(8 / 8) Supplementary Materials
Supplementary Materials
Strategy of identifying biomarkers
The potential biomarkers were ascertained when the values of VIP were more than 1 and the two-sided p-values calculated from nonparametric Kruskal-Wallis rank sum test corrected by Bonferroni correction were less than 0.01, with the retention time (RT) and mass-to-charge ratio (m/z) data as identifiers for each ion.
First, extracted ion chromatograms were obtained from the base peak intensity (BPI) chromatogram of plasma for ions with m/z. Mass spectra from the extracted ion chromatographic peaks at the defined RT and annotation results analyzed by R package CAMERA were analyzed to find quasi-molecular ions and determine the corresponding molecular weights.
Second, the tandem mass spectrometry (MS/MS) experiments were implemented to get fragmentation patterns of potential biomarkers so as to produce their chemical structure information. MS/MS experiments were performed on a Waters Q-TOF micro MS.
Third, the MS/MS product ion spectrum of our metabolomic biomarkers was investigated to try to match with the structure message of metabolites with the same m/z as ours provided by online databases of HMDB (www.hmdb.ca), METLIN (http://metlin.scripps.edu/) and KEGG (http://www.genome.jp/kegg/). The mass tolerance between the measured m/z values and the exact mass of the components of interest was set to within 30 ppm. Further comparison of MS/MS spectra between the online database and experimental data was performed to rule out the unrelated candidates.
Finally, the potential biomarkers were identified according to their molecular weights and MS/MS product ion spectrum compared with those in online database. The list of metabolites after DB matching is shown in Table S2. Metabolites v271 and v282 were not ascertained due to multiple candidates of structural isomers in METLIN database.
Table S1. The metabolic information of the 16 biomarker candidates and their annotations by CAMERA.[1]
Marker / RT / M/Z / Isotopes / Adduct / Pcgroup / VIP / p-value1v14 / 123.2785 / 100.0769 / 114 / 2.19 / 4.95E-08
v379 / 381.2864 / 490.2924 / [M+Na]+ 467.303 / 29 / 1.91 / 6.00E-07
v65 / 139.0177 / 188.0705 / [8][M]+ / [M+H-NH3]+ 204.09 / 2 / 1.00 / 1.09E-06
v285 / 612.0578 / 397.225 / [39][M+2]+ / 16 / 1.89 / 1.26E-06
v39 / 139.0074 / 146.0598 / [7][M]+ / 2 / 1.02 / 1.40E-06
v66 / 139.0034 / 189.0752 / [8][M+1]+ / 2 / 1.02 / 1.97E-06
v271 / 617.3259 / 380.2133 / 218 / 1.73 / 2.58E-06
v282 / 612.0752 / 395.2257 / [39][M]+ / 16 / 1.81 / 3.38E-06
v360 / 380.9193 / 470.3156 / [52][M+2]+ / 29 / 1.83 / 3.86E-06
v358 / 381.2966 / 469.3131 / [52][M+1]+ / 29 / 1.85 / 4.55E-06
v336 / 138.9369 / 447.1366 / [2M+K]+ 204.09 / 2 / 1.22 / 5.37E-06
v46 / 139.0001 / 159.0917 / [M+H-HCOOH]+ 204.09 / 2 / 1.00 / 9.01E-06
v448 / 384.0868 / 540.3093 / [M+Na]+ 517.318 / 19 / 1.86 / 9.61E-06
v284 / 612.0833 / 396.23 / [39][M+1]+ / 16 / 1.76 / 1.02E-05
v40 / 139.0424 / 147.0639 / [7][M+1]+ / 2 / 1.05 / 1.45E-05
v357 / 381.2991 / 468.3094 / [52][M]+ / [M+H]+ 467.303 / 29 / 1.8 / 1.86E-05
Metabolites in bold and italic font were selected as potential biomarkers.
Abbreviation: RT, Retention time; M/Z, Mass to charge ratio; VIP, Variable importance in the project. 1 Nonparametric Kruskal-Wallis rank sum tests corrected by Bonferroni based on the training set.
Table S2. The chemical names and the matched online databases for the 6 metabolic biomarkers
V14 / 2-Piperidinone / HMDB11749 / MID62467 / ---
V65 / L-Tryptophan fr / HMDB00929 / MID33 / C00078
V271 / --- / --- / --- / ---
V282 / --- / --- / --- / ---
V448 / LysoPC(18:3) add / HMDB10387 / MID61697 / C04230
V357 / LysoPC(14:0) / HMDB10379 / MID61689 / C04230
Table S3. Adjusted p-values for each biomarkers in logistic regression models
Marker / The training set / The external validation setEstimate / StdErr / Wald χ2 / p-value1 / Estimate / StdErr / Wald χ2 / p-value1
v14 / -0.6674 / 0.2143 / 9.7018 / 0.0018 / -0.5965 / 0.2508 / 5.6549 / 0.0174
v65 / -0.0053 / 0.0047 / 1.2739 / 0.259 / -0.0677 / 0.0253 / 7.1453 / 0.0075
v271 / 0.5493 / 0.1863 / 8.6954 / 0.0032 / 0.6934 / 0.2223 / 9.7308 / 0.0018
v282 / 0.3728 / 0.1144 / 10.6227 / 0.0011 / 0.3619 / 0.1182 / 9.3813 / 0.0022
v448 / -1.0914 / 0.2934 / 13.8385 / 0.0002 / -0.6571 / 0.3341 / 3.8685 / 0.0492
v357 / -0.0849 / 0.0248 / 11.6707 / 0.0006 / -0.0451 / 0.0253 / 3.1862 / 0.0743
1 Adjusted for age and menopause.
Fig. S1. The identification information of potential biomarker LysoPC(14:0) (m/z 468). (a) The extracted ion chromatographic peak. (b) The mass spectrum of m/z 468 at the retention time 6.36 min. (c) The MS/MS spectrum of ion of m/z 468 at 6.36 min in a plasma sample used to match with the online databases. (d) The chemical structure of LysoPC(14:0).
Fig. S2. The identification information of potential biomarker 2-Piperidinone. (a) The extracted ion chromatographic peak. (b) The mass spectrum of m/z 100 at the retention time 2.05 min (m/z 100). (c) The MS/MS spectrum of ion of m/z 100 at 2.05 min in a plasma sample used to match with the online databases. (d) The chemical structure of 2-Piperidinone.
Fig. S3. The identification information of potential biomarker LysoPC(18:3). (a) The extracted ion chromatographic peak. (b) The mass spectrum of m/z 518 at the retention time 6.40 min (m/z 518). (c) The MS/MS spectrum of ion of m/z 518 at the retention time 6.40 min in a plasma sample used to match with the online databases. (d) The chemical structure of LysoPC(18:3).
Fig. S4. The levels of 2-piperidinone in samples of method blanks (MB, 4 samples), solvent blanks (SB, 35 samples), ovarian cancer (EOC, 80 samples) and benign ovarian tumor (BOT, 90 samples) patients.
[1] The R-package CAMERA is a Collection of Algorithms for Metabolite pRofile Annotation. Its primary purpose is the annotation and evaluation of LC-MS data. It includes algorithms for annotation of isotope peaks, adducts and fragments in peak lists.