tested the prognostic relevance of this AGE-related plasma fluorescence. The effect of circulating AGEs on the NSCLC growth was studied experimentally in vitro and in vivo. NSCLC patients with high (> median) AGE-related plasma fluorescence were characterized by a later reoccurrence of the tumor after curative surgery and a higher survival rate compared with patients with low plasma fluorescence (25% versus 47% 5-y survival, P = 0.011). Treating NSCLC cell spheroids with patients’ plasma showed an inverse correlation between the growth of spheroids in vitro and the individual AGE-related fluorescence of each plasma sample. To confirm the impact of circulating AGEs on the NSCLC progression, we studied the NSCLC growth in mice whose circulating AGE level was elevated by AGE-rich diet. In vivo tumorigenicity assays demonstrated that mice with higher levels of circulating AGEs developed smaller tumors than mice with normal AGE levels. The AGE-related plasma fluorescence has prognostic relevance for NSCLC patients in whom the tumor growth-inhibiting effect of circulating AGEs might play a critical role.


Despite surgical resection of non-small cell lung carcinomas (NSCLCs), many patients die of recurrent tumors; and recent staging of the NSCLC patients by TNM classification is not sufficient for assessing prognosis and selecting therapy after surgery (1,2). Thus, additional analyses of the NSCLC patients are needed to specify tumor diagnosis and patient prognosis. At present, immunohistochemical analyses of the primary tumor support the TNM classification. Since these analyses require tissue material taken from the primary tumor, the analysis of easily accessible and always available specimens may be more favorable in staging NSCLCs. Therefore, our study focused on the prognostic relevance of a simple fluorescence spectrometry of plasma samples derived from NSCLC patients. The fluorescence of either plasma or serum samples is regularly studied in conjunction with the quantification of circulating advanced glycation end products (AGEs) (3, 4, 5, 6, 7). AGEs are generated in the final stages of the Maillard reaction in foods and in biological systems with amino groups of peptides, proteins, amino acids and some phospholipids, reacting with carbonyl groups of reducing sugars, degradation products of carbohydrates, lipids or ascorbic acid (8). Since Maillard-type reactions are highly complex, AGEs exist in a large variety of fluorescent and non-fluorescent chemical compounds, and only for some of them is the chemical structure currently known (8). Therefore, fluorescing AGEs are broadly analyzed at a standard AGE-related fluorescence (370 nm excitation/440 nm emission) to obtain a representative AGE level in biological samples (9).

The pathophysiological increase in the AGE-related fluorescence of plasma or serum samples has been found particularly in patients with chronic renal failure (3,6) and untreated diabetes mellitus (4,5). The generation of AGEs also is associated with oxidative stress (10,11), tobacco smoking (12) and an impaired detoxification of AGE precursors (13). Furthermore, the physiological amount of circulating AGEs depends on nutritional habits including an increased intake of AGE-rich (thermally processed) food (7,14,15) and sugar molecules with higher responsiveness to AGE formation (fructose instead of glucose) (16). Thus, several intrinsic and extrinsic factors determine the biological level of AGEs for each person. Elevated AGE levels are discussed mostly to play an adverse role in the process of degenerative diseases (15,17), but the threshold between physiological and pathophysiological range of the circulating AGEs has not yet been defined. Moreover, a number of nutritional studies indicate a favorable effect of food-derived AGEs, such as an enhanced antioxidant capacity and increase in chemopreventive enzymes (18, 19, 20).

The importance of AGEs in lung carcinomas is studied for extracellular matrix proteins showing an impact of the AGE-modified tissue matrix on the invasive migration of cancer cells (21). This observation is also supported by clinicopathological studies indicating a reduced tumor progression in response to age (22) and diabetes mellitus (23, 24, 25), both of which are characterized by an increase in AGE-modified tissue matrix (26). In contrast, the clinical importance of the circulating AGEs in patients with lung carcinomas or other types of cancer is largely unknown. Therefore, we were interested in the importance of the AGE-related fluorescence for the prognosis of NSCLC patients after curative surgery and, if so, in the impact of circulating AGEs on the NSCLC growth.

Material and Methods

Patients and Animals

Plasma samples were obtained from 78 consecutive NSCLC patients who underwent a pulmonary resection surgery during the hospitalization prior to the surgical procedure. Eight patients were excluded from the analysis because they either died due to a non-tumor-related disease or within 30 d after surgery. Tumor stages were classified according to the TNM system (1) and patients’ data were recorded up to 5 years after surgery. This study was approved by the Ethics Committee of the Medical Faculty (Halle [Saale], Germany).

Blood and tissue samples were also obtained from female athymic mice (Navy Medical Research Institute [NMRI] nu/nu; Harlan Winkelmann, Borchen, Germany) who had received a specific diet for modifying the blood AGE level. All mice were kept under clean room conditions in sterile isolator cages. At an age of about 4 wks, 13 mice of the control group received standard diet (70% (w/w) protein-free standard chow [C1004; Altromin, Lage, Germany], 15% casein, 15% starch), whereas 13 mice of the AGE group received an AGE-rich diet (70% protein-free standard chow, 15% casein, 15% bread crust as AGE-rich compound). All mice were fed each day with an average of about 6 g of the diet per capita. The preparation and partial characterization of the AGE-rich bread crust from rye/wheat-mixed bread was described earlier (27). After a feeding period of 21 d, mice were killed by cervical dislocation. Blood was removed after thoracotomy and a drop of the blood was used for preparing blood films on glass slides. The remaining blood was transferred to microvettes. Lung and liver were removed, frozen in liquid nitrogen and stored at −80°C until use. The local Animal Care and Use Committee, Halle (Saale) approved this study.

In Vivo Tumorigenicity Assay

Female athymic mice (NMRI nu/nu; Harlan Winkelmann) were used for the in vivo tumorigenicity assay. According to the feeding experiments described above, mice received a specific diet for modifying the blood AGE level. At an age of 4 wks, 16 mice of the control group received a standard diet and 16 mice of the AGE group received an AGE-rich diet. After a prefeeding period of 14 d, NSCLC cells were injected subcutaneously into each upper flank of athymic mice. Thus, 32 tumors in each diet group were studied. The National Cancer Institute (NCI) cell line H358 (ATCC, Manassas, VA, USA) was used as NSCLC cell line. H358 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (Invitrogen, Karlsruhe, Germany) at 37°C in a humidified 10% CO2 atmosphere. For tumor cell injection, H358 cells were harvested from 80% confluent monolayers by brief trypsinization, washed and resuspended in ice cold PBS. Then, 2 × 106 H358 cells per site were implanted subcutaneously, and the respective diet was continued during the period of tumor growth. Length and width of the tumors were measured twice a week for the duration of the experiment using a sliding caliper. Individual tumor volume was calculated as the product of 4/3π × (length/2) × (width/2)2 that corresponds to the equation of an ellipsoid. Mice were killed at day 25 of the tumor growth, and the tumors were dissected carefully and weighed. One aliquot of the tumor tissue was frozen in liquid nitrogen and stored at −80°C until used. Another aliquot was fixed in 4% phosphate-buffered formaldehyde and embedded in paraffin. Throughout the experiment, all mice were kept under clean room conditions in sterile isolator cages. Xenograft experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals after approval of the local Animal Care Committee, Halle (Saale).

Blood Analyses

Blood films on glass slides were analyzed for the cytological composition of the blood samples following Giemsa staining and light microscopy. The AGE-specific fluorescence was determined for serum samples of the mice and plasma samples of the NSCLC patients. Mice serum was diluted 1:5 and human plasma was diluted 1:20 in PBS (optimal dilution was tested before); 100 µL of each sample was transferred per well of a black 96-well microplate (Greiner, Frickenhausen, Germany). AGE-related fluorescence was monitored in triplicate on a FLUOstar OPTIMA reader (BMG Labtechnologies, Offenburg, Germany) at 370-10 nm excitation and 440-10 nm emission. Glucose-modified bovine serum albumin (AGE-BSA) was used for creating an internal standard curve (28), and data are given as concentrations equivalent to AGE-BSA. AGE modifications of mice serum proteins were determined additionally by antibody detection of the nonfluorescing AGE variant N-ε-carboxymethyllysine (CML). Briefly, 1 µL of the serum sample was spotted in triplicate onto nitrocellulose membranes using the Minifold Spot Blot Unit (Schleicher & Schüll, Dassel, Germany). After blocking the membrane with 6% nonfat dry milk in TBS buffer containing 0.15% polyethylenesorbitan monolaurate, CML levels were detected by using a rabbit polyclonal anti-CML antibody (ED Schleicher, University of Tuebingen, Germany) followed by a horseradish peroxidase-conjugated anti-rabbit IgG antibody (Dianova, Hamburg, Germany) and enhanced chemiluminescence detection (Amersham Bioscience, Buckinghamshire, UK). The signals were evaluated by use of the LAS 3000 computer-based imaging system (FUJI Film; Tokyo, Japan) and AIDA 3.5 software (Raytest, Straubenhardt, Germany).

Cell Culture

For three-dimensional cell culture, spheroids were formed from H358 lung carcinoma cells in DMEM containing 5 mg/mL methylcellulose (Sigma, Deisenhofen, Germany) and 2.5% fetal calf serum (Invitrogen). Then, H358 cells were seeded to a final number of 300 cells per spheroid into a single well of a 96-well suspension culture plate (Greiner). The formation of H358 spheroid was completed after 24 h. Thereafter, plasma of 28 randomly selected NSCLC patients was added to a final concentration of 7.5%. We studied 10 spheroids per individual plasma for 2 d. The diameter of each spheroid was estimated by using the Axiovert M200 microscope (Carl Zeiss, Jena, Germany) equipped with Spot Camera and Metamorph 4.6.5. software (Visitron Systems, Puchheim, Germany) and subsequently averaged. The spheroid volume was calculated as a product of 4/3π × diameter3 and the mean increase in the spheroid volume was determined for the individual plasma samples in three time-independent experiments. Accordingly, we studied the effect of serum taken from mice with or without AGE-rich diet as well as AGE-modified FCS (AGE-FCS). AGE-FCS was generated by incubating FCS (Invitrogen) with 20 mmol/L glyoxal (Sigma, Deisenhofen, Germany) and control FCS without glyoxal at 37°C for 2 d. Thereafter, control FCS and AGE-FCS were extensively dialyzed against water using ZelluTrans dialysis tubing (Roth, Karlsruhe, Germany). The AGE-related fluorescence was determined as described above.

We also studied the effect of AGE-FCS on the cell proliferation and migration in two-dimensional assays. As the FCS supply of cells in two-dimensional cultures is better than in spheroids, we used less FCS in these assays (5% control FCS or AGE-FCS; tested before). H358 cell proliferation (5),000 cells per well of a 96-well culture plate) was spectrophotometrically determined after 3 d using the alamarBlue reagent (Biosource Europe, Nivelles, Belgium). Moreover, the migration/proliferation of H358 cells was studied in an in vitro wound scratch model. Six scratches per experiment were placed into a monolayer of confluent cells by a pipette tip to generate defined cell-free areas. The distances between the two edges of the migrating/proliferating cells were estimated microscopically. Ten distances per scratch were always measured and finally averaged.

Expression Analyses

For RNA expression analysis, total RNA was isolated with Trizol reagent (Invitrogen), cleaned up (RNeasy Mini Kit; Qiagen, Hilden, Germany) and subjected to high-density oligonucleotide microarray (Affymetrix, Santa Clara, CA, USA) as described previously for human lung tissues (29) and H358 cells (30). The intensity of the hybridization signals was scanned with the GeneArray Scanner 7G and analyzed by using the GeneChip Operating System. For protein expression analysis, standard immunoblot was performed as described (30). Monoclonal mouse anti-Ost-48 antibody was purchased from Santa Cruz Biotechnology (sc-74408; Santa Cruz, CA, USA).

Mouse Tissue Analyses

The vascularization of tumor tissues from the xenografts was determined for tissue sections by staining of endothelial cells for the von Willebrand Factor (vWF) using a rabbit polyclonal anti-vWF (F3520; Sigma, St. Louis, MO, USA) and labeled-streptavidin-biotin detection (DAKO, Hamburg, Germany). Slides were embedded and the vessel density per area was evaluated for two sections each tumor by using the Axiovert microscope (Carl Zeiss) equipped with Spot Camera and Metamorph 4.6.5. software (Visitron Systems). In addition, we isolated and quantified the total protein of mouse lung and liver by SDS lysis. This protein was subjected to the slot blot procedure and subsequent antibody staining for carboxymethyl lysine (CML) as entirely described for the mouse serum. Data were calculated as arbitrary units (aU) per μL of the protein.


Patient data are given either as the median (M) with interquartile range or as the total or percentage number (n). Experimental data are reported as mean ± SD or SEM. All statistical calculations and data presentations were performed by use of the SigmaStat 3.5 and SigmaPlot 10 software (Systat Software Inc., San Jose, CA, USA). The appropriated statistical test is given in the legend of tables and figures respectively. P values ≤0.05 indicate a significant difference.


Plasma Fluorescence and Patients’ Outcome

The retrospective clinical study included plasma samples from NSCLC patients treated by pulmonary resection surgery. None of the patients had distant metastasis, resection stage R2 or died owing to a non-tumor-related disease. The plasma fluorescence was measured at the standard AGE-related fluorescence (370 nm excitation/440 nm emission), and the median plasma fluorescence had been determined with 1.08 µg/mL AGE-BSA (range: 0.73 to 1.61 µg/mL AGE-BSA). According to this median value, the NSCLC patients were divided into patients with either low or high plasma fluorescence. Subsequent analyses of the clinicalpathological data did not find significant differences for age, gender, smoking status, secondary diseases (diabetes mellitus, renal failure) and NSCLC-related parameters such as T and N stages (Table 1). In contrast, we revealed that NSCLC patients with high plasma fluorescence had a significantly better outcome (relapsefree interval and survival) after curative surgery than patients with low plasma fluorescence (Figure 1A). Even within the group of NSCLC patients without nodal involvement (N0), the 5-year survival had been determined as only 30% for patients having low plasma fluorescence compared with a 5-year-survival rate of about 70% for patients with high plasma fluorescence (P = 0.015).
Table 1

Patients’ characteristics depending on the plasma AGE-related fluorescence (FL).a

NSCLC patients parameter


Plasma FL



Lowb (n = 36)

Highb (n = 34)

P valuec

Plasma FL, μg/mL AGE-BSA


0.81 (0.73–0.95)

1.29 (1.21–1.61)


Age, y


69 (60–72)

68 (60–73)














Body mass index


26 (24–28)

27 (24–29)


NSCLC histology

















Tumor (T) stage

















Nodal (N) stage

















Resection (R) stage





























Diabetes mellitus












Blood glucose, mmol/L


5.5 (4.9–7.0)

5.3 (4.9–6.9)


Renal failure












Blood creatinine, μmol/L


79 (69–85)

77 (70–85)


Blood urea, mmol/L


5.3 (4.6–6.5)

5.7 (4.6–6.3)


aData are either given as median (M) with interquartile range or as number of NSCLC patients (n).

bCutoff point at the median plasma FL of all NSCLC patients (1.08 µg/mL AGE-BSA).

cP values were calculated by use of the Student t test (median differences) or the χ2 test (homogeneity of variances) as appropriated.

dADC, adenocarcinoma.

eSCC, squamous cell carcinoma.

Figure 1

Clinical outcome of NSCLC patients with differential levels of AGE-related plasma fluorescence (FL). (A) Kaplan-Meier plot of the 4-year tumor-free interval and the 5-year postoperative survival of NSCLC patients with either low or high plasma fluorescence. (B) NSCLC patients had been further subdivided into quartiles (QI to QIV) according to the level of their plasma fluorescence and the survival curves were calculated for each subgroup. The median plasma FL (μg/mL AGE-BSA) is in QI: 0.73, QII: 0.94, QIII: 1.21 and QIV: 1.61. Significance between the Kaplan-Meier curves was tested by log-rank statistic. All pair-wise multiple comparison procedures calculated a critical P value with <0.008 (Bonferroni method).

On the basis of our observation (see Figure 1A and Table 1), NSCLC patients were also subdivided into quartiles (QI to QIV) according to the level of their plasma fluorescence. Subsequent Kaplan-Meier analyses showed that particularly patients with very low AGE fluorescence (QI) had the lowest survival rate (Figure 1B). Because QI patients did not show a lower body mass index (median BMI of 25.6 compared with median BMI of 26.0 for all patients), a low dietary intake of AGEs due to malnutrition is unlikely for this patient group. Although the survival of NSCLC patients was improving along with an increasing plasma fluorescence (QI to QIII), the survival of patients with a very high plasma fluorescence (QIV) was diminished again (see Figure 1B).

Plasma Fluorescence, Advanced Glycation End Products and Tumor Growth In Vitro

Our clinical data suggest that plasma compounds mediating the AGE-related fluorescence may have an influence on the NSCLC progression. Therefore, we used the multicellular spheroid model to analyze the effect of patients’ plasma on the tumor growth in vitro depending on the individual’s plasma fluorescence. Tumor spheroids were formed from the NSCLC cell line H358, which is suitable for this in vitro growth assay because H358 cells form stable spheroids without simultaneous migration of single cells from the united cell structure (21). H358 spheroids were treated individually with plasma of randomly selected NSCLC patients from our clinical study group (Table 1). Subsequent analyses of the spheroid growth revealed an inverse correlation between tumor spheroid growth in vitro and the level of the AGE-related fluorescence of the individual plasma sample (Figure 2A).
Figure 2

Effect of plasma from NSCLC patients or AGE-modified FCS on the growth of NSCLC cells in vitro. (A) H358 spheroids were treated with the individual plasma of randomly selected NSCLC patients. The spheroid growth after 2 d was plotted per AGE-related fluorescence (FL) of the individual plasma sample and subjected to linear regression analysis. (B) The proliferation and migration of H358 cells (cell proliferation and wound scratch assay) as well as (C) the spheroid growth of H358 cells was analyzed after 2 d in response to AGE-modified versus control (unmodified) FCS. Mean data ± SD are given. P values were calculated by the Student t test.

This inverse correlation could be caused by AGE modifications of plasma compounds, such as growth factors, thereby impairing their functionality. Therefore, we studied the effect of AGE-FCS (AGE-related fluorescence corresponds to 4.35 µg/mL AGE-BSA versus 1.01 µg/mL AGE-BSA in control FCS) on the proliferation and migration of H358 cells in a proliferation and wound scratch assay, respectively. These assays showed a reduced cell proliferation and migration in response to AGE-FCS (Figure 2B). The antiproliferative effect of AGE-FCS is still more obvious by decreasing the FCS concentration (2% instead of 5%; data not shown). An impact of AGE-FCS on the viability of H358 cells has not been observed. The growth of H358 spheroids also was influenced strongly by AGE-FCS (Figure 2C).

As plasma AGEs also could mediate direct biological functions by interacting with defined receptors, we additionally studied the expression of AGE receptors in human lung carcinoma. This study showed a high mRNA expression of the AGE-Rs 1, 2 and 3 in H358 cells and NSCLC tissues. While both AGE-R1 and AGE-R2 are increased in NSCLC tissues compared with control (healthy) lung, AGE-R3 is not differentially expressed in NSCLC (Table 2). The high expression of AGE-R1 in NSCLC also has been observed by immunoblot analysis (data not shown). In contrast, RAGE was down-regulated significantly in H358 cells and NSCLC tissues compared with control lung at mRNA (Table 2) and protein level (31).
Table 2

mRNA expression of AGE receptors in human lung cancer.

AGE receptors

NSCLC cell line H358a (n= 3)

Lung tissuesb

P value

NSCLC (n = 89)

Control (n = 15)

Oligosaccharyl transferase-48 (OST-48, AGE-R1)

1,473 ± 105

775 (720–867)

658 (609–693)


80K-H phosphoprotein (AGE-R2)

3,295 ± 188

351 (316–391)

277 (257–314)


Galectin-3 (AGE-R3)

10,600 ± 1,366

1,209 (1),017–1,392)

1,103 (1),093–1,030)


Receptor for AGEs (RAGE)

84 ± 27

308 (226–445)

1,399 (1),201–1,639)


aData of four independent experiments are given as mean ± SD.

bMedian data are given with interquartile range and P values (NSCLC versus control lung) were calculated by use of the Student t test.

Advanced Glycation End Products and Tumor Growth In Vivo

Our in vitro data suggest an association between the plasma level of fluorescing AGEs and the tumor growth (see Figures 2AC). Therefore, we established a xenograft model (athymic NMRI nu/nu mice) with a physiologically increased level of circulating AGEs. In the first experiment, athymic mice were fed with AGE-rich food (32,33) without injection of the NSCLC cell line H358 (group 1, Figure 3A). After 21 d of the specific diet, we removed blood and tissue from lung and liver for further analyses. In the case of mice, serum instead of plasma was prepared because of the immediate blood coagulation after thoracotomy. This initial feeding experiment showed that AGE-rich diet mediates an increased level of fluorescent AGEs in the mouse serum compared with serum from mice receiving control diet (Table 3). We also determined a slightly increased serum level of the nonfluorescing AGE variant N-ε-carboxymethyllysine (CML) in response to the AGE-rich diet, whereas the tissue level of CML was not influenced by the diet (Table 3). The AGE-rich diet had no impact on the weight and the number of white blood cells (granulocytes, lymphocytes, monocytes) (Table 3).
Table 3

Characteristics of group 1 athymic NMRI nu/nu mice with AGE-rich diet.a





Control (n = 13)

AGE (n = 13)

P value



Start, g

12 ± 1.5

13 ± 2.5


End, g

20 ± 1.6

21 ± 1.9





348 ± 65

402 ± 87



46 ± 1.9

49 ± 2.3




Granulocytes, %

26 ± 13

27 ± 13


Lymphocytes, %

72 ± 13

72 ± 15


Monocytes, %

1.8 ± 1.9

1.8 ± 2.0




Lung CML, a∪/μg

12 ± 2.7

12 ± 2.5


Liver CML, a∪/μg

11 ± 4.1

9.7 ± 2.0


aAll data are given as mean ± SD. P values were calculated by use of the Student t test.

Our previous studies already demonstrated an antiproliferative effect of AGE-rich food in H358 cells (34). As the gastrointestinal digestion alters the food-AGEs regarding quality and quantity, we now treated NSCLC spheroids from H358 cells with the individual serum of mice receiving either AGE-rich or control diets. In accordance with the in vitro experiments using patients’ plasma (see Figure 2A), we determined an inverse correlation between the growth of H358 spheroids in vivo and the level of AGE-related fluorescence of the individual serum sample (Figure 3B, above). This inverse correlation was related mainly to the serum from mice receiving AGE-rich diet (see Figure 3B, above). In contrast, correlating the growth of H358 spheroids in vivo and the CML level of the individual serum did not reach statistical significance (Figure 3B, below).
Figure 3

Effect of mice serum on the growth of lung carcinoma cells in vitro depending on the serum AGE level. (A) Summary of the athymic mice used for studying the influence of AGEs on tumor growth in vitro and in vivo. Group 1 mice received either control or AGE diet without injection of H358 lung carcinoma cells. Group 2 mice received either control or AGE diet with subcutaneous injection of H358 cells. (B) Spheroids of H358 cells were treated with the individual serum of mice receiving either control or AGE-rich diet. The spheroid growth after 2 d was plotted per level of the AGE-related fluorescence (FL; left) or CML (right) of the individual serum and subjected to linear regression analysis.

In the second experiment, athymic mice were fed with an AGE-rich diet to study the impact of circulating AGEs on the tumor growth of H358 cells in vivo (group 2, Figure 3A). The in vivo tumorigenicity assay demonstrated that the H358 tumor growth is less pronounced in the group of mice receiving the AGE-rich diet compared with the control group (Figure 4A, above). In other words, mice with an AGE-poor diet (control) developed more frequently large tumors than mice with an AGE-rich diet (Figure 4A, below). In accordance with the different tumor volume, the mean weight of H358 tumors was reduced significantly in response to the AGE-rich diet, whereas the tumor vascularization was not influenced (Figure 4B).
Figure 4

Effect of AGE-rich diet on the growth of lung carcinoma cells in vivo. (A) Tumor growth of H358 cells subcutaneously implanted into athymic mice receiving either control or AGE-rich diet. The individual regression coefficient a was calculated as determinant of the tumor growth per time. In addition, H358 tumors were subdivided in small and large tumors (cut off at the median volume), and the total number and percentage of tumors with either small or large size was calculated. (B) Weight and vascularization of H358 tumors in mice with or without AGE-rich diet. Data are given as mean ± SEM. P values were calculated by the Student t test.


Plasma or sera prepared from blood samples change in its fluorescence at 370 nm excitation/440 nm emission, which relates to the standard fluorescence of AGEs (4, 5, 6, 7,35). Our clinical study demonstrated a prognostic relevance of the AGE-related plasma fluorescence in NSCLC patients after curative surgery. One reason for its prognostic relevance might be an impact of circulating AGEs on the tumor growth as suggested by the results from our experimental studies. To date, such blood measurements of fluorescent chromophores have not been considered in the clinical analysis of tumor patients. The main advantage of using the AGE-related fluorescence as a prognostic marker is its simple and fast application via a single-wavelength measurement of blood samples.

We determined highly different levels in the AGE-related fluorescence of plasma samples from NSCLC patients and patients having high AGE-related plasma fluorescence showed a significantly higher survival rate after curative surgery than patients having low AGE-related plasma fluorescence. While only 25% of the patients who had low plasma fluorescence survived 5 years after surgery, 47% of the patients with high plasma fluorescence survived at this time. A similar tendency was observed for the relapse-free period after surgery. The highest survival rate after 5 years (60%) has been determined for NSCLC patients having a moderately increased AGE-related fluorescence of plasma samples, whereas this positive correlation was partially abolished in patients with a very high AGE-related fluorescence (35%). Although their 5-year survival was still significantly higher than the survival of patients with very low AGE-related plasma fluorescence (10%), the concentration-dependent observation supports the pathophysiological role of highly increased AGE levels (17,36,37). It is well conceivable that an excess of AGEs induces oxidative stress by exhaustion of antioxidative defense systems (10,11), thereby supporting the tumor progression again. This relation has been suggested recently for patients with advanced, compared with early, breast cancer (38). In this context, pathophysiologically high levels in fluorescent AGEs also have been described for blood samples derived from smokers (12) and patients with chronic renal failure (3,6,35) or untreated diabetes mellitus (4,5). However, none of these factors affected the AGE-related plasma fluorescence in our patients’ group. Some contrasting data regarding the plasma level of AGEs may be explained by the changing smoking behaviors of patients during the time of hospitalization, the few number of patients with renal failure, the well-controlled medical regime in the hospital (diabetes patients) and the concomitant NSCLC disease.

Our clinical observation suggested a protective effect of plasma AGEs on the tumor progression. This can be supported indirectly by previous studies describing an inhibiting effect of reactive carbonyls, which are potential substances for AGE formation, on tumor growth in vivo (39, 40, 41). However, in breast cancer, elevated plasma levels of protein carbonyls were associated with an increased cancer progression (38). Although carbonyls also might affect tumor growth via cellular actions different from AGE actions (42,43), further findings again support the antitumorigenic effect of AGEs. In this regard, experimental and pathological studies especially indicate an effect of extracellular tissue matrix modified with AGEs on tumor progression (21,23).

Quantity and quality of the large number of AGE variants are suggested as crucial factors that finally determine the biological role of AGEs (8,17,19), but what does cause the different levels of AGEs, including fluorescent AGEs, in human plasma? Besides dietary habits (14,17,44), a large number of endogenous factors influence the level of fluorescent AGEs in human blood circulation. This includes the degree of glucose and lipid metabolism with the subsequent formation of reactive carbonyls (26), the activity of enzymatic detoxification systems (45), oxidative stress (11), receptor-mediated AGE uptake by cells (46), targeted degradation of AGE-modified proteins by the proteasomal system (47) and renal excretion (48). In normal conditions, generation and clearance of AGEs are in balance sustaining an AGE homeostasis (17). Low plasma levels of fluorescent AGEs, as observed for our NSCLC patients’ group with worse outcome after curative surgery, may result from a low metabolic formation of AGEs, hypoalbuminemia, high detoxification of reactive carbonyls, low dietary intake of AGEs and/or high degradation and excretion of AGEs.

In contrast, moderately higher levels in plasma AGEs may result from a higher metabolic formation of AGEs and/or dietary intake of AGEs. The moderate increase in the AGE-related plasma fluorescence indicated a better disease-free survival of NSCLC patients, suggesting a protective effect of circulating AGEs and/or secondary factors accompanied by an increased level of circulating AGEs on the reoccurrence and progression of NSCLC. Therefore, plasma samples of NSCLC patients were subjected to in vitro studies. In correspondence with the prognostic relevance of fluorescent AGEs, treating NSCLC spheroids with patients’ plasma showed an inverse correlation between the growth of spheroids and the individual AGE-related fluorescence of each sample. This inverse correlation of fluorescent AGEs with the growth of NSCLC spheroids also has been observed for blood samples from mice, whose circulating AGE level was moderately elevated by an established dietary model using AGE-rich bread crust (32,33). AGE-rich bread crust did not induce oxidative stress in NSCLC cells (34), and the diet with bread crust caused an increase in circulating AGE chromophores of 15% that corresponds approximately with the increase in plasma AGE chromophores determined for NSCLC patients with moderately higher levels (1.21 [µg/mL AGE-BSA versus the cutoff value at 1.08 [µg/mL AGE-BSA). Although the AGE-rich diet also caused a small increase in the nonfluorescent AGE variant CML, the level of plasma CML did not correlate with the spheroid growth. This might be based on the insignificant impact of the single compound CML on the cell proliferation compared with fluorescent AGEs as a set of AGE variants.

Nonetheless, our in vitro data indicate the impact of plasma rich in fluorescent AGEs on the NSCLC growth. This has been supported by our in vivo studies showing that the chronic intake of AGE-rich food was associated with a reduced subcutaneous growth of NSCLC cells in about 30% of athymic mice (final tumor weight; number of large tumors). The reduced tumor growth of NSCLC cells in mice without affecting tumor vascularization as well as the inverse correlation between spheroid growth of NSCLC cells and the AGE-related fluorescence of plasma samples, suggest the antiproliferative effect of AGEs. Moreover, circulating AGEs also may have an antimigrative effect as suggested by the data of our wound scratch assay. Although the detailed chemical structures, digestion and absorption of food-derived AGEs are still incompletely understood, it seems to be clear that particularly low-molecular weight AGEs pass the colon (49,50). Moreover, other compounds than the food-derived AGE chromophores may contribute to the biological effect, because an increased intake of AGEs could also modify the AGE generation in vivo and other factors. In this context, cell surface receptors are identified, which are believed to mediate the biological function of extracellular AGEs (51), among these are the receptors AGE-R1, better known as OST-48, and RAGE, which are the best studied AGE-binding molecules. While the expression of RAGE is strongly downregulated in NSCLC cells and tissues (31), we identified a high expression of AGE-R1, whose activation by AGE-modified serum albumin suppresses the cell signaling cascade induced by the epidermal growth factor (EGF) (52). As EGF-mediated cell signaling contributes to the progression of lung carcinomas (53), the impairment of the EGF receptor signaling by AGE-R1 activation might, at least in part, explain the tumor growth-inhibiting effect of circulating AGEs. However, AGEs also may act independently of AGE receptors. In this regard, we recently showed a reduced activity of the soluble matrix metalloproteinase 2 as well as the growth factor PDGF following AGE modification (54,55). The reduced growth activity of AGE-modified circulating factors also has been demonstrated in this study by a reduced proliferation/ migration of NSCLC cells in response to AGE-FCS. Therefore, several types of biological mechanisms might contribute to the antitumorigenic effect of circulating AGE chromophores.

Moreover, it is conceivable that no singular AGE compound acts alone but in concert with other, probably still unknown, AGE chromophores. The quantification of a set of fluorescent AGEs as a prognostic factor is therefore an advantage compared with the quantification of single compounds in human plasma samples. Nonetheless, the question arises, “which of the AGE structures contribute to the tumor growth-inhibiting effect of AGEs?” Pronyl-lysine is one of the few AGEs presently identified as a chemopreventive factor in bread crust (27), which also affects the formation and progression of chemically induced preneoplastic lesions in rat colon (56). This colon cancer model is not completely consistent with the tumorigenicity model used in our study, but it indicates the potential role of this AGE compound in cancer development. In addition, 3-hydroxy-4-[(E)-(2-furyl)methylidene]methyl-3-cyclopentene-1,2-dione and others have been identified as AGE chromophores formed from carbohydrates upon thermal food processing which impair the proliferation of human mammary carcinoma cells in vitro (57). Although these and our studies indicate the beneficial effect of AGEs in tumorigenicity, it does not exclude the importance of other factors that are directly or indirectly accompanied by an altered AGE level in human.

In summary, we demonstrated that the AGE-related plasma fluorescence is a simple predictor for the outcome of NSCLC patients after curative surgery. The worst prognosis was associated with patients having a low level in the AGE-related fluorescence, which might be based on the tumor growth-inhibiting effect of fluorescent AGEs as suggested by the results of our experimental studies. Since extent and prognosis of NSCLC patients are insufficiently assessed by the current TNM classification, additional analyses of the individual plasma fluorescence may support the patients’ assessment significantly.


The authors declare that they have no competing interests as defined by Molecular Medicine or other interests that might be perceived to influence the results and discussion reported in this paper.



The authors thank Renate Donath, Stephanie Tuche and Sabine Koitzsch for technical assistance and Peter Schmidt (Medical Faculty, Halle [Saale]) for clinical assistance. We also thank Gesine Hansen (Hannover Medical School, Germany), Stefan Burdach (Technical University Munich, Germany) and Vesselin Christov (Medical Faculty, Halle [Saale]) for cooperation with the microarray analysis. This study was supported by Deutsche Krebshilfe (107078), Deutsche Forschungsgemeinschaft (Si-1317/1-1) and Wilhelm Roux grants (FKZ 7/04, FKZ14/14).


  1. 1.
    Wittekind C, Meyer H-J, Bootz F. (2003) TNM classification of malignant tumours. Union Internationale Contre le Cancer. In: Cancer UICl (ed.). Springer, Berlin Heidelberg New York, p. 223.Google Scholar
  2. 2.
    Goldstraw P, et al. (2007) The IASLC lung cancer staging project: proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM classification of malignant tumours. J. Thorac. Oncol. 2:706–14.CrossRefPubMedGoogle Scholar
  3. 3.
    Wrobel K, Wrobel K, Garay-Sevilla ME, Nava LE, Malacara JM. (1997) Novel analytical approach to monitoring advanced glycosylation end products in human serum with on-line spectrophotometric and spectrofluorometric detection in a flow system. Clin. Chem. 43:1563–9.PubMedGoogle Scholar
  4. 4.
    Muench G, et al. (1997) Determination of advanced glycation end products in serum by fluorescence spectroscopy and competitive ELISA. Eur. J. Clin. Chem. Clin. Biochem. 35:669–77.Google Scholar
  5. 5.
    Yanagisawa K, et al. (1998) Specific fluorescence assay for advanced glycation end products in blood and urine of diabetic patients. Metabolism. 47:1348–53.CrossRefPubMedGoogle Scholar
  6. 6.
    Sebekova K, et al. (2001) Plasma levels of advanced glycation end products in children with renal disease. Pediatr. Nephrol. 16:1105–12.CrossRefPubMedGoogle Scholar
  7. 7.
    Uribarri J, et al. (2007) Circulating glycotoxins and dietary advanced glycation endproducts: two links to inflammatory response, oxidative stress, and aging. J. Gerontol. A Biol. Sci. Med. Sci. 62:427–33.CrossRefPubMedPubMedCentralGoogle Scholar
  8. 8.
    Chuyen NV. (2006) Toxicity of the AGEs generated from the Maillard reaction: on the relationship of food-AGEs and biological-AGEs. Mol. Nutr. Food. Res. 50:1140–9.CrossRefPubMedGoogle Scholar
  9. 9.
    Schmitt A, Schmitt J, Munch G, Gasic-Milencovic J. (2005) Characterization of advanced glycation end products for biochemical studies: side chain modifications and fluorescence characteristics. Anal. Biochem. 338:201–15.CrossRefPubMedGoogle Scholar
  10. 10.
    Scivittaro V, Ganz MB, Weiss MF. (2000) AGEs induce oxidative stress and activate protein kinase C-beta(II) in neonatal mesangial cells. Am. J. Physiol. Renal Physiol. 278:F676–83.CrossRefPubMedGoogle Scholar
  11. 11.
    Grimsrud PA, Xie H, Griffin TJ, Bernlohr DA. (2008) Oxidative stress and covalent modification of protein with bioactive aldehydes. J. Biol. Chem. 283:21837–41.CrossRefPubMedPubMedCentralGoogle Scholar
  12. 12.
    Cerami C, et al. (1997) Tobacco smoke is a source of toxic reactive glycation products. Proc. Natl. Acad. Sci. U. S. A. 94:13915–20.CrossRefPubMedPubMedCentralGoogle Scholar
  13. 13.
    Miyata T, van Ypersele de Strihou C, Kurokawa K, Baynes J. (1999) Alterations in non-enzymatic biochemistry in uremia: origin and significance of “carbonyl stress” in long-term uremic complications. Kidney Int. 55:389–99.CrossRefPubMedGoogle Scholar
  14. 14.
    Goldberg T, et al. (2004) Advanced glycoxidation end products in commonly consumed foods. J. Am. Diet. Assoc. 104:1287–91.CrossRefPubMedGoogle Scholar
  15. 15.
    Tessier FJ, Birlouez-Aragon I. (2010) Health effects of dietary Maillard reaction products: the results of ICARE and other studies. Amino Acids. 2010, Oct 15 [Epub ahead of print]Google Scholar
  16. 16.
    Bunn HF, Higgins PJ. (1981) Reaction of monosaccharides with proteins: possible evolutionary significance. Science. 213:222–4.CrossRefPubMedGoogle Scholar
  17. 17.
    Vlassara H. (2005) Advanced glycation in health and disease: role of the modern environment. Ann. N. Y. Acad. Sci. 1043:452–60.CrossRefPubMedGoogle Scholar
  18. 18.
    Esposito F, et al. (2003) Moderate coffee consumption increases plasma glutathione but not homocysteine in healthy subjects. Aliment. Pharmacol. Ther. 17:595–601.CrossRefPubMedGoogle Scholar
  19. 19.
    Somoza V. (2005) Five years of research on health risks and benefits of Maillard reaction products: an update. Mol. Nutr. Food Res. 49:663–72.CrossRefPubMedGoogle Scholar
  20. 20.
    Ruhs S, et al. (2011) Preconditioning with Maillard reaction products improves antioxidant defence leading to increased stress tolerance in cardiac cells. Exp. Gerontol. 45:752–62.CrossRefGoogle Scholar
  21. 21.
    Bartling B, Desole M, Rohrbach S, Silber R-E, Simm A. (2009) Age-associated changes of extracellular matrix collagen impair lung cancer cell migration. FASEB J. 23:1510–20.CrossRefPubMedGoogle Scholar
  22. 22.
    Ershler WB, Socinski MA, Greene CJ. (1983) Bronchogenic cancer, metastases, and aging. J. Am. Geriatr. Soc. 31:673–6.CrossRefPubMedGoogle Scholar
  23. 23.
    Nerlich AG, Hagedorn HG, Boheim M, Schleicher ED. (1998) Patients with diabetes-induced microangiopathy show a reduced frequency of carcinomas. In Vivo. 12:667–70.PubMedGoogle Scholar
  24. 24.
    De Giorgio R, Giovanni B, Cecconi A, Corinaldesi R, Mancini A. (2000) Diabetes is associated with longer survival rates in patients with malignant tumors. Arch. Intern. Med. 160:2217.CrossRefPubMedGoogle Scholar
  25. 25.
    Bartling B, Simm A, Sohst A, Silber R, Hofmann H. (2011) Effect of diabetes mellitus on the outcome of patients with resected non-small cell lung carcinoma. Gerontology. 2011, Mar 11 [Epub ahead of print.]Google Scholar
  26. 26.
    Ulrich P, Cerami A. (2001) Protein glycation, diabetes, and aging. Recent Prog. Horm. Res. 56:1–21.CrossRefPubMedGoogle Scholar
  27. 27.
    Lindenmeier M, Faist V, Hofmann T. (2002) Structural and functional characterization of pronyl-lysine, a novel protein modification in bread crust melanoidins showing in vitro antioxidative and phase I/II enzyme modulating activity. J. Agric. Food Chem. 50:6997–7006.CrossRefPubMedGoogle Scholar
  28. 28.
    Simm A, et al. (1997) Advanced glycation end-products stimulate the MAP-kinase pathway in tubulus cell line LLC-PK1. FEBS Lett. 410:481–4.CrossRefPubMedGoogle Scholar
  29. 29.
    Hofmann H-S, et al. (2004) Classification of human lung neoplasm by two target genes. Am. J. Respir. Crit. Care Med. 170:516–9.CrossRefPubMedGoogle Scholar
  30. 30.
    Bartling B, Rehbein G, Simm A, Silber RE, Hofmann HS. (2010) Porcupine expression is associated with the expression of S100P and other cancer-related molecules in non-small cell lung carcinoma. Int. J. Oncol. 36:1015–21.CrossRefPubMedGoogle Scholar
  31. 31.
    Bartling B, Hofmann H, Weigle B, Silber R, Simm A. (2005) Down-regulation of the receptor for advanced glycation end-products (RAGE) supports non-small cell lung carcinoma. Carcinogenesis. 26:293–301.CrossRefPubMedGoogle Scholar
  32. 32.
    Sebekova K, Faist V, Hofmann T, Schinzel R, Heidland A. (2003) Effects of a diet rich in advanced glycation end products in the rat remnant kidney model. Am. J. Kidney Dis. 41:S48–51.CrossRefPubMedGoogle Scholar
  33. 33.
    Somoza V, et al. (2005) Influence of feeding malt, bread crust, and a pronylated protein on the activity of chemopreventive enzymes and antioxidative defense parameters in vivo. J. Agric. Food Chem. 53:8176–82.CrossRefPubMedGoogle Scholar
  34. 34.
    Bartling B, Rehbein G, Somoza V, Silber RE, Simm A. (2005) Maillard reaction product-rich food impair cell proliferation and induce cell death in vitro. Signal Transduction. 6:303–13.CrossRefGoogle Scholar
  35. 35.
    Gerdemann A, et al. (2002) Plasma levels of advanced glycation end products during haemodialysis, haemodiafiltration and haemofiltration: potential importance of dialysate quality. Nephrol. Dial. Transplant. 17:1045–9.CrossRefPubMedGoogle Scholar
  36. 36.
    Henle T. (2007) Dietary advanced glycation end products: a risk to human health? A call for an interdisciplinary debate. Mol. Nutr. Food. Res. 51:1075–8.CrossRefPubMedGoogle Scholar
  37. 37.
    Birlouez-Aragon I, et al. (2010) A diet based on high-heat-treated foods promotes risk factors for diabetes mellitus and cardiovascular diseases. Am. J. Clin. Nutr. 91:1220–6.CrossRefPubMedGoogle Scholar
  38. 38.
    Tesarova P, et al. (2007) Carbonyl and oxidative stress in patients with breast cancer — is there a relation to the stage of the disease? Neoplasma. 54:219–24.PubMedGoogle Scholar
  39. 39.
    Egyud LG, Szent-Gyorgyi A. (1968) Cancerostatic action of methylglyoxal. Science. 160:1140.CrossRefPubMedGoogle Scholar
  40. 40.
    Ghosh M, et al. (2006) In vivo assessment of toxicity and pharmacokinetics of methylglyoxal. Augmentation of the curative effect of methylglyoxal on cancer-bearing mice by ascorbic acid and creatine. Toxicol. Appl. Pharmacol. 212:45–58.CrossRefPubMedGoogle Scholar
  41. 41.
    Ray M, Ghosh S, Kar M, Datta SC, Ray S. (2006) Implication of the bioelectronic principle in cancer therapy: treatment of cancer patients by methylglyoxal-based formulation. Cancer Therapy. 4:205–22.Google Scholar
  42. 42.
    Du J, et al. (2000) Methylglyoxal induces apoptosis in Jurkat leukemia T cells by activating c-Jun N-terminal kinase. J. Cell. Biochem. 77:333–44.CrossRefPubMedGoogle Scholar
  43. 43.
    Ray S, Dutta S, Halder J, Ray M. (1994) Inhibition of electron flow through complex I of the mitochondrial respiratory chain of Ehrlich ascites carcinoma cells by methylglyoxal. Biochem. J. 303 (Pt 1): 69–72.CrossRefPubMedPubMedCentralGoogle Scholar
  44. 44.
    Krajcovicova-Kudlackova M, Sebekova K, Schinzel R, Klvanova J. (2002) Advanced glycation end products and nutrition. Physiol. Res. 51:313–6.PubMedGoogle Scholar
  45. 45.
    Wu X, Monnier VM. (2003) Enzymatic deglycation of proteins. Arch. Biochem. Biophys. 419:16–24.CrossRefPubMedGoogle Scholar
  46. 46.
    Horiuchi S, Sakamoto Y, Sakai M. (2003) Scavenger receptors for oxidized and glycated proteins. Amino Acids. 25:283–92.CrossRefPubMedGoogle Scholar
  47. 47.
    Grune T, Davies KJ. (2003) The proteasomal system and HNE-modified proteins. Mol. Aspects Med. 24:195–204.CrossRefPubMedGoogle Scholar
  48. 48.
    Henle T, Miyata T. (2003) Advanced glycation end products in uremia. Adv. Ren. Replace. Ther. 10:321–31.CrossRefPubMedGoogle Scholar
  49. 49.
    Finot P, Magnenat E. (1981) Metabolic transit of early and advanced Maillard products. Prog. Fd. Nutr. Sci. 5:193–207.Google Scholar
  50. 50.
    Foerster A, Henle T. (2003) Glycation in food and metabolic transit of dietary AGEs (advanced glycation end-products): studies on the urinary excretion of pyrraline. Biochem. Soc. Trans. 31:1383–5.CrossRefPubMedGoogle Scholar
  51. 51.
    Thornalley PJ. (1998) Cell activation by glycated proteins. AGE receptors, receptor recognition factors and functional classification of AGEs. Cell. Mol. Biol. (Noisy-le-grand). 44:1013–23.Google Scholar
  52. 52.
    Cai W, He JC, Zhu L, Lu C, Vlassara H. (2006) Advanced glycation end product (AGE) receptor 1 suppresses cell oxidant stress and activation signaling via EGF receptor. Proc. Natl. Acad. Sci. U. S. A. 103:13801–6.CrossRefPubMedPubMedCentralGoogle Scholar
  53. 53.
    Domingo G, et al. (2010) EGF receptor in lung cancer: a successful story of targeted therapy. Expert Rev. Anticancer Ther. 10:1577–87.CrossRefPubMedGoogle Scholar
  54. 54.
    Bartling B, Desole M, Silber RE, Simm A. (2007) Dicarbonyl-mediated protein modifications affect matrix metalloproteinase (MMP) activity. Z. Gerontol. Geriatr. 40:357–61.CrossRefPubMedGoogle Scholar
  55. 55.
    Nass N, et al. (2010) Glycation of PDGF results in decreased biological activity. Int. J. Biochem. Cell. Biol. 42:749–754.CrossRefPubMedGoogle Scholar
  56. 56.
    Panneer Selvam J, Aranganathan S, Nalini N. (2008) Aberrant crypt foci and AgNORs as putative biomarkers to evaluate the chemopreventive efficacy of pronyl-lysine in rat colon carcinogenesis. Invest. New Drugs. 26:531–10.CrossRefPubMedGoogle Scholar
  57. 57.
    Marko D, et al. (2003) Maillard reaction products modulating the growth of human tumor cells in vitro. Chem. Res. Toxicol. 16:48–55.CrossRefPubMedGoogle Scholar

Copyright information

© The Feinstein Institute for Medical Research 2011

Authors and Affiliations

  • Babett Bartling
    • 1
    Email author
  • Hans-Stefan Hofmann
    • 1
    • 2
  • Antonia Sohst
    • 1
  • Yvonne Hatzky
    • 1
  • Veronika Somoza
    • 3
  • Rolf-Edgar Silber
    • 1
  • Andreas Simm
    • 1
  1. 1.Department of Cardio and Thoracic Surgery, University Hospital Halle (Saale)Martin Luther University Halle-WittenbergHalle (Saale)Germany
  2. 2.Department of Thoracic SurgeryHospital Barmherzige Brüder RegensburgRegensburgGermany
  3. 3.Research Platform for Molecular Food ScienceUniversity of ViennaViennaAustria

Personalised recommendations