Supplementary Material for

1H NMR based metabolomic approach to monitoring of the head and neck cancer treatment toxicity.

Ł. Boguszewicz 1, A. Hajduk 2, J. Mrochem-Kwarciak 3, A. Skorupa 1, M. Ciszek 1, A. Heyda 2, K. Skladowski 2 M. Sokol 1,

1 Department of Medical Physics, Maria Sklodowska-Curie Memorial Cancer Center and Institute of Oncology, Gliwice Branch, Poland.

2 I Radiotherapy Clinic, Maria Sklodowska-Curie Memorial Cancer Center and Institute of Oncology, Gliwice Branch, Poland.

3 Analytics and Clinical Biochemistry Department, Maria Sklodowska-Curie Memorial Cancer Center and Institute of Oncology, Gliwice Branch, Poland.

Maria Sklodowska-Curie Memorial Cancer Center and Institute of Oncology, Gliwice Branch

Street: Wybrzeze Armii Krajowej 15,

44-101 Gliwice, Poland.

Corresponding author:

Lukasz Boguszewicz

Phone: +48322788047

Fax: +48322313512

Acknowledgments

The work has been funded by National Science Centre grant 2015/17/B/NZ5/01387
Patients

45 patients with head and neck squamous cell carcinomas (HNSSC) were involved in the study (33 men and 12 women, all Caucasians, at median age of 58). The primary sites of disease and primary tumor stages as listed in Table S1. 62% of patients has locoregional disease.

Table S1. List of anatomical regions affected by tumor.

ICD-10 / Anatomical region / No. of patients / T1 / T2 / T3 / T4 / N+
C32 / Larynx / 13 / 3 / 7 / 3 / 3
C13 / Hypopharynx / 11 / 1 / 5 / 5 / 10
C11 / Nasopharynx / 2 / 1 / 1 / 2
C10 / Oropharynx / 10 / 3 / 6 / 1 / 8
C09 / Tonsil / 5 / 1 / 4 / 3
C06 / Retromolar triangle / 1 / 1 / 1
C05 / Palate / 1 / 1 / 1
C02 / Tongue / 1 / 1
C01 / Base of tongue / 1 / 1


Design of the study

The subsequent steps of identification of metabolic signatures of ARS are described below and visualized in Figure S1. The main parts of the procedure are as follows:

1.  In the first step the four sets of 45 1H NMR bucketed spectra from NOESY, CPMG, DIFF and J-resolved sequences were analyzed with PCA in order to detect outliers and to find the directions of the highest variation in the data.

2.  Supervised multivariate analyses were performed only on the data for which the directions of the greatest variation were correlated with the severity of ARS (i.e. there was a clear trend in the spectra clustering in the PCA scores plot).

3.  Identification and quantification of metabolites was performed by signal integration.

4.  Quantified metabolites were analyzed using classical statistical tools.

The separation of the high and low ARS groups was unambiguously seen only in the scores plot from PCA analysis of the J-resolved spectra. Thus, only J-resolved spectra were subjected to supervised OPLS-DA analysis and quantified.

The remaining data sets (NOESY, CPMG, DIFF) were also analyzed and the results of these studies will be published in the future. However the general conclusions were applied in the current paper.

Figure S1. Design of the study.


NMR pulse sequences parameters

Table S2. Pulse sequence parameters.

Pulse program / NOESYGPPR1D / CPMGPR1D / LEDBPGPPR2S1D / JRESGPPRQF
TD / 65536 / 65536 / 65536 / 8192
SW [ppm] / 30 / 20 / 30 / 16.62
AQ [sec] / 2.73 / 4.09 / 2.73 / 0.62
D1 [sec] / 4 / 4 / 4 / 2
D8 [sec] / 0.01 / - / - / -
D16 [sec] / - / - / 0.0002 / 0.0002
D20 [sec] / - / 0.0003 / 0.12 / -
D21 [sec] / - / - / 0.005 / -
DS / 4 / 4 / 4 / 16
L4 / - / 126 / - / -
NS / 32 / 64 / 64 / 1
DELTA1 [sec] / - / - / 0.11572488 / -
DELTA2 [sec] / - / - / 0.004172 / -