Germline variants observed in pediatric cancer patients related to hereditary breast and ovarian cancer in adults

Authors

Katharina Daugs

Danielle Brandes

Layal Yasin

Ammarah Anwar

Jubayer Alam

Yash Prasad

Jil Bartrina y Manns

Melina Mescher

Ute Fischer

Arndt Borkhardt

Triantafyllia Brozou

Stefanie V. Junk

Doi

10.1002/ijc.70097

PMID: 29489754 · DOI: 10.1002/ijc.70097 · Journal: International Journal of Cancer (2025)

TL;DR

Daugs et al. performed whole-exome sequencing on a single-center prospective cohort of 372 pediatric cancer patients (Düsseldorf, 2015–2023) and assessed (likely) pathogenic germline variants (LP/PVs) in 25 genes associated with adult-onset hereditary breast and ovarian cancer (HBOC). Twenty-seven of 372 (7.3%) patients carried LP/PVs; 12/27 (44%) had no clinically suspected cancer predisposition syndrome (CPS) prior to genotyping. LP/PV carriers had a markedly elevated risk for second malignant neoplasms (SMNs; 18.5% vs. 3.8%, OR=5.8, p=.0021), with all five SMNs occurring in carriers of TP53 (n=3), NBN (n=1), or ATM (n=1) variants. Single-cohort burden testing against gnomAD v3.1.1 non-cancer controls flagged monoallelic LP/PVs in TP53, CHEK2, ATM, NF1, and NBN as significantly enriched; a joint analysis with the 1120 patients from Zhang et al. (2015) confirmed strong associations for TP53, CHEK2, NF1, and MSH2. The authors argue for systematic genotype-driven screening because clinical criteria miss ~44% of carriers.

Cohort & data

  • 372 pediatric cancer patients enrolled prospectively in the “Germline Mutations in Children with Cancer” study at the Department of Pediatric Oncology, Hematology and Clinical Immunology, Heinrich Heine University Düsseldorf (January 2015 – January 2023). Median age at first diagnosis 5.1 years (range 0–22.2); 160 girls (43%) PMID:29489754.
  • Disease distribution: 212 (57%) hematologic neoplasms, 71 (19%) brain tumors, 89 (24%) various solid entities. Among LP/PV carriers (n=27): 10 leukemias (37%), 5 lymphomas (18.5%), 3 brain tumors (11.1%), 8 solid tumors (29.6%), 1 other (3.7%) PMID:29489754.
  • Assay: trio whole-exome sequencing; variant pathogenicity assessed by standard diagnostic guidelines (ACMG-style classification). Variant positions on GRCh38 (hg38) PMID:29489754.
  • Candidate gene set: 25 selected HBOC-related genes, drawn from the literature (Table S1) PMID:29489754.
  • Statistical comparator: gnomAD v3.1.1 non-cancer dataset (n=74,023 healthy adults) for gene-level burden testing using logistic regression and two-sided Fisher’s exact tests; results considered relevant at p<.05 PMID:29489754.
  • Joint-analysis cohort: Genotype data from 1120 pediatric cancer patients in Zhang et al. (2015) added to augment statistical power (Table S6, Figure S3) PMID:29489754.
  • Data availability: WES data for the 27 LP/PV carriers deposited in EGA under accession EGAD50000001561 PMID:29489754.

Key findings

  • 7.3% of pediatric cancer patients carried LP/PVs in HBOC genes. 28 LP/PVs were found across 27/372 patients (7.3%): 13 likely pathogenic and 15 pathogenic, plus 187 (31.5%) variants of uncertain significance among the 593 total variants in 297 patients PMID:29489754.
  • 44% of carriers were clinically unsuspected. 12 of 27 LP/PV carriers (44.4%) did not meet the German Cancer Society (DKG) CPS criteria prior to genotyping; only 14/135 (10.4%) of clinically suspected CPS cases harbored an HBOC-related LP/PV PMID:29489754.
  • Carriers had a 5.8-fold increased risk of second malignant neoplasms. SMN incidence: 5/27 (18.5%) in carriers vs. 13/345 (3.8%) in non-carriers; OR=5.8 (95% CI=1.89–17.75), p=.0021. All five SMNs occurred in carriers of TP53 (n=3), NBN (n=1), or ATM (n=1) PMID:29489754.
  • Twelve of the 25 candidate genes harbored at least one LP/PV. Distribution across 28 LP/PVs: TP53 6 (21.4%), CHEK2 5 (17.9%), ATM 4 (14.3%), NBN 3, BRIP1 2, NF1 2, BRCA1 1, BLM 1, FANCC 1, FANCM 1, MSH2 1, STK11 1 PMID:29489754.
  • Single-cohort burden testing flagged five HBOC genes as significantly enriched vs. gnomAD non-cancer controls: TP53 (OR=32.8, 95% CI 13.8–78.1, p=7.83×10⁻⁸), CHEK2 (OR=9.6, 95% CI 3.9–23.7, p=2.42×10⁻⁴), ATM (OR=3.2, 95% CI 1.2–8.6, p=.040), NF1 (OR=9.8, 95% CI 2.4–40.5, p=.020), and NBN (OR=9.4, 95% CI 2.3–38.7, p=.021) PMID:29489754.
  • STK11 showed an extreme but wide-CI association (OR=199.5, 95% CI 12.5–3195.8, p=.010); the single carrier (LPP_27) presented with congenital Peutz-Jeghers syndrome features and a T-cell ALL with maternal-line family history PMID:29489754.
  • Joint analysis with Zhang et al. 2015 (n=1120) confirmed associations for TP53, CHEK2, NF1, and MSH2 with pediatric cancers; MSH2 became significant only after pooling (OR=7.1, 95% CI 1.6–31.2, p=.0390). ATM, NBN, BRCA1, and other less-constrained genes did not retain significance in the joint analysis PMID:29489754.
  • Most LP/PVs were pathogenic loss-of-function (pLoF) variants. All LP/PVs in ATM, NBN, BRIP1, STK11, MSH2, FANCM, BLM, and FANCC were pLoF. Six missense LP/PVs were found in TP53 (n=2), CHEK2 (n=3), and BRCA1 (n=1); one truncating intronic splice-region LP/PV was found in NF1 PMID:29489754.
  • **One patient (LPP_12) was homozygous for the NBN c.657_661del p.(Lys219Asnfs*16) (rs587776650) founder mutation** with typical clinical features of Nijmegen breakage syndrome (NBS) and a T-cell lymphoma; he was excluded from monoallelic burden testing. Two additional patients carried this same NBN founder variant in heterozygous form and developed hematologic neoplasms without NBS-related symptoms PMID:29489754.
  • One patient (LPP_20) carried two distinct heterozygous LP/PVs in different genes: CHEK2 c.1165C>T p.(Arg389Cys) and NF1 c.4137dup p.(Ala1380Serfs*15). She was an 11.4-year-old girl with Langerhans cell histiocytosis, no neurofibromatosis type 1 features, but a positive family history of late-onset cancers (breast cancer, acute and chronic myeloid leukemias) PMID:29489754.
  • **The recurrent LFS-related TP53 variant c.586C>T p.(Arg196*) was detected in two unrelated patients** (LPP_07, de novo, T-cell ALL with lineage switch to BCP-ALL; LPP_10, who developed adenocarcinoma at 1.8y, BCP-ALL at 4.8y, then AML during ALL maintenance) PMID:29489754.
  • Highly constrained genes drove the strongest pediatric-cancer associations. Among candidate genes, NF1, MSH2, and STK11 are highly constrained (LOEUF<.35); TP53 and ATM are moderately constrained; CHEK2 (LOEUF=1.522) and NBN (LOEUF=1.010) are less constrained. Constraint correlated with pLoF tolerance, consistent with prior work (Stoltze et al., Nat Commun 2024) PMID:29489754.

Genes & alterations

  • TP53 — Six LP/PV carriers (21.4% of LP/PV carriers); strongest single-gene burden-test association (OR=32.8, p=7.83×10⁻⁸). Recurrent c.586C>T p.(Arg196*) detected twice (LPP_07 de novo, LPP_10 with LFS). Three of six TP53 carriers (50%) developed SMNs; carriers presented with rhabdomyosarcoma, osteosarcoma, or BCP-ALL. Five of six had a positive family history of cancer; two had relatives who also experienced pediatric cancer PMID:29489754.
  • CHEK2 — Five LP/PVs (17.9% of LP/PV carriers); burden OR=9.6, p=2.42×10⁻⁴. Four in BCP-ALL patients and one in a Langerhans-cell-histiocytosis patient (LPP_20, with co-occurring NF1 LP/PV). Three of three pLoF (40% of CHEK2 LP/PVs were pLoF) plus three missense PMID:29489754.
  • ATM — Four monoallelic LP/PVs (all pLoF, 100%); burden OR=3.2, p=.040 in single-cohort analysis but not significant in joint analysis. Two BCP-ALL patients (LPP_02, LPP_03) and one each with medulloblastoma and hepatoblastoma (LPP_01, LPP_04). One ATM carrier developed a secondary event. None had ataxia-telangiectasia features PMID:29489754.
  • NF1 — Two LP/PVs (one pLoF, one truncating intronic splice-region variant); burden OR=9.8, p=.020 single-cohort; confirmed in joint analysis. One carrier (LPP_20) co-carried CHEK2 LP/PV and lacked typical NF1 clinical features PMID:29489754.
  • NBN — Three LP/PVs in total: one homozygous (LPP_12, classical NBS phenotype + T-cell lymphoma) excluded from monoallelic burden testing; two heterozygous carriers of the same c.657_661del p.(Lys219Asnfs*16) (rs587776650) founder mutation developed hematologic neoplasms without NBS features. Burden OR=9.4, p=.021 single-cohort; one heterozygous carrier developed an SMN PMID:29489754.
  • BRIP1 — Two pLoF LP/PVs; burden test not significant (OR=4.2, p=.085) PMID:29489754.
  • STK11 — One pLoF LP/PV: c.179dup p.(Tyr60*) in patient LPP_27, a boy with congenital Peutz-Jeghers syndrome and T-cell ALL; mother and maternal grandfather also have PJS. Single-cohort OR=199.5 (95% CI 12.5–3195.8), p=.010 — extreme estimate driven by very low gnomAD baseline (1/74,022 controls) PMID:29489754.
  • MSH2 — One pLoF LP/PV; not significant in single cohort but reached significance in joint analysis with Zhang et al. (OR=7.1, p=.0390). Authors highlight MSH2 as the only DNA-repair gene consistently associated with childhood cancer in a prior meta-analysis PMID:29489754.
  • BRCA1 — One missense LP/PV; burden test non-significant (OR=1.9, p=.406) PMID:29489754.
  • BLM — One pLoF LP/PV; non-significant (OR=1.7, p=.450) PMID:29489754.
  • FANCC — One pLoF LP/PV; non-significant (OR=1.3, p=.534) PMID:29489754.
  • FANCM — One pLoF LP/PV; non-significant (OR=2.2, p=.370) PMID:29489754.
  • BRCA2 — Listed among the 25 candidate HBOC genes; no LP/PVs were detected in this cohort PMID:29489754.

Clinical implications

  • Genotype-driven screening identifies clinically unsuspected carriers. 44% of HBOC-related LP/PV carriers in this cohort failed standard DKG CPS criteria; the authors argue for systematic WES regardless of phenotype-based suspicion to enable risk-adapted therapy and surveillance PMID:29489754.
  • HBOC LP/PV carriers warrant SMN surveillance. A 5.8-fold elevated SMN risk concentrated in TP53, NBN, and ATM carriers supports intensified surveillance protocols (analogous to the Toronto Protocol for TP53 carriers cited via Villani et al. 2016) and consideration of reduced-toxicity treatment regimens PMID:29489754.
  • Therapy de-escalation can mitigate genotype-related toxicity. Patient LPP_10 (LFS, BCP-ALL → AML) received a personalized chemotherapy regimen with reduced doses of cytarabine and gemtuzumab and, after stem-cell rejection from her mother, achieved remission following haploidentical HSCT from her maternal grandmother PMID:29489754.
  • Severe neurotoxicity in an ATM / TP53-spectrum context can be lethal. Patient LPP_07 developed muscle atrophy progressing to tetraparesis during HSCT conditioning with fludarabine and died one month later — illustrating chemotoxicity risk in carriers of DNA-damage-response gene LP/PVs PMID:29489754.
  • Monoallelic variants in autosomal-recessive HBOC genes may also matter. Four monoallelic ATM carriers (no AT features) and two heterozygous NBN carriers (no NBS features) developed pediatric cancers; the authors hypothesize that single-allele ATM / NBN impairment contributes to risk and merits further investigation PMID:29489754.
  • Family-history alone underestimates HBOC predisposition in pediatric oncology. Only 22% of HBOC LP/PV carriers had a positive family history (vs. ~50% for general CPGs in prior literature), suggesting either lower familial penetrance for these genes or underrepresentation in the studied cohort PMID:29489754.
  • Cascade testing of relatives is implied but not formally evaluated. Identifying HBOC LP/PVs in pediatric probands may also enable identification of at-risk adult relatives PMID:29489754.

Limitations & open questions

  • Limited candidate-gene set. Restricting analysis to 25 HBOC-related genes excludes related DNA-repair / mismatch-repair genes (e.g., MSH6, ERCC2, CDKN2A) that prior work has implicated in pediatric cancer predisposition PMID:29489754.
  • No copy-number variation analysis. Technical limitations of the WES pipeline excluded CNVs; carrier frequencies in genes susceptible to large deletions (NF1, BRCA1/2) may be underestimated PMID:29489754.
  • No second-hit / tumor analysis. Lack of available tumor material precluded investigation of somatic second hits to corroborate driver status in carriers PMID:29489754.
  • Variable follow-up time. Heterogeneous follow-up durations may underestimate SMN incidence, particularly in recently enrolled patients PMID:29489754.
  • Single-center cohort. 372 patients from one Düsseldorf institution; the joint analysis with Zhang et al. (2015) only partially mitigates statistical-power constraints. Wide confidence intervals on rare-variant ORs (e.g., STK11 OR=199.5, 95% CI 12.5–3195.8) reflect this.
  • Comparator may inflate ORs. Using gnomAD non-cancer adults as the control population assumes no enrichment of rare HBOC LP/PVs in adults who simply have not yet developed cancer; ascertainment bias is possible PMID:29489754.
  • Multiple-testing correction not explicitly applied. Twelve gene-level burden tests with p<.05 cutoff; only TP53 and CHEK2 survive any reasonable Bonferroni correction in the single-cohort analysis.
  • Penetrance and age-at-onset for monoallelic AR variants are unknown. Whether monoallelic ATM or NBN variants act as low-penetrance alleles or modifiers remains an open question that this study highlights but cannot resolve.
  • No prospective evaluation of risk-adapted therapy. The clinical-utility argument is implicit; the study does not benchmark outcomes for genotype-guided treatment vs. standard care.

Citations from this paper used in the wiki

  • “27 of 372 patients (7%) carried LP/PVs in the candidate genes; for 12 of 27 (44%) no CPS was suspected prior to genotyping” (Abstract).
  • “LP/PV carriers were particularly at risk for second malignancies (SMN; 5/27 vs. 13/345; OR=5.8; p=.0021); yet, LP/PVs in SMN-developing patients resided exclusively in TP53 (n=3), NBN (n=1), and ATM (n=1)” (Abstract).
  • “Burden testing of our single-center cohort revealed considerable associations between monoallelic LP/PVs in five HBOC-related genes (TP53, CHEK2, ATM, NF1, and NBN) and pediatric cancers compared to healthy adults (gnomAD v.3.1.1, non-cancer dataset)” (Abstract).
  • “Joint analyses adding 1120 individuals from a previous study Zhang et al. (2015) confirmed significant associations for TP53, CHEK2, NF1, and MSH2” (Abstract).
  • “Overall, 593 variants were identified in 297 of 372 analyzed pediatric cancer patients: 291 (49.1%) benign, 87 (14.7%) likely benign, 13 (2.2%) likely pathogenic, 15 (2.5%) pathogenic, and 187 (31.5%) variants of uncertain significance (VUS)” (Results).
  • “We classified a total of 28 variants in 27 patients (7.3%) as LP/PVs” (Results).
  • “one girl (LPP_20) carried two heterozygous LP/PVs in distinct genes (CHEK2 and NF1) and one boy with typical clinical features of NBS (LPP_12) was homozygous for NBN(ENST00000265433.7):c.657_661del p.(Lys219Asnfs*16), rs587776650” (Results).
  • “we determined LP/PVs in 12 of the 25 genes …: six of 28 (21.4%) in TP53, five (17.9%) in CHEK2, and four (14.3%) in ATM. A further 13 LP/PVs were detected in NBN (n=3), BRIP1 (n=2), NF1 (n=2), BRCA1 (n=1), BLM (n=1), FANCC (n=1), FANCM (n=1), MSH2 (n=1), and STK11 (n=1)” (Results).
  • “Burden testing revealed significantly higher ORs for patients in our cohort for heterozygous LP/PVs in TP53, NF1, NBN, CHEK2, and ATM” (Results).
  • “this joint burden testing confirmed the strong associations of TP53, NF1, and CHEK2 with pediatric cancers. In contrast to our initial approach, we determined that MSH2 LP/PVs carriers are also particularly at risk of developing pediatric cancer (OR=7.1, 95% CI=1.6–31.2, p=.0390)” (Results).
  • “Overall, LP/PV carriers in our cohort had a 5.8-fold increased SMN risk (95% CI=1.89–17.75, p=.0021)” (Discussion).
  • “The WES data of the 27 LPP cases identified with LP/PVs in this study is available in EGA under accession number EGAD50000001561” (Data Availability).

This page was processed by crosslinker on 2026-05-15.