Investigators screened serum IgG repertoires from a multicohort collection and found enrichment of antibodies against DNA repair proteins in pediatric cancer patients. The primary discovery used HuProt v4.0 proteome microarrays to profile sera from patients with nine cancer types and healthy age-matched controls. The cohort included 175 patients at diagnosis (a mix of pediatric and adult cases) and 141 healthy controls in the discovery stages, later expanded to include additional pediatric controls. Data indicated that autoantibodies targeting DNA repair proteins were detected across multiple tumor types, with a distinct age-specific distribution noted by the study team. The authors ranked significant proteins using z-scores and reported a core list of antigens that were positive in at least 10% of patients in two or more cancer groups.
The study team examined pre-diagnostic sera from UK birth cohorts (ALSPAC, BiB) and identified some of the same enriched autoantibodies in healthy children who later received cancer diagnoses. For the pre-diagnostic analysis the investigators evaluated six core antigens individually (ADRM1, DDX58, DHX58, DMC1, MGMT, and UNG) and then performed pooled analyses. Initial odds ratios were estimated with standard logistic regression; because some individuals had multiple samples, mixed-effects logistic regression models with random intercepts were fitted to account for within-individual correlation. The Benjamini–Hochberg false discovery rate procedure was applied to adjust p-values across the six antigen-specific tests. The authors created a composite binary variable indicating presence of at least one core antigen and classified individuals with multiple samples as positive if any sample tested positive. The discovery cohort lacked childhood controls, a limitation the authors acknowledged; subsequent stages included a broader age span of controls (0.6–19 years) which the investigators used to adjust pediatric analyses for false positives and to confirm the observed patterns.
To probe potential pathogenic significance, the authors tested whether serum IgG could enter live, intact cells. Cell lines and primary cells (including osteosarcoma lines, primary human fibroblasts, and primary osteoblasts) were exposed to donor serum at a 1:40 dilution for 2 hours at 37 °C. This dilution was selected after testing a range of concentrations and judged optimal for internalized IgG signal with low background. Following fixation and permeabilization, anti-human IgG antibodies were used for detection and confocal microscopy (Leica SP8-STED at 100X, z-stacks) was applied to evaluate subcellular localization. The investigators reported a specific intracellular staining pattern—referred to in the methods as a quantifiable TISa pattern—that was more prevalent in pediatric patients. Quantitative image analysis used QuPath to detect DAPI-stained nuclei and subcellular IgG spots within defined regions of interest; spot size thresholds were optimized to limit false-positive and false-negative detections. Flow cytometry was used to compare cell cycle distributions between IgG+ and IgG– cell populations; each IgG-defined subpopulation was normalized internally to analyze G0/G1, S, and G2/M distributions, reflecting internal composition rather than proportions of the total sample.
Arrays were processed according to the manufacturer’s protocol: blocking with 2% BSA overnight at 4 °C, serum diluted 1:1000 in blocking buffer and incubated 2 hours at room temperature, washes, and detection using anti-GST and anti-human IgG secondaries. Scanning was performed on a GenePix scanner with uniform settings across batches. Preprocessing included neighborhood background correction, subtraction of negative-control arrays, and duplicate-spot rsd filtering. Positive protein calls used cutoffs of z-score >2.51 (approximate p = 0.02 as per the protocol), rsd <0.35, and sample-to-control ratio >3.5. The authors identified and removed proteins affected by batch effects; they explicitly stated that proteins with inconsistent detection across batches were excluded from downstream analyses. For antigen-level inference in pre-diagnostic samples the two-stage logistic regression approach and FDR adjustment were implemented; pooled core antigen positivity was used to assess group-level associations.
The authors documented several constraints. Protein microarrays can suffer from batch effects; proteins impacted by these effects were removed from analyses. The discovery cohort did not include childhood controls, which could produce false positives for autoantibodies common in healthy children; the authors addressed this in later study stages by adding pediatric controls and confirming earlier patterns. The ALSPAC pre-diagnostic data lacked complete information on histological diagnoses for some cases. The source did not report mechanistic experiments demonstrating that the identified autoantibodies directly induce oncogenic transformation, nor did it report in vivo functional validation. The complete proteome array dataset is deposited in ArrayExpress (accession E-MTAB-13995) and ALSPAC data access is available via the cohort’s searchable data dictionary as described by the authors. Remaining questions recorded by the study team include whether the identified autoantibodies are causal, contribute to tumor initiation, or represent biomarkers of an underlying predisposition; the source did not provide definitive answers to these mechanistic uncertainties.