Fig. 2: Differential expression analyses comparing patients later diagnosed with cancer to non-cancer in patients with non-specific symptoms of cancer.

A Volcano plot summarizing differential expression results comparing cancer and non-cancer patients in the discovery cohort (n = 456, cancer = 160, non-cancer = 296). Proteins upregulated also in the replication cohort are colored in red. P-values were derived from two-sided moderated t-tests from limma linear models and adjusted for multiple testing using the Benjamini–Hochberg method. B Left: Forest plot displaying odds ratios (ORs) per 1 NPX increase and 95% confidence intervals (log scale) from logistic regression models adjusted for age and sex, for proteins differentially expressed between cancer and non-cancer patients. Error bars represent the 95% confidence intervals, and data are centered on the estimated OR. Corresponding p-values are listed in Supplementary Table S5. Right: Characterization of the differentially expressed proteins according to whether they are cancer-related, secreted or tissue enriched according to the Human Protein Atlas annotation (v24.proteinatlas.org), or associated with pan-cancer according to Álvez et al.²⁹. C Protein levels of the top significant differentially expressed proteins in the discovery cohort, presented in the discovery (n = 456, cancer = 160, non-cancer = 296) and replication cohorts (n = 238, cancer = 35, non-cancer = 203). Box plots display the median (center line), interquartile range (box bounds), and whiskers representing the 1.5× interquartile range. Statistical analyses were performed on data derived from independent biological samples. No technical replicates were included. Sample sizes (n) indicate the number of individuals included in each analysis.