As genetic and genomic testing is now the standard of care for identifying hereditary susceptibility to many conditions, the reporting of VUS has increased considerably. It is difficult to be prevented through optimized technical approaches. Since VUS cannot be acted upon clinically, this classification may delay or prohibit precise diagnosis and genetic counseling [10] and prevent patients from accessing gene-specific therapies and clinical trials, particularly in adult genetic disorders patients who often have atypical phenotype or polygenic burden which may confound a primary diagnosis [9]. However, the clinical effects of VUS for adult genetic disorders patients cannot well be predicted by lack of parental samples which may limit use of computational and predictive evidence and segregation evidence. But a definitive genetic diagnosis is more pivotal for adult genetic disorders patients who often underwent a long diagnostic odyssey, with multiple specialist evaluations and countless investigations without a satisfactory diagnostic outcome due to the effects of environmental exposure or the phenotypes of more common/complex non-Mendelian inherited diseases which may interfere with the phenotypes of genetic diseases [9]. To further explore the classification of VUS adult genetic disorders patients, we analyzed 458 Chinese adult patients WES data to reveal characteristics of qualitatively distinct lines of evidence used for VUS in different phenotype adult genetic disorders patients.

Totally 1502 phenotype-associated VUS were detected from 408 patients. The most common were missense variants (1260/83.89%), with computational and predictive evidence PP3, BP4 and functional evidence PP2 was used much more frequently. All evidence could be used for missense variants contained population data (PM2, PS4, BA1, BS1, BS2), computational and predictive data (PS1, PM5, PP3, BP4, BP1), functional data (PS3, PM1, PP2, BS3), segregation data (PP1, BS4), de novo data (PS2, PM6), allelic data (PM3, BP2), and other data (PP4, BP5). But for novel or previously uncharacterized missense variants, the available evidence for pathogenic classification was only population allele frequency and in silico prediction (e.g. REVEL/SIFT/PolyPhen/CADD/Z-score) [10], meanwhile PM1, PM5, PP4 could only be used in a tiny minority of missense variants. The PM1 evidence was related to missense variants located in a mutational hot spot and/or critical and well-established functional domain. These regions are typically enriched with pathogenic non-truncating variants and devoid of benign non-truncating variants (missense, inframe indels). While the classification of missense variants might greatly benefit from the PM1 evidence, but there are still analytical obstacles specific to hotspot regions, such as the lack of hotspot region databases or gold standard [22]. The PP4 evidence could be for conditions with a phenotype highly specific to a single gene. According to the guidelines, the patient should be with a highly specific phenotype, the testing need be extremely sensitive, the gene should be reported limited benign variants, and family history of the patient should be consistent with the mode of inheritance [23]. However, among the nine CSER laboratories assessing the ACMG/AMP standards, PP4 was the most inconsistently implemented evidence which was probably because determining the specificity of a phenotype was subjective [24]. On the other hand, the interpretation of most VUS missense variants did not only need silico prediction but also co-segregation analysis, functional assays, the comparison of allele frequency in case and control [25]. Especially when the patients had the typical genetic disease family history, and no pathogenic variants were found the expanded verification of the VUS missense variants in each patient and their parents or offline functional verification tests on the VUS missense variants should be conducted with genetic counseling and informed consent.

As most of P/LP variants were LOF variants, PVS1 was used much more frequently for P/LP variants than VUS. Meanwhile population evidence PM2_Supporting was found to be used most both for the variants classified as VUS and P/LP. It may result from the rapid increase in human genome sequencing data, more and more frequency data were accumulated in the public population databases, making the use of population evidence the most convenient [26]. In this study, most of the P/LP variants were rare and the VUS were phenotype-associated and potentially harmful rare variants, thus PM2_Supporting could be used for most of these variants.

In addition, within 15 pathogenic evidence PS1, PS2, PM1, PM6 and PP4 code were not used for VUS, PS1 code were not used for P/LP variants. Except for PM1 and PP4 mentioned above, PS2 and PM6 could only be used after conducted expanded verification of the variants in parents in the reported literature or this study [27], but expanded verification were often difficult to conduct by the adult probands often did not want to inform genetic testing result to family members and lack parental samples which resulted in PS2 and PM6 often could not be used in this study. Then the use of PS1 and PM1 evidence often relied on public database and in-house databases to check whether there was pathogenic variant lead to the same amino acid substitution with the variant to be classified [28], the lack of study on the adult genetic disorders may lead to data could be used for PS1 was less and finally leaded to the limit use of PS1. These results suggested that genetic counseling for the adult genetic disorders patients was more important, genetic study in these patients should be performed more widely and in-house data about adult patients genetic testing results could be released to the public database as much as possible, these may increase the utilization rate of PS1, PS2, PM1, PM6 and PP4 in VUS re-classification in adult genetic disorders patients.

There were 31 null variants (nonsense, frameshift, canonical ±1 or 2 splice sites) were classified as VUS. All these null variants had three features: (1) LOF (Loss-of-function) was not a known mechanism of these diseases; (2) Variants were at the extreme 3′ end of these genes; (3) Relatively common LOF variants. Most of these variants were probably the disease-causing variants of the patients, but only could be classified as VUS based on current evidence. Thus the functional studies for VUS null variants re-classification in adult genetic disorders patients could be conduct based on the above three aspects, it could provide powerful insight into the effect of a variant on protein function and have the capacity to reclassify variants [29]. Such as confirm LOF was a known mechanism of these diseases, or this variant would lead to nonsense-mediated mRNA decay (NMD) [30]. This result would expand the genetic spectrum and improve the diagnostic rate of adult genetic disorders.

In order to explore which adult genetic disorders patients were more likely to be reported phenotype associated VUS, we analyzed the number of VUS in patients by department and found that patients with nervous system disease would be reported the most phenotype-associated VUS with each patient carried 5.12 VUS averagely. It may result from: (1) clinical features of nervous system disease were diverse and under-recognized, particularly those of adult-onset, (2) The clinical symptoms of patients with nervous system disease were atypical and usually spanned multiple neurological and medical subspecialties (3) In contrast to other adult genetic diseases, the positive family history of patients with nervous system disease was more difficult to accurately obtain which resulted in a lack of family variants validation data [31]. Thus, it was challenging to identify disease-causing pathogenic variants in these patients for the reasons mentioned above but only many VUS would be showed in the report. Then through noted the general pattern we found that the lower the phenotypic specificity, the more reported VUS. These results suggested that in the pretest genetic counseling, the patient’s family history should be investigated once as much as feasible of the patient’s specific clinical information had been obtained. The patient’s distinctive phenotype should also be distinguished. This information would significantly help us in identifying candidate variants and reporting lower VUS. In addition, it also could assist clinical judgment regarding the necessity of genetic testing and reduce the financial pressure on patients.

In conclusion, by analyzed 458 Chinese adult patients WES data, totally 1502 phenotype-associated VUS were detected from 408 patients with the most common were missense variants (1260/83.89%). The PS1, PS2, PM1, PM6 and PP4 were not used for VUS pathogenic classification and the PP3, BP4, PP2 was used much more frequently for VUS than P/LP. There were also 31 null variants which were probably the disease-causing variants of the patients were classified as VUS. These results suggested that appropriate genetic counseling, reliable releasing of in-house data, expanded verification in patient family, co-segregation analysis, functional assays and allele frequency comparison between case and control were urgent need to gather more evidence to reclassify VUS. We also found adult patients with nervous system disease were reported the most phenotype-associated VUS and the lower the phenotypic specificity, the more reported VUS. This result emphasized the importance of pretest genetic counseling which should be more precise and specific to get more information for identifying candidate pathogenic variants and reporting less VUS.

Our result revealed the characteristics of the criteria according to ACMG/AMP guideline used for VUS pathogenic classification in adult genetic disorders patients for the first time. We recommend a rules-based process to evaluate the pathogenicity of VUS in adult genetic disorders patients to provide a strong basis for accurately evaluating the pathogenicity and clinical grade information of VUS and avoiding the repeated genetic testing on the same patients to reducing social and family burdens. Meanwhile, we further expand the genetic spectrum and improve the diagnostic rate of adult genetic disorders, finally improve the clinical application value of WES. In the current study, we showed results of preliminary statistical analyses, as the sample size increases, more data will be added to improve these results.