Flowchart showing the number of patients included and the workflow of the data analysis.

UCT: University Cancer Center Frankfurt, PCA: principal component analysis, ESO: emergent self-organizing maps, cABC analysis: computed ABC analysis. The figure was created using Microsoft PowerPoint® (Redmond, WA, USA) on Microsoft Windows 11 running in a virtual machine powered by Virtual Box 6.1.36 (Oracle Corporation, Austin, TX, USA) as guest on Linux, and then further modified with the free vector graphics editor “Inkscape (version 1.2 for Linux, https://inkscape.org/

Results of a projection of the z-standardized log-transformed lipidomics data onto a lower-dimensional space by means of a self-organizing map of artificial neurons (bottom).

a): 3D display of an emergent self-organizing map (ESOM), providing a 3-dimensional U-matrix visualization (78) of distance-based structures of the serum concentrations of d = 255 lipid mediators following projection of the data points onto a toroid grid of 4,000 neurons where opposite edges are connected. The dots represent the so-called “best matching units” (BMU), i.e., neurons on the grid that after ESOM learning carried a data vector that was most similar to a data vector of a sample in the data set. Only those neurons of the originally 4000 neurons are shown that carried vectors of cases from the present data set. Please also note that one BMU can carry vectors of several cases, i.e., the number of BMUs is not necessarily equal to the number of cases. A cluster structure emerges from visualization of the distances between neurons in the high-dimensional space by means of a U-matrix (79). The U-matrix was colored as a geographical map with brown or snow-covered heights and green valleys with blue lakes, symbolizing high or low distances, respectively, between neurons in the high-dimensional space. Thus, valleys (left and right of the “mountain range” in the middle indicate clusters and watersheds, i.e.,, i.e., the line of large distances between neighboring points, indicate borderlines between different clusters. Tat is, the mountain range with “snow-covered” heights separates main clusters according to probe acquisition at day 1 or day 2, i.e., before and after treatment with paclitaxel. BMUs belonging to the two different clusters are colored green or bluish. b): Mosaic plot of the prior classes (day 1 or day 2) versus the ESOM/Umatrix based clusters. The separation corresponded to the previous classification into pre- and post-therapy probes (day1/2). Cluster #1 was composed of more probes taken on day #1, while probes from day 2 were overrepresented in cluster #2. The figure has been created using the R software package (version 4.1.2 for Linux; https://CRAN.R-project.org/ (26)), R library “ggplot2” (https://cran.r-project.org/package=ggplot2 (80)) and our R package “Umatrix” (https://cran.r-project.org/package=Umatrix (37)).

Lists of lipid mediators that were most informative in assigning a sample (i) to the first or second sampling time point or (ii) a sample from the second time point to a patient with or without neuropathy.

Abbreviations: SA1P: sphinganine-1-phosphate, S1P: sphingosine-1-phosphate, LPE: lysophosphatidylethanolamine, LPC: lysophosphatidylcholine, 2-AG: 2-arachidonoylglycerol, OEA. Oleoylethanolamide.

Internal validation of the sets of lipid mediators resulting from the feature selection analysis.

The different classifiers (linear support vector machine, SVM, random forests, and logistic regression) were trained with subsets of the training data set with all variables (d = 255 lipid mediators as “full” feature set and with the d = 77 or d = 27 lipid mediators that had resulted from the recursive cABC analysis applied on the sum score of selections by 17 different feature selection methods as “reduced” or “sparse” feature sets, respectively. The trained classifiers were applied to a validation sample comprising 20% of the data that had been removed in a class-proportional manner from the dataset at the beginning of feature selection and had not been touched until used in the classifier validation task presented in this table. In addition, the validation task was repeated with training the classifiers with permuted lipid mediators to observe possible overfitting. Shown are the medians and nonparametric 95% confidence intervals (2.5th to 97.5th percentiles) from 5 x 20 nested cross-validation runs. Results of external validation in an independent cohort are shown in the supporting information (Table S3).

Log10-transformed concentrations of lipid mediators shown to be informative for assigning a post-therapy sample to a patient with neuropathy or a patient without neuropathy.

Individual data points are presented as dots on violin plots showing the probability density distribution of the variables, overlaid with box plots where the boxes were constructed using the minimum, quartiles, median (solid line inside the box) and maximum of these values. The whiskers add 1.5 times the interquartile range (IQR) to the 75th percentile or subtract 1.5 times the IQR from the 25th percentile. a): Concentrations of SA1P (top hit for sample 1 versus sample 2 segregation) are presented separately for the first and second samples. b): Concentrations of the top lipid mediators for neuropathy versus no neuropathy in the second sample presented separately for neuropathy-positive and -negative samples. The results of the group comparison statistics (Kruskal-Wallis tests (58)) are given at the top of the graphs. The figure has been created using the R software package (version 4.1.2 for Linux; http://CRAN.R-project.org/ (26)) and the R library “ggplot2” (https://cran.r-project.org/package=ggplot2 (80)).

Effects of sphinganine-1-phosphate on primary sensory neurons.

a) Neurons were stimulated with SA1P (1 or 10 µM, 1min or vehicle (0.7% methanol (v/v)). b) percentage of responding neurons to vehicle (0.7% methanol (v/v), 1 min), (SA1P (1 µM, 1min), AITC (allyl isothioncyanate, 75 µM, 30s) or capsaicin (caps, 200 nM, 20s). c) representative traces of SA1P-responding neurons and their response to AITC, capsaicin and KCl. d) percentage of SA1P-responding neurons responding to AITC, capsaicin (caps), AITC and capsaicin and KCl. Data are shown as mean ± SEM from at least six measurements per condition with at least 40 neurons per measurement, * p < 0.05, ** p < 0.01, *** p < 0.01, One-way ANOVA.

Contribution of TRPV1 and S1P receptors to SA1P-mediated calcium-influx in sensory neurons.

Sensory neurons were stimulated with SA1P twice (1 µM, 1 min) and either a) vehicle (DMSO 0.003% (v/v), 2 min) or b) the TRPV1 antagonist AMG9810 (1 µM, 2 min) prior to the second SA1P stimulus. Cells were depolarized with KCl (50 mM, 1 min) at the end of each experiment. c) Statistical analysis of the amplitude of SA1P-mediated calcium transients in sensory neurons treated with either vehicle or AMG9810 (blue). d) Statistical analysis of the amplitude of capsaicin-mediated calcium transients (100 nM, 20s) in sensory neurons treated with either vehicle or AMG9810 (blue). e), f) Sensory neurons were stimulated with SA1P after preincubation with the S1P1 receptor modulator fingolimod (1 µM, 1 h) or control. g) Statistical analysis of the amplitude of SA1P-mediated calcium transients (1 µM, 1min) in sensory neurons treated with either vehicle or fingolimod (1 µM, 1h, orange). h) Statistical analysis of the number of SA1P-responding neurons (as % of KCl-positives) after treatment with either vehicle or fingolimod (1 µM, 1h, orange). Data represents mean ± SEM from at least five measurements per condition with at least 25 neurons per measurement, * p < 0.05, ** p < 0.01, *** p < 0.01, Student’s t-test with Welch’s correction.

Log10-transformed concentrations of SA1P in the second patient cohort.

Individual data points are presented as dots on violin plots showing the probability density distribution of the variables, overlaid with box plots where the boxes were constructed using the minimum, quartiles, median (solid line inside the box) and maximum of these values. The whiskers add 1.5 times the interquartile range (IQR) to the 75th percentile or subtract 1.5 times the IQR from the 25th percentile. a): Concentrations of SA1P (top hit for sample 1 versus sample 2 segregation) are presented separately for the first and second samples. b): Concentrations of S1AP in the second sample are shown separately for neuropathy-positive and -negative samples. Day 1 represents the timepoint before starting chemotherapy. Day 2 represents the timepoint after 12 cycles of paclitaxel chemotherapy. The results of the t-test group comparison statistics are given at the top of the graphs. The figure has been created using the R software package (version 4.1.2 for Linux; http://CRAN.R-project.org/ (26)) and the R library “ggplot2” (https://cran.r-project.org/package=ggplot2 (80)).

Sphingolipids and Ceramides

(SPT: Serine palmitoyl-transferase; 3KR: 3-ketosphinganine reductase; SPHK: Sphingosine kinase; CerS: Ceramide synthase; DEGS: Dihydroceramide desaturase, GlCerS: Glucosylceramide synthase; LacCerS: Lactosylceramide synthase; SMS: Sphingomyelin synthase; CDase: Ceramidase). Structures were drawn with ChemDraw 20.