A Comprehensive Follow-Up Study Identifying Diagnostic Evidences for Secondary Finding

Background (cid:0) There was a lacking of clinical diagnostic evidence in follow-up studies for reporting of secondary variants in 59 genes in American College of Medical Genetics and Genomics recommendations for reporting secondary ndings and various strategies were applied to interpret the secondary variants. Results: Out of 1330 participants performed whole-exome sequencing, we identied 15 families with convincing clinical evidence. After Sanger validation and a comprehensive clinical follow-up, 10 families with both convincing clinical evidence and convincing genetic evidence of hereditary variants were found. Detailed clinical presentations and related clinical evidence were collected. Conclusions: Our research is a comprehensive follow-up study to identify secondary variants with convincing genetic and clinical evidence and it could help improve the strategy of screening actionable secondary variants and contribute to translation of genetic ndings into medical practice. DM/LP, c.3133G>A, BRCA1 c.427G>A) DM?/VUS, and 1 variant c.35_39del) classied as LP. The gene-associated genotype were cardiogenic diseases (TNNT2-related dilated cardiomyopathy, MYBPC3-related hypertrophic cardiomyopathy); connective tissue diseases (COL3A1-related Ehlers-Danlos syndrome-vascular type); hypercholesterolemia (APOB-familial hypercholesterolemia); and oncogenic or BRCA2-related hereditary breast and ovarian cancer, MSH2-related Lynch syndrome). signicance of secondary variants 22–25 . In the few previous reports on relevant medical ndings in individuals carrying secondary variants in 59 ACMG genes, clinical presentations of familial hypercholesterolemia reected by TC and LDL-C levels were the most easily accessible clinical evidence reports 6,8,26 . As for the two most common disease categories of secondary ndings, oncogenic diseases and cardiogenic disease, carriers of variants in the cancer-susceptible genes BRCA1/2 and MLH1 have been reported to have a family history of cancer in previous cohort study on secondary ndings 7 . Individuals carrying variants in genes such as MYBPC3 and KCNQ1 have also been reported to be related to cardiogenetic disease in researches about secondary ndings 7,26 . However, cardiovascular symptoms are very common and due to the lack of specicity of symptoms and diculty in obtaining clinical evidence, it is impossible to determine whether cardiovascular symptoms are related to cardiac diseases such as cardiomyopathy and arrhythmia 6 . In our follow-up study, the secondary variant-associated disease categories with clinical diagnostic evidence included oncogenic diseases (BRCA1 variants c.811G > A, c.427G > A and BRCA2 c.1832C > G-related breast cancer, MSH2 c.2197G > A-related Lynch syndrome) in 4 families, hypercholesterolemia (APOB c.1342G > A, c.35_39del-related familial hypercholesterolemia), cardiogenic diseases (TNNT2 c.422G > A-related dilated cardiomyopathy, MYBPC3 c.2543C > T-related hypertrophic cardiomyopathy) and connective tissue diseases (COL3A1 c.3133G > A-related Ehlers-Danlos syndrome-vascular type) in 2 families separately. Clinical and genetic evidence was collected to conrm the disease diagnosis and cosegregation of the variants. In addition to reporting of three most common disease categories- oncogenic diseases, hypercholesterolemia and cardiogenic diseases, our research rst provided clinical diagnostic evidence for variants related to rare connective disease Ehlers-Danlos syndrome-vascular type in secondary variant research. In family

DNA fragments were hybridized and captured by IDT's xGen Exome Research Panel (Integrated DNA Technologies, San Diego, USA) according to the manufacturer's protocol. The hybrid products were eluted and collected. Then, DNA was PCR ampli ed and puri ed.
The libraries were tested for enrichment by qPCR, and size distribution and concentration were determined using an Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA).WES was performed on the NovaSeq6000 platform (Illumina, San Diego, USA), with 150 bp pair-end reads used for sequencing the genomic DNA of the family. Raw image les were processed using CASAVA v1.82 for base calling and generating raw data.
After the annotation process, variants within the 59 genes recommended by the ACMG were extracted.

Variant ltering
Our goal was to screen out family members of participants with hereditary secondary variants and participants with de novo secondary variants who developed secondary variant related diseases to clarify the clinical signi cance of the secondary variants; therefore, sensitivity was sacri ced.
Step 1. Filtering variants according to HGMD and ACMG guidelines Disease-causing mutations (DM) and probable/possible pathological mutation (DM?) in the HGMD (Professional version 2019.1) database, and pathogenic (P) and likely pathogenic (LP) variants interpreted by ACMG guidelines for interpretation of genetic variants were included 12 .
Step 2. Filtering variants according to allele frequency, variant type and inheritance mode Variants with minor allele frequencies (MAF) <0.1%, variant depth of coverage >=20 and alteration base depth of coverage >=4 were taken on for further analyses. Filtering was conducted on the remaining variants according to variant type and inheritance model of the associated disease of ACMG 59 medically actionable genes 2 . Variants in the exonic region or with splicing impact were included. Among exonic variants, loss-of-function variants (frameshift variants, start lost, stopgain, etc.) and nonsynonymous single nucleotide variants (SNVs) were retained, and synonymous SNVs were excluded. Heterozygous variants in two genes, ATP7B and MUTYH, in autosomal recessive inheritance were excluded, and homozygous variants in the genes remained. Heterozygous variants in the other 57 genes that t the dominant inheritance mode were retained. Pass-lter variants were selected for clinical and genetic follow-up studies (Supplementary Table 1

-4).
Clinical and genetic follow-up study based on families A comprehensive follow-up study was conducted. Family members of participants with hereditary secondary variants and participants with de novo secondary variants were enrolled. Primary medical evaluation: Current and past medical history was assessed by email, WeChat software or telephone to identify con rmed individuals according to diagnostic criteria or possible individuals tting at least one clinical manifestation in OMIM database (https://omim.org/). Secondary medical evaluation: physical examination and necessary examinations by face-to-face evaluation in the clinic or home visits were also used to identify individuals with convincing clinical evidence. All clinical information was collected and evaluated by experienced clinicians, including pediatricians and neurosurgeons. The diagnostic criteria of the gene-associated diseases are listed in the detailed clinical manifestations of each family separately.
Individuals and families with convincing clinical evidence were performed Sanger sequencing was performed to validate the variant. Experienced geneticists analyzed the variants, including variant types and inheritance modes. All family members who took the test signed the informed consent form.
Published articles in the HGMD database related to secondary variants with convincing clinical and genetic evidence were reviewed to explore the pathogenicity of the variants.

Results
The owchart showing research steps including variant screening and follow-up study is presented in Figure 2. A total of 8851 secondary variant counts classi ed as DM or DM? in the HGMD database or P or LP in the ACMG guidelines in 1330 participants were obtained after variant ltering step 1 Figure  3b).

Follow-up study
A comprehensive follow-up study was performed in 638 individuals and their family members. Among these participants, 105 people were lost to follow-up (38 people could not be contacted, 67 people refused to participate in the study), and 532 people nally took part in the follow-up study including family members of 420 participants with hereditary secondary variants and 112 participants with de novo secondary variants. The detailed screening strategies to identify individuals with convincing clinical evidence in the follow-up study is shown in Figure 5. Exampli ed by follow-up study of participants carrying MPBPC3 variants, strategies including primary medical evaluation and secondary medical evaluation were performed to identify individuals with convincing clinical evidence ( Figure 5). Based on the clinical features of different genes associated with secondary ndings, 13 families with su cient clinical evidence were found in the follow-up study while no participants with de novo variants presented convincing clinical manifestation (Table 1).
Sanger sequencing was conducted to validate variants in families with convincing clinical evidence. There was 1 family who refused Sanger sequencing (Table 2). After Sanger sequencing of the 12 families, 10 families with both convincing clinical and genetic evidence were detected (Table 3). 2 variants did not cosegregate with disease phenotypes in 2 families (Table 2). Among these nal identi ed individuals having clinically and genetically con rmed secondary variants, 8 individuals were from the WES 1330 cohort. Therefore, the frequency of the secondary variants with both convincing clinical and genetic evidence in our cohort was 0.6% (8/1330). Detailed clinical and genetic evidence of the 9 secondary variants in 10 families are shown as follows along with penetrance calculation of variants in families based on age at onset of gene-associated diseases.
Family 1 TNNT2 c.422G>A was identi ed in the mother of the proband with microcephaly (III:2) by trio-WES in family 1. The heterozygous variant was validated by Sanger sequencing in 5 family members who agreed to do the test. Among them, the variant was detected in 4 family members (II:1, II:3, III:2, III:3) and was not found in 1 family member (IV:1) ( Figure 6-7). The diagnosis of dilated cardiomyopathy was made by rst-line imaging test echocardiography 13 . The sister of the grandmother of the proband (II:1) was previously diagnosed with dilated cardiomyopathy by echocardiography (Figure 8a). The grandmother of the proband (II:3) often suffered from chest tightness and dizziness, especially after exercise. Echocardiography showed dilated cardiomyopathy (Figure 8b). The mother of the proband (III:2) had occasional chest distress and palpitation, but she did not seek medical advice from the doctors. After we reported the variant to her, she came to the hospital and underwent echocardiography examination. Dilated cardiomyopathy was presented as a result (Figure 8c) ( Table 4). As a gene related to child/adult-onset disease, the penetrance of TNNT2 c.422G>A was 3/4 in family 1.

Family 2
Trio-WES was performed, and APOB c.1342G>A was observed in the mother and the proband who presented with developmental and epileptic encephalopathy (III:3, IV:1). Sanger sequencing revealed the heterozygous variant in 6 family members (II:1, II:2, II:6, III:1, III:3, IV:1) who wanted to undergo the test (Figure 9-10). The diagnosis was made according to the Dutch Criteria for Familial Hypercholesterolemia 14 . Two family members (I:2, II:1) died of cerebral infarction accompanied by hyperlipidemia at the ages of 71 and 69. The grandfather of the proband had hyperlipidemia with serum total cholesterol (TC) 7.68 mmol/L and low-density lipoprotein cholesterol (LDL-C) 4.97 mmol/L. The younger brother of the grandfather of the proband (II:6) had facial paralysis after cerebral infarction at the age of 50. The TC was 9.22 mmol/L, and LDL-C was 5.04 mmol/L in the recent blood lipid test. The older brother of the mother of the proband (III:1) had hyperlipidemia (TC 7.91 mmol/L, LDL-C 5.17 mmol/L) and had fatty liver detected by abdominal ultrasound examination. The mother of the proband (III:3) also had hyperlipidemia with TC 8.18 mmol/L and LDL-C 5.46 mmol/L ( Table 4). As the age at onset of APOB-related familial hyperlipidemia was child/adult, the penetrance of APOB c.1342G>A in family 2 was 5/6. Family 3 APOB c.35_39del was identi ed in the mother of the proband with microcephaly (III:2) after trio-WES. The heterozygous variant was validated in four family members by Sanger sequencing. Among them, three family members carried the variant (II:2, II:3, III:2), and 1 family member (IV:1) did not ( Figure 11-12).
Familial hypercholesterolemia was diagnosed according to Dutch criteria 14 . The father of the grandmother of the proband (I:1) was diagnosed with coronary heart disease and hypercholesterolemia 20 years ago. The grandmother (II:2), the brother of grandmother (II:3) and the mother of the proband (III:2) all had hypercholesterolemia with high TC and LDL-C (TC 8.05 mmol/L LDL-C 5.39 mmol/L; TC 7.72 mmol/L LDL-C 4.98 mmol/L; TC 7.99 mmol/L LDL-C 5.51 mmol/L) ( Table 4). According to the age at onset of APOB-related familial hyperlipidemia (child/adult), the penetrance of APOB c.35_39del was 3/3 in family 3.  Table 4). The penetrance of adult-onset MSH2-associated Lynch syndrome was 2/3.

Family 5
After WES sequencing of trios, the proband with microcephaly (IV:3) and mother (III:5) of family 1 carried the secondary heterozygous variant COL3A1 c.3133G>A. Sanger sequencing was applied to the family members who agreed to perform the test. Five people carried the variant (III:2, III:3, III:5, IV:2, IV:3), and 2 people (III:1, IV:1) did not ( Figure 16-17). The diagnosis for Ehlers-Danlos syndrome-vascular type was made by clinical features and COL3A1 mutations 16,17 . The brother of the proband's grandfather (II:1) reported a history of inguinal hernia. The grandfather of the proband (II:2) had an acute onset severe headache at the age of 47, and subarachnoid hemorrhage was detected by brain tomography angiography (CT) ( Figure 18). Brain computed tomography angiography (CTA) was performed, and an aneurysm of the internal carotid system was identi ed (Figure 19a). Surgical interventions were conducted. The younger sister of the mother (III:3) had headache of unknown reason at the age of 28. Due to the cerebral aneurysm hemorrhage of her father, brain CTA was performed, and an internal carotid system aneurysm was found ( Figure 19b). After being informed that he carried the same variant as his family members, the asymptomatic brother of the mother (III:2) was suggested to undergo brain CTA, and an aneurysm of the internal carotid system was also identi ed ( Figure 19c). The mother and sister of the proband showed dermatopathological features of Ehlers-Danlos syndrome-vascular type. The mother of the proband (III:5) had a brown-black plaque above the left ankle manifesting atrophic skin and ulcers ( Figure 20a Table 4). The penetrance was 4/5 based on child/adult age at onset of COL3A1-related Ehlers-Danlos syndrome-vascular type.

Family 6
The WES of trios in family 2 identi ed the same heterozygous variant COL3A1 c.3133G>A in the proband with developmental delay (IV:1) and her mother (III:2). Sanger sequencing validated the variant in 4 family members. The diagnosis of Ehlers-Danlos syndrome-vascular type was made as stated in family 5. Three family members (II:5, III:2, IV:1) carried the variant, and 1 family member (III:3) did not ( Figure 21-22). The younger brother of the grandfather of the proband died of acute myocardial infarction at the age of 22 (II:4). The grandfather (II:5) and mother (III:2) of the proband had a history of gastric perforation (Table 4). The penetrance was 2/3 based on child/adult age at onset of COL3A1-related Ehlers-Danlos syndrome-vascular type.

Family 7
Trio-WES results showed that the father of the proband with microcephaly (IV:1) carried the heterozygous MYBPC3 c.2543C>T in family 7. Three family members of family 7 underwent Sanger sequencing. Two family members were veri ed to carry the variant (III:1, IV:1), and 1 family member (V:1) was not ( Figure 23-24). The European Society of Cardiology (ESC) on the diagnosis and management of hypertrophic cardiomyopathy was used as a diagnostic criterion 18 . Four family members died suddenly. Among them, two family members (I:2, II:3) had sudden death before 40 with unknown cause. Two family members (II:1 and III:3) died of cardiac arrest at the ages of 45 and 38, respectively. The grandfather of the proband (III:1) was admitted to the hospital due to shortness of breath and chest tightness, and echocardiography detected features of hypertrophic cardiomyopathy (Figure 25a). The asymptomatic father of the proband (IV:1) was advised to undergo echocardiography after she was informed that she was a variant carrier. Echocardiography also showed evidence of hypertrophic cardiomyopathy (Figure 25b) ( Table 4). The penetrance of MYBPC3 c.2543C>T in family 7 was 2/2 (child/adult onset).  Table  4). The penetrance of the variant was 2/4 in total and 2/3 in females in family 8 according to adult-onset BRCA2-associated breast cancer.
Family 9 BRCA1 c.811G>A was detected in the proband with microcephaly by WES. Sanger sequencing was performed on 4 family members who agreed to the test. Three family members (I:2, II:1, II:2) were carriers of the variant, and 1 family member (II:4) was a noncarrier (Figure 29-30). Breast cancers were diagnosed histologically in two family members (I:2, II:2) (Figure 31a-b) ( Table 4). Due to the adult onset of BRCA1-related breast cancer, the penetrance was 2/3 in total and 2/3 in females in family 9.
Family 10 BRCA1 c.427G>A was identi ed in the mother (III:3) and the proband who presented with developmental and epileptic encephalopathy (IV:1) by trio-WES. Sanger sequencing validated the variant in four family members. Among them, 3 family members (II:2, III:1, III:3) carried the variant, and 1 family member (II:3) did not (Figure 32-33). Pathological ndings proved that the three family members had breast cancer in the right breast (Figure 34a-b) ( Table 4). The penetrance of the variant was 3/3 according to adult-onset BRCA1-related breast cancer.

Discussion
Our research is a comprehensive clinical and genetic follow-up study of individuals carrying secondary variants. A total of 9 variants in 10 families with actual clinical and genetic evidence were identi ed. These secondary variants cosegregated with previously reported breast or ovarian cancer, colon cancer, cardiomyopathy, and hypercholesterolemia, and we also rst reported cosegregation of COL4A1 variant and Ehlers-Danlos syndrome-vascular type in the secondary nding cohort. Additionally, we used these secondary variants shown by actual evidence as an indicator of developing screening strategies for secondary variants. We found that P/LP variants classi ed by ACMG guidelines could be reportable secondary variants, and DM/DM? variants in the HGMD database, even with VUS classi cations, could be of clinical signi cance. In addition, we calculated the penetrance of these actual secondary variants based on families. Our results further con rmed the clinical signi cance and necessity of interpretation of secondary ndings based on 59 genes recommended by ACMG SF v2.0 3 .
In cohort studies on secondary ndings, the frequencies of identi ed variants in secondary genes varied among different studies 6,7,19 . In our research, 10 families with convincing clinical and genetic evidence were detected, and the frequency of the secondary variants with actual evidence in our cohort was 0.6%, which was in accordance with ACMG's estimation of a 1% rate of reportable incidental ndings 7 . The most frequent secondary variants in genes after bioinformatic analyses were related to hereditary breast/ovarian cancer, familial hypercholesterolemia, arrhythmogenic right ventricular cardiomyopathy, hypertrophic cardiomyopathy, and catecholaminergic polymorphic ventricular tachycardia. The frequency of secondary variants in disease categories from the maximum to the minimum was ontogenetic disease, cardiogenetic diseases, hypercholesterolemia, connective tissue diseases and others. The disease categories were in accordance with ACMG recommendations for secondary ndings that the disorders of 59 genes were primarily cancer-predisposition syndromes and inherited cardiovascular disorders such as hypertrophic cardiomyopathy 20 .
In our research, we provided actual clinical and genetic evidence of the secondary variants that were reliable information on variant segregation. So far, there was limited actual clinical evidence that met the gold standard of secondary ndings-related diseases or ful lled the diagnostic criteria of the diseases used for identify secondary variants 19 . If the compatible symptoms for secondary ndings-related diseases were not speci c, it was di cult to verify the diseases with the subjective reports of participants without convincing evidence. Additionally, there were possibilities that the variants did not cosegregate with clinical phenotypes similar to the 2 families in our follow-up study ( Table 3). Calculation of reportable secondary variant frequency based on convincing clinical evidences combined with genetic evidence may provide a clue for identifying real secondary variants of clinical signi cance, which may bene t patients psychologically and help reduce the unnecessary costs of medical actions 21 .
There is a lack of clinical follow-up studies in cohort studies to explore the clinical signi cance of secondary variants [22][23][24][25] . In the few previous reports on relevant medical ndings in individuals carrying secondary variants in 59 ACMG genes, clinical presentations of familial hypercholesterolemia re ected by TC and LDL-C levels were the most easily accessible clinical evidence reports 6,8,26 . As for the two most common disease categories of secondary ndings, oncogenic diseases and cardiogenic disease, carriers of variants in the cancer-susceptible genes BRCA1/2 and MLH1 have been reported to have a family history of cancer in previous cohort study on secondary ndings 7 . Individuals carrying variants in genes such as MYBPC3 and KCNQ1 have also been reported to be related to cardiogenetic disease in researches about secondary ndings 7,26 . However, cardiovascular symptoms are very common and due to the lack of speci city of symptoms and di culty in obtaining clinical evidence, it is impossible to determine whether cardiovascular symptoms are related to cardiac diseases such as cardiomyopathy and arrhythmia 6 . In our follow-up study, the secondary variant-associated disease categories with clinical diagnostic evidence included oncogenic diseases (BRCA1 variants c.811G > A, c.427G > A and BRCA2 c.1832C > G-related breast cancer, MSH2 c.2197G > A-related Lynch syndrome) in 4 families, hypercholesterolemia (APOB c.1342G > A, c.35_39del-related familial hypercholesterolemia), cardiogenic diseases (TNNT2 c.422G > Arelated dilated cardiomyopathy, MYBPC3 c.2543C > T-related hypertrophic cardiomyopathy) and connective tissue diseases (COL3A1 c.3133G > A-related Ehlers-Danlos syndrome-vascular type) in 2 families separately. Clinical and genetic evidence was collected to con rm the disease diagnosis and cosegregation of the variants. In addition to reporting of three most common disease categories-oncogenic diseases, hypercholesterolemia and cardiogenic diseases, our research rst provided clinical diagnostic evidence for variants related to rare connective disease Ehlers-Danlos syndrome-vascular type in secondary variant research. In family members carrying COL3A1 c.3133G > A in two unrelated families, typical cutaneous features of the disease, such as brown-black plaque, pityriasis versicolor-like lesions and cerebrovascular complications aneurysm accompanied by hernias, and perforation of organs were observed in family members carrying the variant 27,28 .
Past studies have applied different criteria ltering procedures and classi cation of secondary variants. Interpreting secondary variants not related to primary disease requires a large amount of exploration work from researchers and clinicians and could be very challenging. Using ACMG guidelines alone or as the nal screening step, as most published articles did, could indeed identify the majority of secondary variants with actual evidence re ected by our results.
However, restricting the variant classi cations to P/LP in the guidelines might decrease the sensitivity of the identi cation of secondary variants such that the frequency of these variants was lower than 5% after variant ltrations 24,29 . Some studies relied on the HGMD database accompanied by manual review of primary reviews to de ne the pathogenicity of secondary variants 24 . The HGMD database is one of the most common medically curated databases used for determining the pathogenicity of variants. However, the variants in HGMD may be nonspeci c and oversensitive 8 . There were limited primary articles for variant curation, which might lead to inevitable misclassi cations of variants. For example, one of the primary articles included in the curation of MSH2 c.2197G > A in the HGMD database explored the role of the variant in medulloblastoma 30 . However, the disease is not well-established in Lynch syndromerelated tumors, such as cancers in the colon, rectum, and endometrium 15 . Therefore, our preliminary exploration suggestions on the screening strategy of secondary variants included variants according to ACMG guidelines (P/LP) or the HGMD database (DM/DM?) to elevate the detection rate of clinical diagnostic variants and thus re ne the screening strategies. Additionally, from the variants included in our research (Fig. 4), secondary variants classi ed as both P/LP and DM accounted for the majority of either P/LP or DM classi cations. However, there were a large number of secondary variants, especially in DM? classi cation (Fig. 3-4), and our research showed that this classi cation was also worthy of attention. We identi ed two DM?/VUS variants (COL3A1 c.3133G > A, BRCA1 c.427G > A) in three families with con rmed clinical and genetic evidence indicating that VUS variants could also be actual disease-related variants. Although the possibility of a given variant of solely classi cation of VUS as a disease-causing variant is extremely low 1 , the "clinically signi cant" VUS subgroup variants such as the two DM?/VUS variants should be considered in future research to develop strategies of identifying secondary variants.
However, future studies are needed to explore further ltration strategies for secondary variants classi ed as DM?/VUS.
The penetrance of secondary variants could range from 20-100% at different evaluation ages, and such variable variant penetrance may cause the genetic counseling of secondary variants to be very challenging 3 . We addressed penetrance of speci c variants in families with actual clinical and genetic evidence, and the penetrance calculation was carried out based on more than 3 generations in families that had a broader age range than routine cohort studies collected at a certain age stage. The family-speci c penetrance could decrease age-related bias and was more accurate in re ecting the actual variant penetrance, especially for variants in child/adult-onset genes. It has been reported that the evaluation age of a speci c age might lter a number of highpenetrance cancer variants for mortality at younger ages 7 . In our 10 families, the penetrance of secondary variants ranged from 67-100%, which was in accordance with the high-penetrance feature of secondary variants. Sexual penetrance differences of BRCA1-2 were also demonstrated in our research; female penetrance was higher than male penetrance.
The value of reporting secondary ndings was shown in family members who did not exhibit overt disease phenotypes. These secondary ndings may provide signi cant opportunities for asymptomatic individuals to prevent disease by surveillance for family members. It is crucial for the prediction and prevention of severe diseases by screening deleterious secondary variant carriers. For example, after learning about their grandfather's medical history of subarachnoid hemorrhage, two family members carrying COL3A1 c.3133G > A underwent CTA examination and were surgically treated by neurosurgeons. As signi cant symptoms do not usually appear before the rupture of aneurysms, the action has played a preventive role in this disease, which may be lethal and seriously affect the quality of life of patients. In the family of multiple sudden deaths and the grandfather with hypertrophic cardiomyopathy, the father of the proband was also suggested to undergo a relevant test after being reported as having the secondary variant MYBPC3 c.2543C > T and was also diagnosed with the disease. It is a life-changing action that leads to early detection of disease and early prevention of asymptomatic patients. Therefore, our research indicated that if a secondary variant with genetic and clinical evidence is identi ed in a patient, it is worth checking for the actual secondary variant in their asymptomatic family members to prevent future risk of actionable high-penetrance diseases by lifestyle changes or medical and/or surgical intervention.
Although our analysis is a comprehensive follow-up study of secondary ndings, the research had some inevitable limitations. First, the included populations were from the clinics of our hospital. A total of 1330 participants could not compare to the sample size of multicenter WES studies, which may increase the generalizability of the variant interpretations. Second, the people who were lost to follow-up or refused the test would lead to lower nal positive results of the follow-up study. Third, although our exploration of the ltration of actionable secondary variants may increase the sensitivity of the variants by incorporation of VUS/DM? variants, the speci city of the variants was compromised. Further bioinformatic studies based on clinical and genetic evidence need to be performed to seek appropriate procedures identifying reportable actionable secondary variants.

Conclusions
In conclusion, our comprehensive study provided a choice of re nement in the report of secondary ndings so that the secondary variants could be more likely    Figure 1 Pie chart of the frequency distribution of primary diseases in the population.    The detailed screening strategies to identify individuals with convincing clinical evidence in the follow-up study exempli ed by participants carrying MPBPC3 variants.

Figure 6
Pedigrees of family 1 with TNNT2 c.422G>A. Individuals with heterozygous variants and without the variant are indicated by +/-and -/-, respectively.

Figure 7
Sanger sequencing results of TNNT2 c.422G>A in family members.  Echocardiographic ndings in TNNT2 c. Sanger sequencing results of APOB c.1342G>A in family members.

Figure 11
Pedigrees of family 3 with APOB c.35_39del. Individuals with heterozygous variants and without the variant are indicated by +/-and -/-, respectively.

Figure 12
Sanger sequencing results of APOB c.35_39del in family members.

Figure 13
Pedigrees of family 4 with MSH2 c.2197G>A. Individuals with heterozygous variants and without the variant are indicated by +/-and -/-, respectively.

Figure 14
Sanger sequencing results of MSH2 c.2197G>A in family members.

Figure 15
Histologic features of MSH2 c.2197G>A family members. 15a. Histologic image showed middle to low differentiated adenocarcinoma, part of which was mucinous adenocarcinoma. The tumor invaded the deep myometrium (ulcerative mass, 3.5 × 2 × 1 cm in size); tumor thrombus and perineural invasion were seen in the vessel; no cancer metastasis was found in the lymph nodes (0 / 7). 15b. Histologic image showed low differentiated adenocarcinoma (tumor size 6 × 4 × 2 cm). The tumor invaded the whole layer, and cancer tissue could be seen in the abdominal wall. Mesenteric lymph nodes showed cancer metastasis (1 / 40); Peri-intestinal lymph nodes showed no cancer metastasis (0 / 8).  Sanger sequencing results of COL3A1 c.3133G>A in family members. Figure 19 19a-c. Brain CTA showed an aneurysm of the internal carotid system.  Sanger sequencing results of COL3A1 c.3133G>A in family members.

Figure 23
Pedigrees of family 7 with MYBPC3 c.2543C>T. Individuals with heterozygous variants and without the variant are indicated by +/-and -/-, respectively.

Figure 24
Sanger sequencing results of MYBPC3 c.2543C>T in family members.   Sanger sequencing results of BRCA2 c.1832C>G in family members.

Figure 28
Histologic features of BRCA2 c.1832C>G family members. 28a. Histologic image showed (left breast) invasive carcinoma, nonspecial type, grade II histology, no clear tumor thrombus in the vessel, no nerve invasion; (base, skin margin and nipple) no cancer; (left armpit) lymph node with cancer metastasis (5 / 6).
28b. Histologic image showed (left breast) grade I invasive ductal carcinoma. No cancer was found in the upper, lower, inner, outer and bottom cutting edges, and no cancer metastasis was found in the lymph nodes (0 / 14).

Figure 29
Pedigrees of family 9 with BRCA1 c.811G>A. Individuals with heterozygous variants and without the variant are indicated by +/-and -/-, respectively.  Histologic features of BRCA1 c.811G>A family members. 31a. Histologic image showed (left breast) invasive carcinoma, inclined to tubular carcinoma, grade I histology and (right breast) invasive cancer; the focus tends to be apocrine adenoid type, histology grade II. 31b. Histologic image showed (left breast) invasive carcinoma, nonspecial type, grade II histology, with more lymphocytic in ltration, 5 × 4 × 3 cm in size, no tumor thrombus in vessel, no de nite nerve invasion, (basal, skin margin and papilla) no cancer invasion, (left axillary) lymph node no cancer metastasis (0 / 18).

Figure 32
Pedigrees of family 10 with BRCA1 c.427G>A. Individuals with heterozygous variants and without the variant are indicated by +/-and -/-, respectively. Figure 33