Lynch syndrome: influence of additional susceptibility variants on cancer risk

Some patients with Lynch syndrome (LS) have extreme phenotypes, i.e. cancer before the recommended screening age, or cancer for which there are no screening guidelines. We made the hypothesis that additional germline variants in cancer susceptibility genes (CSG) could explain some of these phenotypes. We compared the prevalence of additional CSG variants in LS patients with a cancer diagnosis before age 30 (early-onset, EO group) and after 40 (usual-onset, UO group). While there was no overall difference, we did find an excess of pathogenic variants and variants of unknown significance in EO cases when only gastrointestinal CSG were considered (OR 2.25; 95% CI: 1.01–5.06, p value = 0.04). Four EO cases stood out: two with POLE/POLD1 variants in the key exonuclease domain, one with a BMPR1A duplication and one with an EPCAM deletion. Additional germline variants should be considered in future screening recommendations, as they might influence cancer risk.


INTRODUCTION
Recent European and British guidelines recommend gene-based colorectal cancer (CRC) screening in patients with Lynch syndrome (LS) [1,2].More precisely, carriers of pathogenic variants (PV) in MSH6 and PMS2 are advised to start colonoscopy screening at the age of 35 years, opposed to age 25 for MLH1 and MSH2 PV carriers.In addition, European guidelines propose increased screening intervals for PMS2 PV carriers, i.e., colonoscopy every 5 years versus every 2-3 years.A simulation model-based analysis supports such gene-specific surveillance, as it seems to have favorable impact regarding quality of life and cost, with no deleterious consequences regarding cancer incidence [3].
The later CRC onset associated with MSH6 and PMS2 in the Prospective Lynch Syndrome Database (PLSD) justifies these geneadjusted guidelines [4].If no early-onset CRC were associated with MSH6 and PMS2 in the PLSD cohort, there are isolated reports in the literature.For example, two Austrian siblings carrying a PMS2 PV developed CRC in their teenage years [5].A likely explanation was the presence of a contributing variant in the POLD1 exonuclease domain.Both CRC showed isolated PMS2 expression loss, and mutational signatures in favor of MMR-deficiency combined with POLD1-associated proofreading defect.In addition, we knew of local LS patients who had developed cancer before the recommended screening age.
A continuum between LS and constitutional mismatch repair deficiency (CMMRD) has been suggested, with the description of LS patients presenting a CMMRD-like phenotype, and carrying candidate variants, in addition to the LS causing variant [6].More generally, cancer phenotypes associated with multiple variants in cancer susceptibility genes (CSG) have been designated as Multilocus Inherited Neoplasia Allele Syndrome (MINAS) [7].We made the hypothesis, as speculated by the MINAS investigators, that additional variants in other CSG might explain some of our extreme LS phenotypes.We therefore studied retrospectively LS patients with an associated cancer diagnosis under the age of 30 years and looked for second germline susceptibility variants.We compared our observations to a series of LS cases with a cancer diagnosis at more usual ages.

SUBJECTS AND METHODS Patient selection
The study was retrospective.LS patients were selected from four Paris university hospitals: Pitié-Salpêtrière, Saint-Antoine (both Sorbonne Université), Cochin Hospital, and Hôpital Européen Georges Pompidou (both Université Paris Cité).We selected two patient subgroups: 1. earlyonset (EO), with a Lynch-associated cancer diagnosis between ages 16-30; 2-usual-onset (UO) with a Lynch-associated cancer diagnosis ≥age 40.All had received a LS diagnosis between 2016 and 2021.They had signed an informed consent form before germline testing.CMMRD had been excluded in EO patients fulfilling clinical criteria using functional tests [8].

Panel testing and variant analysis
Germline testing was performed in the Pitié-Salpêtrière and Cochin cancer genetics laboratories.A 34-CSG panel was analyzed (Table S1).Coding sequences were sequenced after enrichment by capture (Roche Nimble-Gen kit).PV were confirmed by Sanger sequencing and copy number variations by MLPA.Large PMS2 rearrangements were confirmed by longrange PCR.Variants were classified according to the ACMG criteria [9] and/ or national or international variant databases.

Tumor mutational burden and mutational signatures in selected tumors
Tumor mutational burden (TMB) and mutational signatures were studied using Maftools [10] and the R DeconstructSigs package (COSMIC signatures), respectively, on sequencing data generated with the AVENIO Tumor Tissue Comprehensive Genomic Profiling Kit (Roche Sequencing Solutions, Pleasanton, CA) (duodenal carcinoma) and the Dana Farber Cancer Institute OncoPanel (glioblastoma) [11].

Ethics approval and patient information
The study was approved by the ethics evaluation committee (IRB00003888, IORG0003254, FWA00005831) of the French Institute of medical research and Health (INSERM, Opinion number 22-878).All patients (or their next of kin if deceased) were informed in writing of this project, with a possibility to opt out, in accordance with French law.

RESULTS
One hundred and forty-one patients were eligible, 45 and 96 in the EO and UO groups, respectively.Six declined participation, leaving us with 43 EO cases and 92 UO cases (Table 1; Table S2).
EO patients with a second gastrointestinal CSG PV/VUS are listed in Table 3.Two EO cases were reminiscent of the young PMS2 PV carriers who developed CRC [5].Both carried a variant in a polymerase gene exonuclease domain alongside the MMR PV.One teenage female aged 17 had duodenal adenocarcinoma with PMS2 expression loss in the tumor.Family history was negative.She had no other features suggesting CMMRD.Germline testing showed the PMS2: NC_000007.13(NM_000535.5):c.989-2A > G, p.(?) splice acceptor PV.In addition, she carried the c.895A > G, p.(Met299Val) VUS in POLE(NM_006231.4).This variant is located in the key exonuclease domain.In silico analysis predicts a deleterious effect on protein structure/function.It involves a highly conserved amino acid across species [13] and is only observed at a frequency of 7.9×10 -6 in gnomAD v2.1.1.TMB was 111 mutations/Mb.The SBS14 MMR/POLE signature was identified in the tumor, alongside four MMR signatures (SBS6, 15, 21 and 26) (Figure S1).The second case was diagnosed with glioblastoma,  Fisher's exact tests were used to determine associated odd-ratios (OR) and 95% confidence intervals (CI).
a Some cases carried more than one variant, see Table 3.
Table 3. Early-onset patients with a second pathogenic variant or variant of unknown significance in a gastrointestinal cancer susceptibility gene.IDH-wild type aged 22. Expression was lost for MSH2/MSH6.A paternal uncle had developed pancreatic cancer in his fifties.An MSH2 exon 9-10 deletion and the POLD1(NM_001256849.1):c.961 G > A, p.(Gly321Ser) VUS were identified.The POLD1 VUS is located within the exonuclease domain, is only reported at a 3.5×10 -4 frequency in gnomAD and involves a highly conserved amino acid across species [12].Again, a pathogenic effect is predicted in silico.TMB was 34.2 mutations/Mb.We could not detect a typical POLD1 signature.Neither case had gastrointestinal adenomatous polyps.Importantly, no POLD1/POLE exonuclease domain variants were identified among UO cases.
Another patient was diagnosed with CRC aged 21, four years earlier than the recommended screening age.There was MLH1/ PMS2 expression loss.He carried two PV, MLH1(NM_000249.3):c.199 G > T, p.(Gly67Trp) and a complete EPCAM deletion.
Finally, a case was diagnosed with CRC aged 25.The tumor was MMR-deficient, with MSH2/MSH6 expression loss.He carried a MSH2 PV and a BMPR1A exon 3-13 duplication, of unknown significance.No juvenile polyps were observed.Twenty-five is the age at which the first screening colonoscopy is recommended in MSH2 PV carriers.

DISCUSSION
Recent literature supports the existence of a continuum of phenotypes ranging from CMMRD to classical LS.However, publications so far have consisted of case reports or small-size case series [6].In this study including a total of 135 LS patients, we explored whether the prevalence of additional germline susceptibility variants was higher in patients with extreme phenotypes, i.e. early-onset cancer, versus those with cancer at more usual ages.The overall prevalence of additional PV and VUS in CSG was similar in the EO and UO groups.However, when selecting CSG specifically associated with gastrointestinal cancer, more PV/VUS were observed in LS EO cases.
Among EO cases, four stood out.One PMS2 PV carrier who developed duodenal cancer aged 17 also carried a POLE-exonuclease domain variant.TMB was very high, with an MMR/POLE mutational signature, indicating an involvement of both variants in carcinogenesis.A MSH2 case with glioblastoma aged 22 also carried a POLD1-exonuclease domain variant.A patient with CRC aged 21 turned out to carry a MLH1 missense PV and a whole EPCAM deletion, encompassing known pathogenic 3ʹUTR deletions inducing MSH2 promoter methylation in cis [13].It would have been interesting to study MSH2 methylation on the CRC, but no tissue was available.Finally, a patient with CRC aged 25 carried a MSH2 PV and a BMPR1A exon 3-13 duplication.
In these four cases, cancer occurred either in at-risk organs for which no screening is formally recommended (duodenal cancer, glioblastoma), before the recommended screening age or just when screening should be initiated (colorectal cancer) [2,14].They resemble previously reported cases with extreme phenotypes linked to a MMR-polymerase genes variant combination [5].
We have grouped all additional CSG PV and VUS together.We acknowledge that VUS are not all the same, and that variants that actually have no biological effect might have been included.The study was too small for secondary analyses based on pathogenicity probability.In addition, creating two or three categories of VUS would have introduced subjectivity, while we preferred to follow the ACMG criteria.
Six cases were heterozygotes for a MUTYH PV, among them two EO cases with the common c.1145 G > A, p.(Gly382Asp) PV.MUTYHassociated polyposis is indeed a recessive disease, but we cannot rule out a moderate CRC risk increase in older patients [15].
We have focused on rare CSG PV and VUS.Other genetic contributing factors, such as polygenic risk scores could be relevant [16].Llach et al. recently envisaged the possibility of personalized cancer screening in LS patients on additional contributing factors [17].Our study suggests that additional gastrointestinal CSG PV/VUS are such factors, and that they should be considered in future screening recommendations.More precisely, earlier CRC screening, and expanded screening for cancers less commonly seen in LS, might be warranted.POLD1 and POLE are of particular interest.

Table 2 .
Number of Lynch syndrome patients with additional pathogenic variants (PV) or variants of unknown significance (VUS) in all 34, or among the 14 digestive cancer susceptibility genes (CSG).