In around 30% of families with colorectal adenomatous polyposis, no germline mutation in the previously-implicated genes APC, MUTYH, POLE, POLD1, or NTHL1 can be identified, although a hereditary etiology is likely. To uncover further genes with high-penetrance causative mutations, exome sequencing of leukocyte DNA from 102 unrelated individuals with unexplained adenomatous polyposis was performed. We identified two unrelated individuals with differing compound-heterozygous loss-of-function germline mutations in the mismatch repair gene MSH3. The impact of the MSH3 mutations (c.1148delA, c.2319-1g>a, c.2760delC, c.3001-2a>c) was indicated on RNA and protein level. Analysis of the diseased individuals’ tumor tissue demonstrated high microsatellite instability of di- and tetranucleotides (EMAST) and immunohistochemical staining illustrated a complete loss of nuclear MSH3 in normal and tumor tissue, confirming the loss-of-function effect and causal relevance of the mutations. The pedigrees, genotypes, and the frequency of MSH3 mutations in the general population are consistent with an autosomal recessive mode of inheritance. Both index persons had an affected sibling carrying the same mutations. The tumor spectrum in these four persons comprised colorectal and duodenal adenomas, colorectal cancer, gastric cancer, and an early-onset astrocytoma. Additionally, we detected one unrelated individual with biallelic PMS2 germline mutations, representing Constitutional Mismatch Repair Deficiency Syndrome (CMMRD). Potentially causative variants in 14 more candidate genes identified in 26 other individuals require further workup. In the present study we describe biallelic germline mutations of MSH3 in individuals with a suspected hereditary tumor syndrome. Our data suggest that MSH3 mutations represent an additional recessive subtype of colorectal adenomatous polyposis.

The majority of monogenic disorders cause craniofacial abnormalities with characteristic facial morphology. These disorders can be diagnosed more e ciently with the support of computer-aided nextgeneration phenotyping tools, such as DeepGestalt. These tools have learned to associate facial phenotypes with the underlying syndrome through training on thousands of patient photographs. However, this "supervised" approach means that diagnoses are only possible if they were part of the training set. To improve recognition of ultra-rare diseases, we created GestaltMatcher, which uses a deep convolutional neural network based on the DeepGestalt framework. We used photographs of 21,836 patients with 1,362 rare disorders to de ne a "Clinical Face Phenotype Space". Distance between cases in the phenotype space de nes syndromic similarity, allowing test patients to be matched to a molecular diagnosis even when the disorder was not included in the training set. Similarities among patients with previously unknown disease genes can also be detected. Therefore, in concert with mutation data, GestaltMatcher could accelerate the clinical diagnosis of patients with ultra-rare disorders and facial dysmorphism.

Colorectal cancer (CRC) arising in Lynch syndrome (LS) comprises tumours with constitutional mutations in DNA mismatch repair genes. There is still a lack of whole-genome and transcriptome studies of LS-CRC to address questions about similarities and differences in mutation and gene expression characteristics between LS-CRC and sporadic CRC, about the molecular heterogeneity of LS-CRC, and about specific mechanisms of LS-CRC genesis linked to dysfunctional mismatch repair in LS colonic mucosa and the possible role of immune editing. Here, we provide a first molecular characterization of LS tumours and of matched tumour-distant reference colonic mucosa based on whole-genome DNA-sequencing and RNA-sequencing analyses. Our data support two subgroups of LS-CRCs, G1 and G2, whereby G1 tumours show a higher number of somatic mutations, a higher amount of microsatellite slippage, and a different mutation spectrum. The gene expression phenotypes support this difference. Reference mucosa of G1 shows a strong immune response associated with the expression of HLA and immune checkpoint genes and the invasion of CD4+ T cells. Such an immune response is not observed in LS tumours, G2 reference and normal (non-Lynch) mucosa, and sporadic CRC. We hypothesize that G1 tumours are edited for escape from a highly immunogenic microenvironment via loss of HLA presentation and T-cell exhaustion. In contrast, G2 tumours seem to develop in a less immunogenic microenvironment where tumour-promoting inflammation parallels tumourigenesis. Larger studies on non-neoplastic mucosa tissue of mutation carriers are required to better understand the early phases of emerging tumours.