Association between plasma proteome profiles

0 downloads 0 Views 360KB Size Report
Sep 2, 2011 - 2009). Mass spectrometry-based analysis of the serum proteome also has been performed on blood samples collected before and during RT of ...
padmavathym 9/2/11 16:35

TRAB_A_556174

(XML)

Int. J. Radiat. Biol., Vol. 87, No. 5, May 2011, pp. 1–9

60 5

10

Association between plasma proteome profiles analysed by mass spectrometry, a lymphocyte-based DNA-break repair assay and radiotherapy-induced acute mucosal reaction in head and neck cancer patients

65

70 15

20

25

30

35

40

´ SKA2*, ANNA WALASZCZYK1, MONIKA PIETROWSKA1*, JOANNA POLAN 1 ANDRZEJ WYGODA , TOMASZ RUTKOWSKI1, KRZYSZTOF SKŁADOWSKI1, ŁUKASZ MARCZAK3, MACIEJ STOBIECKI3, MICHAŁ MARCZYK2, ´ SKI2,4, & PIOTR WIDŁAK1 ANDRZEJ POLAN Maria Skłodowska-Curie Memorial Cancer Center and Institute of Oncology, Gliwice, 2Silesian University of Technology, Gliwice, 3Polish Academy of Science, Institute of Bioorganic Chemistry, Poznan, and 4Polish-Japanese Institute of Information Technology, Bytom, Poland

75

1

80

(Received 11 June 2010; Revised 19 December 2010; Accepted 17 January 2011) Abstract Purpose: The plasma proteome was analysed as a potential source of markers of radiosensitivity in patients treated with definitive radiotherapy for head and neck cancer. Materials and methods: Acute mucosal reactions that developed during radiotherapy were assessed in 55 patients. Blood samples were collected from each patient before the treatment and also from 50 healthy donors. The low-molecular-weight fraction of the plasma proteome (2,000–10,000 Da range) was analysed by the Matrix-Assisted Laser Desorption Ionisation mass spectrometry. The capacity for DNA break repair was assessed by the comet assay using lymphocytes irradiated in vitro. Results: Spectral components registered in plasma samples were used to build classifiers that discriminated patients from healthy individuals with about 90% specificity and sensitivity (components of 4469, 6929 and 8937 Da were the most essential for cancer classification). Four spectral components were identified (2219, 2454, 3431 and 5308 Da) whose abundances correlated with a maximal intensity of the acute reaction. Several spectral components whose abundances correlated with the rate of DNA repair in irradiated lymphocytes were also detected. Additionally, a more rapid escalation of an acute reaction was correlated with a higher level of unrepaired damage assessed by the comet assay. Conclusions: The plasma proteome could be considered as a potential source of predictive markers of acute reaction in patients with head and neck cancer treated with radiotherapy.

85

90

95

Keywords: Acute mucosal reaction, comet assay, head and neck cancer, mass spectrometry, proteomics, normal tissue radiosensitivity

100 45

50

55

Introduction Surgery and radiotherapy (RT) are the main treatment options for patients with head and neck cancer (HNC). The potential to preserve structure and function of a target organ is an obvious advantage of RT. Unconventional fractionation schemes or radiochemotherapy can improve outcomes of treatment, particularly in advanced cases of cancer (Bourhis

et al. 2004a, 2004b). However, RT-related toxicity may decrease the gain in therapeutic ratio. A severe acute mucosal reaction (AMR) significantly affects the quality of life of patients and in some cases may cause treatment discontinuation. Although the AMR is likely to appear in all patients, its escalation and duration is rather individual. Most of the patients can go through RT in an ambulatory way, while others require hospitalisation and intensive anti-inflamma-

Correspondence: Prof. Piotr Widlak, PhD, Maria Skłodowska-Curie Memorial Cancer Center and Institute of Oncology, Gliwice, Poland. Tel: þ48 32 2789672. Fax: þ48 32 2789808. E-mail [email protected] *Monika Pietrowska and Joanna Polan´ska contributed equally to this paper. ISSN 0955-3002 print/ISSN 1362-3095 online Ó 2011 Informa UK, Ltd. DOI: 10.3109/09553002.2011.556174

105

110

2 115

120

125

130

135

140

145

150

155

160

165

170

M. Pietrowska et al.

tory and supportive treatment due to mucositis (Epstein and Schubert 2003, Sonis 2004, VeraLlonch et al. 2006, Treister and Sonis 2007, Wygoda et al. 2009). A more tailored treatment could be proposed if individual radioresistance/radiosensitivity is known before the beginning of the RT. However, no reliable predictive marker of the AMR exists in clinical practice so far. Factors related to regulation of cell proliferation and DNA damage are among the most likely affecting the biological response to radiation, and have been intensively tested as potential biomarkers of individual radiosensitivity (Bentzen 2008). However, none of them has been generally accepted in clinical practice as yet. In recent years molecular diagnosis of cancer has been taking great advantage of high throughput methods of genomics and proteomics. Thus, it might be expected that genomics and proteomics would also facilitate delivery of real predictive markers of radiosensitivity (Nuyten and van de Vijver 2008, Wouters 2008). The major goal of clinical proteomics is identification and characterisation of dynamic changes related to a disease and its therapy (Hanash 2003, Wulfkuhle et al. 2003). Mass spectrometry-based analyses of the low-molecular-weight components of the blood proteome has become a valuable tool (Liotta et al. 2003, Liotta and Petricoin 2006). Several papers have been published that aimed to verify the applicability of mass spectrometry-based profiling of the serum or plasma proteome. The multiplepeptide sets selected for numerical testing have shown a potential value for diagnosis of different types of cancer. In addition, a few studies have also used mass spectrometry-based analysis of the blood proteome to address possible therapy-related changes or to identify prognostic/predictive factors (reviewed in: Conrads et al. 2004, Posadas et al. 2005, Azad et al. 2006, Solassol et al. 2006, Cho 2007a, 2007b, Cho and Cheng 2007, Palmblad et al. 2009). Mass spectrometry-based analysis of the serum proteome also has been performed on blood samples collected before and during RT of patients with different type of cancer. These samples were compared and several potential markers of exposure to ionising radiation were identified after a pair-wise analysis (Menard et al. 2006). In this study, the association between plasma proteome profiles analysed by mass spectrometry (MS) and the severity of radiotherapy-related acute mucosal reaction was investigated in individual head and neck cancer patients. Assuming that combination of different assays might increase the chance of defining a potential marker(s) of radiosensitivity, analysis of the parameters of a lymphocyte-based DNA repair comet assay was also carried out.

Materials and methods Characteristics of patient and control groups Fifty-five patients (46 men, nine women, mean age 57 years with a range 40–75 years) with laryngeal (n ¼ 24), pharyngeal (n ¼ 15) and oral cavity (n ¼ 16) squamous cell cancer were enrolled into this study. Primary tumour was scored as: T1 (15%), T2 (34%), T3 (24%) and T4 (27%); nodal status was scored as: N0 (49%), N1 (20%) and N3 (31%); none of patients had diagnosed metastases (all M0). All patients were treated with definitive radiotherapy, either alone (n ¼ 45) or combined with systemic treatment based on cisplatin and 5fluorouracil (n ¼ 10). Total radiation doses were in the range of 52–76 Gy (median 68 Gy), given in 5–7 weeks. The AMR was assessed using the Dische scoring system (Dische 1994) each 2–4 days during the radiotherapy. A modified Dische system places emphasis on both morphologic and functional symptoms, especially dysphagia and pain, and has been used in our Centre for more then 15 years (Wygoda et al. 2009). The AMR was characterised by two parameters: A maximal intensity of the reaction and a rate of escalation of the reaction. The latter parameter was described as a period between time points when the observed reaction started to grow and reached the maximal intensity (lower number represented faster escalation). Blood samples were collected before the start of the treatment. Fifty volunteers (45 men) were included in the study as a control group; mean age 54 years (range 32–77 years). All of them were free of any known acute or chronic illness and were not treated with any anticancer therapy in the past. The study was approved by the local Ethics Committee at the Maria Sklodowska-Curie Memorial Cancer Center and Institute of Oncology, Poland, and all participants provided informed consent indicating their conscious and voluntary participation. All participants were Caucasians. There was a similar proportion of smokers (about 73%) and alcohol consumers (about 71%) in both analysed groups.

175

180

185

190

195

200

205

210

215

Mass spectrometry analysis of plasma samples Blood samples (5 ml collected into BD Vacutainer Tubes containing sodium heparin (Becton Dickinson, Franklin Lake, NJ, USA) were centrifuged on a Ficoll gradient (LymphoprepTM; MP Biochemicals, Solon, OH, USA) to obtain the plasma and lymphocyte fractions; purified plasma was aliquoted and stored at 7708C. Plasma samples were analysed using an Autoflex MALDI-ToF (Matrix-Assisted Laser Desorption Ionization with Time of Flight analyzer) mass spectrometer (Bruker Daltonics,

220

225

Plasma proteome and radiation-induced acute reactions Bremen, Germany); the analyzer worked in the linear mode and positive ions were recorded in the mass range between 2,000 and 10,000 Da. Mass calibration was performed after every four samples using appropriate standards (Bruker Daltonics). Each sample was passed repeatedly 10 times through C18 ZipTip-microcolumns (Millipore, Billerica, MA, USA), columns were washed with water and then eluted with 1 ml of matrix solution (30 mg/ml sinapinic acid in 50% acetonitrile and 0.1% trifluoroacetic acid with addition of 1 mM n-octyl glucopyranoside; all general reagents from Merck, Darmstadt, Germany) directly onto the 600 mm AnchorChip plates (Bruker Daltonics). ZipTip extraction/loading was repeated twice for each sample and for each spot on the plate two spectra were acquired after 120 laser shots (i.e., four spectra were recorded for each sample).

230

235

240

245

The comet assay 250

255

260

265

270

275

1 280

285

Freshly isolated lymphocytes were irradiated on ice using a linear accelerator (6 MeV energy Clinac 600; Varian, Palo Alto, CA, USA) at a 2 Gy dose (a 1 Gy/ min rate). Aliquots of cells were collected immediately before irradiation (not irradiated control), and immediately after irradiation (time 0) or after different time of incubation at 378C (15, 30, 60, 120 and 180 min). Collected cells were placed on ice and all further steps were performed at 0–48C. The amount of DNA breaks was assessed by the alkaline single-cell gel electrophoresis method (the comet assay) (Green et al. 1992). About 1,000 cells suspended in 1% low melting agarose were placed on a microscope slide, the gel was allowed to set, and then the slides were incubated for 60 min in the lysis solution comprising of 2.5 M NaCl, 100 mM ethylenediaminetetraacetic acid (EDTA), 10 mM Tris-HCl pH 7.5, 1% Triton X-100), which was followed by incubation for 20 min in the denaturation solution (300 mM NaOH, 1 mM EDTA, pH 13). Electrophoresis was performed in the denaturation solution for 20 min at 1 V/cm, and then gels were neutralised for 5 min in 0.4 M Tris-HCl pH 7.5, and stained with ethidium bromide. Cells were observed at 400 6 magnification using a fluorescence microscope (Axiophot, Carl Zeiss, Jena, Germany) and classified into five categories (A0– A4) according to the classification of Collins et al. (1993). The experimental scores were fitted to the exponential model constructed by us previously (Palyvoda et al. 2003), which describes the kinetics of DNA break repair. The model is characterised by three parameters that correspond to the initial level of DNA breaks, the time constant inversely related to the rate of repair and the level of residual unrepaired damage.

3

Data processing and statistical analysis The preprocessing of spectral data that included averaging of technical repeats, binning of neighboring points to reduce data complexity, removal of the spectral area below baseline and normalisation of the total ion current (TIC), was performed according to procedures considered to be standard in the field (Hilario et al. 2006, Karpievitch et al. 2007). In the second step the spectral components, which reflected [M þ H]þ peptide ions recorded at defined mass to charge (m/z) values, were identified using decomposition of mass spectra into their Gaussian components as described elsewhere (Pietrowska et al. 2009). The average spectrum was decomposed into a sum of 300 Gaussian bellshaped curves by using a variant of the expectation maximisation (EM) algorithm (Hastie et al. 2001). The Gaussian components were used to compute features of registered spectra (termed spectral components hereafter) for all samples by the operation of convolutions of spectral signals with Gaussian masks. These spectral components were characterised by their abundances (or intensities), location along the m/z axis and standard deviation of corresponding Gaussian. The knowledge base EPO-KB (Empirical Proteomic Ontology Knowledge Base) (Lustgarten et al. 2008), which links m/z values to known peptide/proteins, was used for hypothetical identification of registered spectral components assuming their mono-protonation and allowing for a 0.5% mass accuracy limit. The univariate analysis of differentiating components was performed using the Mann-Whitney U test and the Bonferroni correction for multiple testing (the corrected p-value 5 0.05 was considered as statistically significant). The classification was based on the spectral components described above and used a version of the Support Vector Machine (SVM) algorithm developed by Scho¨lkopf et al. (2000) with a feature selection based on the entropy measure. The performance of classifiers built of different numbers of spectral components was estimated by the level of errors and receiver operating curves (ROC) as described previously (Pietrowska et al. 2009). Correlation between abundances of spectral components and intensity of the AMR or capacity of DNA repair was analysed using both the Person’s correlation test and the Spearman’s rank correlation test. The significance of a correlation was characterised by the Storey test q-value, which adjusted statistical significance for multiple testing using the False Discovery Rate (FDR) approach (a 0.05 value was chosen as the significance cut-off level, which corresponded to 5% FDR).

290

295

2

300

305

310

315

320

325

330

2 335

340

4

M. Pietrowska et al.

Results 345

350

355

In the first step of analysis, spectral components, i.e., registered [M þ H]þ ions corresponding to blood peptides, that were specific for plasma of patients were identified. Figure 1A shows average mass spectrum, in the range of 2,000–10,000 Da, computed for all analysed plasma samples. Mass spectra registered for healthy donors and patients were compared after their decomposition into Gaussian components, and then components showing significantly different abundances were identified. About 15% of components showed significantly different intensities between the two types of samples. However, to obtain more reliable classification of samples,

360

365

370

375

380

385

390

395

Figure 1. Classification of head and neck cancer patients using plasma proteome mass profiles. (A) The average MALDI-ToF mass spectrum of plasma peptides. (B) Estimation of the performance of cancer classification; the error rate was plotted against the number of spectral components in the classifier (dotted lines show 95% CI). (C) Characteristics of three spectral components that were present in the majority of cancer classifiers; shown are levels of each component in samples from healthy controls and HNC patients (box-plots show minimum, lower quartile, median, upper quartile and maximum values; outliers are marked by asterisks).

multi-component classifiers based on proteome profiles characteristic for both groups were built. The best performance of classification (about 10% of total errors) was obtained with classifiers built of 6– 10 components (Figure 1B). For further analyses, classifiers built statistically of eight components were selected, which allowed the classification of head and neck cancer patients with about 90% sensitivity and specificity. Table I shows characteristic of eight spectral components that were the most frequent in such cancer classifiers. All of them were present in at least 45% of classifiers, and also had high potency to differentiate control and cancer samples in univariate analysis (statistical significance of the differences was at the level of a corrected p-value better than 1076); combined marker built intentionally from these 8 components revealed 92% sensitivity and 90% specificity (Table I). Three spectral components (i.e., plasma peptides) that had nominal m/z values of 4469, 6929 and 8937 Da were present in at least 90% of classifiers and had very high potential to differentiate control and cancer samples also when analysed separately (a combined marker built of these three components revealed 86% sensitivity and 88% specificity). Figure 1C shows abundances of these spectral components in plasma samples of healthy controls and cancer patients. We conclude that several peptides have different abundances in plasma of healthy individuals and patients, and plasma proteome mass profiles specific for cancer patients could be clearly established. Spectral components specific for plasma samples collected from patients who responded differently to RT were identified in the second step of analysis. Individual intensities of AMR were assessed based on the modified Dische score (Dische 1994, Wygoda et al. 2009). The maximal intensities of the AMR were noted in the range between 8 and 15 points (median ¼ 11 points), while the rates of escalation of the reaction were noted in the range between 1 and 33 days (median ¼ 10 days). The patient group was split into the sub-groups: ‘lower’ and ‘higher’ maximal intensity of the reaction, as well as into ‘slower’ and ‘faster’ rate of the reaction escalation, according to their median values. When patients with lower and higher maximal intensity of the reaction were compared, several spectral components showed different abundances in corresponding plasma samples. However, the difference was statistically significant for only one plasma component with a registered m/z value 5308 Da (p ¼ 0.00015). When the two subgroups of patients differing in the rate of escalation of AMR were compared, none of the differences in the abundance of the plasma components were statistically significant. The correlation between the AMR noted for each patient and the abundance of each spectral compo-

400

405

410

415

420

425

430

435

440

445

450

455

5

Plasma proteome and radiation-induced acute reactions Table I. Characteristics of spectral components that differentiated plasma samples from cancer patients and healthy controls.

Component m/z [Da]

Frequency [%]

p-value

Corrected p-value

99% 93% 90% 63% 59% 56% 48% 46% 9.4

3.46E-15 1.32E-12 2.38E-12 2.42E-10 1.16E-09 1.32E-09 1.26E-09 9.38E-10 91.7

1.04E-12 3.96E-10 7.15E-10 7.26E-08 3.49E-07 3.96E-07 3.79E-07 2.81E-07 90.1

Total error [%]

Sensitivity [%]

Specificity [%]

13.3 19.7 23.6 24.6 25.9 24.8 28.9 30.6

83.4 77.5 72.2 65.6 71.1 74.1 65.5 66.3

91.6 84.9 83.2 88.8 79.1 77.8 79.7 75.1

515

460

465

470

4469 8937 6929 8895 8132 6772 8876 8149 combined marker built of 8 components:

Shown are the most frequent spectral components (m/z values) and their relative frequencies in a statistical cancer classifiers built of 8 features. The p-values are for differences between patients and healthy controls measured by the Mann-Whitney U test for each individual component (also shown after the Bonferroni correction against multiple testing). Total error rate, sensitivity and specificity were calculated for one-feature classifiers (markers) built for each component, as well as for combined marker built intentionally of the 8 top components (last line). Hypothetical identity of peptides based on annotation of registered m/z values at the EPO-KB knowledge base: 4469 Da – Serpin C1 (426–464 fragments), 8149 Da – component C3 (672–739 fragments), 8895 Da – vitronectin (402–478 fragments), 8937 Da – component C3 (672–747 fragments) or apolipoprotein A2 (24–100 fragments).

520

525

530

475

480

Table II. Features of plasma proteome components specific for patients showing different intensities of maximal acute reaction during radiotherapy.

Component m/z[Da]

485

490

495

500

505

510

3431 2454 2219 5308

Spearman’s rank correlation coefficient

Storey-test q-value (correlation)

U-test p-value (two group comp.)

70.509 70.492 70.481 70.295

0.027 0.027 0.029 0.181

0.061 0.068 0.099 0.00015*

Listed are three spectral components showing statistically significant correlation between their abundances and maximal intensities of the AMR, and one component that significantly differentiated samples from two subgroups (‘lower’ and ‘higher’ maximal intensity of the AMR; n ¼ 47). Each component is characterised by its m/z value, correlation coefficient, q-value of the correlation significance, and p-value of the significance in the univariate analysis of the difference between the two subgroups (asterisk denotes significance retained after the Bonferroni correction).

nent registered in a corresponding plasma sample was investigated in the next step of analysis. We found that abundances of three spectral components (nominal m/z values 2219, 2454 and 3431 Da) correlated significantly with the maximal intensity of AMR (q-value 50.05 for the Spearman’s rank correlation test), while abundance of none of spectral components correlated with the rate of reaction escalation (Table II and Figure 2 show characteristics of differentiating spectral components). We conclude that abundances of a few peptides detected in plasma samples collected before the start of radiotherapy correlate with the level of maximal intensity of the AMR. In the analysed material we found no statistically significant differences in the maximal intensities of AMR or features of plasma

535

540

545 Figure 2. The example of spectral component with nominal m/z value 3431 Da, whose abundance correlated with the intensity of maximal acute mucosal reaction to irradiation. (A) Actual spectral plots; grey and black lines represent patients with lower and higher intensity of the reaction, respectively (white line represents the model Gaussian). (B) Abundance of the component in samples from patients with lower and higher intensity of the AMR (boxplots show minimum, lower quartile, median, upper quartile and maximum values; outliers are marked by asterisks). (C) Correlation between abundance of the component and intensity of the AMR; each dot represents one patient (Pearson’s correlation coefficient r ¼ 70.214, p ¼ 0.052; n ¼ 47).

proteome profiles between subgroups of patients with different location and size of a tumour or scheme of a treatment (possibly because of small size of subsets), hence multiple subgroup analyses of correlation between proteome features and the AMR were not performed. Spectral components specific for plasma samples collected from cancer patients whose lymphocytes had different capacities to repair radiation-induced DNA damage were identified in the next step. The in vitro repair rates were assessed using the comet assay and then three features were noted: The initial level of DNA breaks in irradiated cells, the rate of repair, and the level of unrepaired residual damage (Palyvoda

550

555

560

565

570

6

575

580

585

590

595

600

605

610

615

620

625

M. Pietrowska et al.

et al. 2003). The analyses were performed in two ways described above: Spectral components that differentiated plasma samples from patients with lower/higher value of each parameter (two subgroups defined according to the median value) were identified, or correlations between abundances of each spectral component and each parameter of the comet assay were established. When two groups with different initial levels of DNA breaks were compared, several spectral components showed different abundances in the corresponding plasma samples, but none of these differences remained statistically significant after the Bonferroni correction. However, for 34 spectral components highly significant correlations between their abundances and individual levels of the initial DNA breaks were found (q 50.005 for the Spearman’s rank correlation test); Table IIIa shows characteristics of spectral components with the highest significance of correlation. When two groups with different rates of DNA repair were compared several spectral components showed different abundances in the corresponding plasma samples, and for two plasma components (m/z ¼ 2737 and 9736 Da) the differences remained significant after the Bonferroni correction. In addition, for 10 spectral components a highly significant correlation between their abundances and individual levels of repair rate was found (q 50.01 for the Spearman’s rank correlation test); Table IIIb shows characteristics of spectral components with the highest significance of correlation. In marked contrast, for none of spectral components was found any significant correlation between its abun-

Table III. Features of plasma proteome profiles specific for patients whose lymphocytes showed different capacities for in vitro repair of DNA breaks.

Component m/z [Da]

Spearman’s rank correlation coefficient

Storey-test q-value (correlation)

dance in plasma samples and the level of residual DNA damage analysed by the comet assay. We conclude that abundances of several plasma peptides correlate with either the level of DNA breaks induced in lymphocytes irradiated in vitro or the rate of repair of such damage. The correlation between different responses of patients to radiation in vivo and capacities to repair DNA damage in lymphocytes irradiated in vitro was finally investigated. Parameters that described individual severity of the AMR and parameters of the comet assay that characterised DNA repair in lymphocytes isolated before the start of RT were analysed. We found that the rates of escalation of the AMR correlated with the levels of residual unrepaired DNA breaks in lymphocytes. Importantly, a faster rate of the AMR in vivo was related to a higher level of unrepaired damage in vitro (p ¼ 0.031); however, the statistical significance of this correlation was observed only when the rank correlation test was applied (Figure 3). Discussion The low molecular weight (515 kDa) component of the blood proteome analysed by mass spectrometric methods appears to be a promising source of markers for cancer diagnostics (reviewed in: Azad et al. 2006, Solassol et al. 2006, Cho 2007a, 2007b, Cho and Cheng 2007, Palmblad et al. 2009). The proteomics approach that takes into consideration characteristic features of the whole proteome but does not rely on a particular protein is called proteome pattern analysis or proteome profiling. In this approach multi-component sets of peptides or proteins (which are exemplified by ions registered at defined m/z values in the mass

635

640

645

650

655

660

665

U-test p-value (two group comp.)

(a) Correlation with the initial radiation-induced DNA breaks 9033 0.583 0.0007 0.0006 2668 70.622 0.0007 0.0037 5697 0.596 0.0007 0.009 4716 0.593 0.0007 0.012 4581 0.532 0.003 0.032 (b) Correlation with the rate of DNA breaks repair 5533 0.640 0.0005 0.0006 2737 70.483 0.009 0.00004* 9304 0.484 0.009 0.0003 5697 0.507 0.009 0.002 5808 0.479 0.009 0.025 Listed are five spectral components with the highest significance of correlation between their abundances and parameters of the comet assay (n ¼ 48). Each component is characterised by its m/z value, correlation coefficient, q-value of the correlation significance, and also p-value of the significance in the univariate analysis of the difference between the two subgroups (asterisks denote differences that retained significance after the Bonferroni correction).

630

670

675

680 Figure 3. The correlation between the level of residual unrepaired DNA breaks assessed by the comet assay and the rate of escalation of acute mucosal reaction; Spearman’s rank correlation coefficient rho ¼ 70.306, p ¼ 0.031 (dotted lines show 95% confidence interval; n ¼ 50).

Plasma proteome and radiation-induced acute reactions 685

690

695

3 700

705

710

715

720

725

730

735

740

spectrum) define specific proteomic patterns (or profiles) that can be used for sample identification and classification, even though their particular components may lack differentiating power when analysed separately (Li et al. 2004, Dworzanski and Snyder 2005, Somorjai 2008). A few studies have addressed the possibility of applying mass spectrometry-based blood proteome pattern analysis in the diagnostics of the head and neck cancer. These works allowed identification of potential markers, i.e., plasma or serum peptides, specific for patients with head and neck squamous cell cancer at different locations and clinical stages (Soltys et al. 2004, Wadsworth et al. 2004, Cheng et al. 2005, Schaaij-Visser et al. 2010). Similarly, blood proteome profiling revealed several serum or plasma peptides whose abundances changed after definitive treatment of patients with HNC (Cho et al. 2004, Gourin et al. 2007, Freed et al. 2008, Gourin et al. 2009). Here we used MALDI-based analysis of plasma proteome to identify peptides whose abundances were significantly different in the plasma of healthy individuals and patients with HNC. Those peptides were eventually used to build a multicomponent cancer classifier with about 90% sensitivity and specificity. Based on the m/z value of registered spectral components one could try to annotate corresponding peptides using existing experimental data published for MALDI or SELDI (SurfaceEnhanced Laser Desorption Ionization) MS analyses. Using the knowledge base EPO-KB (Empirical Proteomic Ontology Knowledge Base) (Lustgarten et al. 2008), the hypothetical identities were found for four out of eight of the most frequent components of the classifier. These were: 4469 kDa (fragment of serpin C1), 8149 Da (fragment of component C3), 8895 Da (fragment of vitronectin) and 8937 Da (fragments of component C3 or apolipoprotein A2). Noteworthy, three spectral components that were the most important for proposed cancer classifier (namely m/z ¼ 4469, 6929 and 8937 Da) apparently corresponded to spectral peaks previously reported among MS peaks differentiating serum or plasma samples from HNC patients and healthy controls (Soltys et al. 2004, Wadsworth et al. 2004). As far as we know this was the first attempt to search for the low-molecular-weight plasma peptides specific for patients who showed a different severity of the AMR due to RT. One spectral component (m/ z ¼ 5308 Da), whose abundance was significantly higher in blood of patients with a low maximal intensity of AMR was detected. In addition, three spectral components (m/z ¼ 2219, 2454 and 3431 Da), whose abundances in plasma samples correlated with individual intensities of their AMR were found. One of the spectral components, with a nominal m/z value 3431 Da, whose abundance decreased in the plasma of patients with higher

7

intensities of AMR, was hypothetically identified (using the EPO-KB tool) as a fragment of defensin 1 (DEFA1, fragment 65–94) or/and secretory granule neuroendocrine protein 1 (SCG5, fragment 183– 212). Defensin 1 is a microbicidal peptide made by neutrophils, which participates in the innate immune response (Lehrer et al. 1991), while SCG5 is a secretory chaperone protein expressed in different endocrine glands (Mbikay et al. 2001). Neither of these proteins has been reported to be involved in responsiveness to ionising radiation so far. However, antimicrobial functions of defensin 1 could in part explain its relation to radiation-induced reaction because bacterial infections are important symptoms of the AMR (Sonis 2004). In addition, several plasma peptides were identified whose abundances correlated markedly with the capacity of isolated lymphocytes to repair DNA breaks induced by radiation in vitro. Abundance of some of them significantly correlated with the rate of DNA break removal, while others were correlated with the level of initial damage. One component (m/z ¼ 5697 Da) was associated with both features of the comet assay. Noteworthy, the level of unrepaired residual damage did not correlate with the abundance of any detected plasma peptides. The applicability of the lymphocyte-based comet assay for prediction of radiation-induced acute reaction has been tested previously. A few studies analysed possible associations between DNA repair capacity in vitro and acute skin reaction in breast cancer patients treated with radiotherapy. However, the conclusions were contradictory probably because of different clinical and biological endpoints (Oppitz et al. 2002, Popanda et al. 2003). The possible association between DNA repair capacity in cryopreserved lymphocytes and acute skin reaction was also analysed in patients with nasopharyngeal cancer. This study showed that, although individuals with defects of DNA repair could be identified with the comet assay, the correspondence between repair deficiency and acute reaction to RT was only marginal (Wang et al. 2005). In this study we found that the escalation of the AMR in the oral cavity of patients was correlated with capacity for DNA break repair in the ‘model’ cells in vitro. It may have functional meaning because higher levels of unrepaired residual damage were associated with a faster escalation of the AMR.

745

750

755

760

765

770

775

780

785

790

Conclusions Several plasma components were detected by MALDI-ToF mass spectrometry whose abundances correlated with the response to ionising radiation of either oral cavity mucosa or isolated lymphocytes. The observed associations, possibly confounded by

795

8

M. Pietrowska et al.

the heterogeneity of the group tested, were not strong enough to suggest any clinical applicability at this stage of the research. Nevertheless, presented data indicated that the plasma proteome could be considered as a potential source of predictive biomarkers of normal tissue radiosensitivity in patients with HNC treated with RT.

800

805

Acknowledgements This work was supported by the Polish Ministry of Science and Higher Education, Grant No. 402– 450339.

810

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

815

References 820

825

830

835

3

840

845

850

855

1

Azad NS, Rasool N, Annunziata CM, Minasian L, Whiteley G, Kohn EC. 2006. Proteomics in clinical trials and practice. Molecular and Cellular Proteomics 5:1819–1829. Bentzen SM. 2008. From cellular to high-throughput predictive assays in radiation oncology: Challenges and opportunities. Seminars in Radiation Oncology 18:75–88. Bourhis J, Etessami A, Wilbault P, Lusinchi A, Calais G, Lapeyre M, Pignon JP. 2004a. Altered fractionated radiotherapy in the management of head and neck carcinomas: Advantages and limitations. Current Opinion in Oncology 16:215–219. Bourhis J, Calais G, Lapeyre M, Tortochaux, Alfonsi M, Sire C, Bardet E, Rives M, Bergerot P, Rhein B, Desprez B. 2004b. Concomitant radiochemotherapy or accelerated radiotherapy: Analysis of two randomized trials of the French Head and Neck Cancer Group (GORTEC). Seminars in Oncology 31:822–826. Cheng AJ, Chen LC, Chien KY, Chen YJ, Chang JTC, Wang HM, Liao CT, Chen IH. 2004. Oral cancer plasma tumor marker identified with bead-based affinity-fractionated proteomic technology. Clinical Chemistry 51:2236–2244. Cho WCS, Yip TTC, Yip C, Yip V, Thulasiraman V, Ngan RKC, Yip TT, Lau WH, Au JSK, Law SCK, Cheng WW, Ma VWS, Lim CKP. 2004. Identification of serum amyloid A protein as a potentially useful biomarker to monitor relapse of nasopharyngeal cancer by serum proteomic profiling. Clinical Cancer Research 10:43–52. Cho WC. 2007a. Contribution of oncoproteomics to cancer biomarker discovery. Molecular Cancer 6:e25. Cho WC. 2007b. Proteomics technologies and challenges. Genomics Proteomics Bioinformatics 5:77–85. Cho WC, Cheng CH. 2007. Oncoproteomics: Current trends and future perspectives. Expert Review of Proteomics 4:401–410. Collins AR, Dusinska M, Franclin M, Somorovska M, Petrovska H, Duthie S, Fillion L, Panayiotidis M, Raslova K, Vaughan N. 1997. Comet assay in human biomonitoring studies: Reliability, validation and applications. Environmental and Molecular Mutagenesis 30:139–146. Conrads TP, Hood BL, Issaq HJ, Veenstra TD. 2004. Proteomic patterns as a diagnostic tool for early-stage cancer: A review of its progress to a clinically relevant tool. Molecular Diagnostics 8:77–85. Dische S. 1994. The uniform reporting of treatment-related morbidity. Seminars in Radiation Oncology 4:112–118.

Dworzanski JP, Snyder AP. 2005. Classification and identification of bacteria using mass spectrometry-based proteomics. Expert Review of Proteomics 2:863–878. Epstein JB, Schubert MM. 2003. Oropharyngeal mucositis in cancer therapy. Oncology 17:1767–1779. Freed GL, Cazares LH, Fichandler CE, Fuller TW, Sawyer CA, Stack BC Jr, Schraff S, Semmes OJ, Wadsworth JT, Drake RR. 2008. Differential capture of serum proteins for expression profiling and biomarker discovery in pre- and posttreatment head and neck cancer samples. Laryngoscope 118:61–68. Gourin CG, Moretz WH, Weinberger PM, Xia ZS, Liu Z, Terris DJ, Adam BL. 2007. Serum protein profile analysis following definitive treatment in patients with head and neck squamous cell carcinoma. Archives of Otolaryngology Head and Neck Surgery 133:1125–1130. Gourin CG, Zhi W, Adam BL. 2009. Proteomic identification of serum biomarkers for head and neck cancer surveillance. Laryngoscope 119:1291–302. Green M, Low J, Harcourt SA, Akinluyi P, Row T, Cole J, Anstey AV, Arlett CV. 1992. UV-C sensitivity of unstimulated and stimulated human lymphocytes from normal and Xeroderma pigmentosum donors in the comet assay: A potential diagnostic technique. Mutation Research 273:137–144. Hanash S. 2003. Disease proteomics. Nature 422:226–232. Hastie T, Tibshirani R, Friedman JH. 2001. The elements of statistical learning. New York: Springer Verlag. Hilario M, Kalousis A, Pellegrini C, Mu¨ller M. 2006. Processing and classification of protein mass spectra. Mass Spectrometry Review 25:409–449. Karpievitch YV, Hill EG, Smolka AJ, Morris JS, Coombes KR, Baggerly KA, Almeida JS. 2007. PrepMS: TOF MS data graphical preprocessing Tool. Bioinformatics 23:264–265. Lehrer RI, Ganz T, Selsted ME. 1991. Defensins: Endogenous antibiotic peptides of animal cells. Cell 64:229–230. Li L, Tang H, Wu Z, Gong J, Gruidl M, Zou J, Tockman M, Clark RA. 2004. Data mining techniques for cancer detection using serum proteomic profiling. Artificial Intelligence in Medicine 32:71–83. Liotta LA, Ferrari M, Petricoin EF. 2003. Clinical proteomics: Written in blood. Nature 425:905. Liotta LA, Petricoin EF. 2006. Serum peptidome for cancer detection: Spinning biological trash into diagnostic gold. Journal of Clinical Investigations 116:26–30. Lustgarten JL, Kimmel C, Ryberg H, Hogan W. 2008. EPO-KB: A searchable knowledge base of biomarker to protein links. Bioinformatics 24:1418–1419. Mbikay M, Seidah NG, Chretien M. 2001. Neuroendocrine secretory protein 7B2: Structure, expression and functions. Biochemical Journal 357:329–342. Menard C, Johann D, Lowenthal M, Muanza T, Sproull M, Ross S, Gulley J, Petricoin E, Coleman CN, Whiteley G, Liotta L, Camphausen K. 2006. Discovering clinical biomarkers of ionizing radiation exposure with serum proteomic analysis. Cancer Research 66:1844–1850. Nuyten DS, van de Vijver MJ. 2008. Using microarray analysis as a prognostic and predictive tool in oncology: Focus on breast cancer and normal tissue toxicity. Seminars in Radiation Oncology 18:105–114. Oppitz U, Schulte S, Stopper H, Baier K, Muller M, Wulf J, Schakowski R, Flentje M. 2002. In vitro radiosensitivity measured in lymphocytes and fibroblasts by colony formation and comet assay: Comparison with clinical acute reactions to radiotherapy in breast cancer patients. International Journal of Radiation Biology 78:611–616. Palmblad M, Tiss A, Cramer R. 2009. Mass spectrometry in clinical proteomics – from the present to the future. Proteomics Clinical Applications 3:6–17.

860

865

870

875

880

885

890

895

900

905

910

Plasma proteome and radiation-induced acute reactions

915

920

925

930

935

Palyvoda O, Polanska J, Wygoda A, Rzeszowska-Wolny J. 2003. DNA damage and repair in lymphocytes of normal individuals and cancer patients: Studies by the comet assay and micronucleus tests. Acta Biochimica Polononica 50:181–190. Popanda O, Ebbeler R, Twardella D, Helmbold I, Gotzes F, Schmezer P, Thielmann HW, von Fournier D, Haase W, Sautter-Bihl ML, Wenz F, Bartsch B, Chang-Claude J. 2003. Radiation-induced DNA damage and repair in lymphocytes from breast cancer patients and their correlation with acute skin reactions to radiotherapy. International Journal of Radiation Oncology Biology and Physics 55:1216–1225. Posadas EM, Simpkins F, Liotta LA, MacDonald C, Kohn EC. 2005. Proteomic analysis for the early detection and rational treatment of cancer-realistic hope? Annals of Oncology 16:16– 22. Schaaij-Visser TB, Brakenhoff RH, Leemans CR, Heck AJ, Slijper M. 2010. Protein biomarker discovery for head and neck cancer. Journal of Proteomics 73:1790–803. Scho¨lkopf B, Smola AJ, Williamson RC, Bartlett PL. 2000. New Support Vector algorithms. Neural Computation 12:1207– 1245. Solassol J, Jacot W, Lhermitte L, Boulle N, Maudelonde T, Mange´ A. 2006. Clinical proteomics and mass spectrometry profiling for cancer detection. Expert Review of Proteomics 3:311–320. Soltys SG, Le QT, Shi G, Tibshirani R, Giaccia AJ, Koong AC. 2004. The use of plasma surface-enhanced laser desorption/ ionization time-of-flight mass spectrometry proteomic patterns for detection of head and neck squamous cell cancers. Clinical Cancer Research 10:4806–4812.

9

Somorjai RL. 2008. Pattern recognition approaches for classifying proteomic mass spectra of biofluids. Methods in Molecular Biology 428:383–396. Sonis ST. 2004. The pathobiology of mucositis. Nature Review Cancer 4:277–284. Treister N, Sonis S. 2007. Mucositis: Biology and management. Current Opinion in Otolaryngology Head and Neck Surgery 15:123–129. Vera-Llonch M, Oster G, Hagiwara M, Sonis S. 2006. Oral mucositis in patients undergoing radiation treatment for head and neck carcinoma. Cancer 106:329–336. Wadsworth JT, Komers KD, Cazares LH, Malik G, Adam BL, Stack BC, Wright GL, Semmes OJ. 2004. Serum protein profiles to identify head and neck cancer. Clinical Cancer Research 10:1625–1632. Wang WD, Chen ZT, Li DZ, Cao ZH, Sun SL, Pu P, Chen XP. 2005. Correlation between DNA repair capacity in lymphocytes and acute side effects to skin during radiotherapy in nasopharyngeal cancer patients. Clinical Cancer Research 11:5140–5145. Wouters BG. 2008. Proteomics: Methodologies and applications in oncology. Seminars in Radiation Oncology 18:115–125. Wulfkuhle JD, Liotta LA, Petricoin EF. 2003. Proteomic applications for the early detection of cancer. Nature Review Cancer 3:267–275. Wygoda A, Maciejewski B, Składowski K, Hutnik M, Pilecki B, Golen´ M, Rutkowski T. 2009. Pattern analysis of acute mucosal reactions in patients with head and neck cancer treated with conventional and accelerated irradiation. International Journal of Radiation Oncology Biology and Physics 73:384–390.

970

975

980

985

990

995 940

1000 945

1005 950

1010 955

1015 960

1020 965

1025