Case-Control Study for 23 Cancer Types With Functional Analysis of CHEK2: Risk Estimation and Clinical Recommendations in East Asia

Abstract

PURPOSE

CHEK2 is the frequently detected cancer-predisposing gene in female breast cancer. In addition, the association with the risks of other cancer types has been suggested, and clinical management has also been discussed. Although clinical relevance of germline variants differs across population, there is little evidence of the clinical relevance of CHEK2 germline variants in East Asia.

METHODS

Targeted sequencing and functional analyses of missense variants for the coding region of CHEK2 in 111,571 East Asian individuals were performed. Variants classified as pathogenic/likely pathogenic in ClinVar, predicted loss-of-function, or functionally impaired in functional analysis were defined as germline damaging variants (gDVs). We evaluated the association between CHEK2 gDVs and the risk of 23 cancer types. We also compared the clinical characteristics of carriers and noncarriers among patients with CHEK2-associated cancers.

RESULTS

We identified 77 gDVs including 36 functionally impaired missense variants. CHEK2 gDVs were significantly associated exclusively with prostate cancer (odds ratio [OR], 1.8 [95% CI, 1.2 to 2.6]; P = 1.7 × 10⁻³), in addition to female breast cancer (OR, 1.8 [95% CI, 1.3 to 2.6]; P = 1.2 × 10⁻³), among 23 cancer types. There were no differences in age at diagnosis, pathologic status, and prognosis between carriers and noncarriers. Besides, there was no association with the risk of cancer types with high incidence rates in East Asian countries.

CONCLUSION

CHEK2 gDVs were associated with female breast and prostate cancer risks in East Asia. The necessity of additional systematic clinical management for all CHEK2 gDV carriers should be carefully discussed, and standard cancer screening is recommended unless no other clinical features suggestive of cancer predisposition are noted in East Asia.

INTRODUCTION

With the progression of next-generation sequencing technology, a rapid expansion of multigene panel testing enables the identification of individuals with germline pathogenic variants (gPVs) of cancer-predisposition genes, leading to optimization of risk reduction and surveillance for unaffected individuals.¹ BRCA1/2 is the best example; after the establishment of association with the risk of many cancer types, the existence and identification of differences between the clinical characteristics of gPV carriers and noncarriers, such as age at diagnosis, pathology, and prognosis, has led to the development of appropriate clinical guidelines.² After gPVs in BRCA2, gPVs in CHEK2 are the very frequent alterations detected in multigene panel testing for patients with female breast cancer among the Breast Cancer Association Consortium.³ Among this study, the c.1100delC (p.Thr367fs) variant accounted for approximately 80% of the protein-truncating variants; however, its prevalence varied widely across population and affected the prevalence of gPVs in CHEK2.³ Almost all previous reports on CHEK2-associated cancer types have come from the European population²; given that excessive management on the basis of the evidence from a skewed subset of individuals is sometimes harmful, population-specific robust genetic evidence for clinical relevance should be required.^4,5

CONTEXT

• Key Objective

• Cancers associated with CHEK2 germline variants are still uncertain, and the appropriate clinical management for such carriers is unconclusive especially for East Asian population.

• Knowledge Generated

• Functional analysis categorized 25.9% (36/139) of CHEK2 missense variants as functionally impaired. This case-control study of 23 cancer types in 112,141 individuals found a significant association between female breast and prostate cancers and germline damaging variants (gDVs), which we defined as pathogenic or likely pathogenic variants in ClinVar, loss-of-function variants, or functionally impaired variants. In contrast with BRCA1/2 germline pathogenic variants, no differences were observed between the clinical characteristics of CHEK2 gDVs carriers and noncarriers.

• Relevance

• Carriers of CHEK2 gDVs should not uniformly be considered for additional clinical management rather than regular screening.

Although several associations with other cancers, such as prostate, colorectal, kidney, and thyroid cancers, have been suggested in the European population, even associations with cancer risks were inconsistent in the literature.^2,6,7 There are several possible reasons for this. One is the limited sample size or the use of a skewed subset limited to individuals with suspected hereditary diseases.^6,8 Among the others were lacking clinical interpretation of many missense CHEK2 variants and moderate penetrance of all gPVs in this gene, which hinder the understanding of their clinical relevance.^6,7 Large-scale unselected case-control analyses that include the relevance of missense gPVs using functional analysis⁹ could overcome these limitations.

To support pathogenicity interpretation for these variants, the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) guidelines are commonly used. However, they are generally considered appropriate for interpretation of variants in high-penetrance genes,^10,11 and gene-specific ACMG/AMP guidelines for CHEK2 are still under development. In addition, applying the ACMG/AMP guidelines across multiple cancer types presents additional challenges.

Moreover, the inconsistency of associated cancers has not led to the development of clinical management of CHEK2 gPV carriers, similar to that of BRCA1/2 gPV carriers, even in the European population.^2,6 Besides, the landscape of related cancer type may differ because of the differences in cancer incidence,¹² indicating the importance of region-specific evaluation of impact of gPVs.

However, there is little evidence of the clinical relevance of CHEK2 gPVs in East Asia. Here, we performed a large-scale unselected case-control analysis of 23 cancer types among the East Asian population, including the relevance of missense variants with functional analysis.⁹

METHODS

Study Population

Individuals included in the study were registered in Biobank Japan (BBJ), a multicentered hospital-based registry across Japan originally developed for human genetic research, with diagnoses made by physicians. The selection was not on specific criteria such as suspicion of hereditary diseases.^13-15 DNA samples were obtained from BBJ and were collected between April 2003 and March 2018.¹³ This study included all available 73,853 individuals with 23 different cancer types. Some of these patients were included in our previous studies.^16-22 The study also included 38,288 controls with no history of malignant disease who were frequency-matched for age, sex, and area of hospital. All the participants provided written informed consent. The study was approved by the ethics committees of the Institute of Medical Sciences, the University of Tokyo, and the RIKEN Center for Integrative Medical Sciences.

Sequencing and Bioinformatics Analyses

We analyzed the complete coding region and the 2-bp flanking intronic sequences (total: 1,761 bp) of all CHEK2 transcripts registered in the Consensus CDS, release 15, using a multiplex polymerase chain reaction–based targeted sequencing method.²³ The pooled DNA libraries were sequenced using 2 × 150-bp paired-end reads on a HiSeq2500 or NovaSeq6000 (Illumina Inc, San Diego, CA), and the genetic variants were identified using Genome Analysis Toolkit (GATK, version 3.7-0; Broad Institute, Cambridge, MA).²⁴

Variant Interpretation and Functional Analysis

As the ACMG/AMP guidelines were not universally applicable to moderate-penetrance genes including CHEK2 and the gene-specific criteria remained under development, we defined the germline damaging variants (gDVs) on the basis of three types of evaluations. Variants that were classified as pathogenic or likely pathogenic according to ClinVar, version 2023-12-09,²⁵ or had HIGH impact on protein function in SnpEff ver4.3t²⁶ were considered damaging in this study. Additionally, missense variants not categorized as damaging by these evaluations were annotated using functional analysis with two established kinase assays, KAP1-pS473 and CHK2-pS516, as previously described.⁹ We excluded variants affecting the first or last two coding nucleotides of an exon, which may result in aberrant splicing.⁹ Variants showing impaired function in both kinase assays were considered damaging. CHK2 isoform localization was also evaluated using the nuclear-to-cytoplasm ratio: variants causing mislocalization outside the nucleus (low nuclear-to-cytoplasm ratio) were defined as functionally impaired and considered damaging.

Somatic Analysis

Considering that CHEK2 plays a role in cell cycle control in response to double-stranded DNA damage, we investigated the influence of gDVs on the patterns of somatic mutations and homologous recombination deficiency (HRD). We predicted the HRD status using the R package Classifier of HOmologous Recombination v2.0, as described previously.^16,27 Since BBJ does not have data on somatic analysis, we used data from the Pan-Cancer Analysis of Whole Genomes (PCAWG) Consortium of the International Cancer Genome Consortium and The Cancer Genome Atlas²⁸ that are available for download.²⁹

Statistical Analyses

First, we estimated the odds ratios (ORs) and corresponding 95% CIs to assess the association between gDVs and each cancer type. Logistic regression models were applied under the dominant model, with adjustments for age at registration, sex, and hospital area recorded in the BBJ. Breast cancer was separately analyzed in females and males. In addition, we evaluated the combined impact of CHEK2 gDVs and family history of the same cancer. Second, we investigated the clinical characteristics of CHEK2 gDV carriers and noncarriers diagnosed with CHEK2 gDV-associated cancers. We used the Mann-Whitney U test for continuous variables and Fisher's exact or chi-square tests for categorical variables.

We estimated the probability of overall survival using Kaplan-Meier curves and compared carriers with noncarriers using the log-rank test. To estimate hazard ratios (HRs), we applied Cox proportional-hazards models for multivariable analysis, adjusted for age at diagnosis and disease stage. Overall survival was defined as the period from the date of diagnosis to all-cause mortality or the day of the last follow-up.

All statistical tests were two-sided, and statistical significance was set at P < .05. The Bonferroni correction was applied when appropriate. All statistical analyses were performed using the R software (version 4.2.1; R Foundation for Statistical Computing, Vienna, Austria) and Stata (version 18.0; StataCorp, College Station, TX). See Supplementary Methods for detailed methodology.

RESULTS

Study Population

The study included 73,853 patients (mean [standard deviation] age at registration, 66.8 [11.3] years; 31,644 [42.8%] female) and 38,288 controls (60.9 [12.3] years; 16,786 [43.8%] female; Table 1). The number of registered patients was the highest for colorectal cancer (n = 16,481), followed by gastric (n = 12,931), prostate (n = 12,240), and breast cancers (n = 11,691). Among the total, 11.7% of cases carried a family history of the same cancer type.

TABLE 1.

Characteristics of the Participants

Cancer Type	No. of Samples			Age at Registration, Years, Mean (SD)	No. of Family History of the Same Cancer Type (%)a
	Female	Male	Total
Case
Biliary tract	373	659	1,032	70.1 (9.2)	28 (2.7)
Bladder	158	972	1,130	73.0 (8.4)	41 (3.6)
Bone	36	43	79	65.1 (12.0)	NA
Brain	104	97	201	65.5 (13.5)	4 (2.0)
Breast	11,601	90	11,691	60.9 (12.3)	1,177 (10.1)
Cervical	2,547	NA	2,547	56.1 (14.0)	71 (2.8)
Colorectal	6,105	10,376	16,481	68.4 (10.4)	2,279 (13.8)
Endometrial	2,286	NA	2,286	60.5 (11.3)	50 (2.2)
Esophageal	362	2,110	2,472	67.6 (8.5)	140 (5.7)
Gastric	3,324	9,607	12,931	69.0 (10.0)	3,290 (25.4)
Head neck	127	770	897	69.9 (9.6)	25 (2.8)
Liver	1,111	3,200	4,311	68.9 (9.1)	414 (9.6)
Lung	2,855	5,371	8,226	68.7 (9.4)	1,118 (13.6)
Lymphoma	892	1,158	2,050	64.0 (13.5)	47 (2.3)
Ovarian	1,718	NA	1,718	57.9 (12.3)	57 (3.3)
Pancreatic	431	709	1,140	68.0 (9.8)	83 (7.3)
Prostate	NA	12,240	12,240	72.6 (7.4)	973 (7.9)
Renal	349	1,066	1,415	67.2 (10.6)	31 (2.2)
Sarcoma	18	23	41	62.5 (16.5)	NA
Skin	127	204	331	71.7 (9.9)	4 (1.2)
Testis	NA	79	79	58.4 (13.6)	NA
Thyroid	532	202	734	65.4 (11.8)	17 (2.3)
Ureteral	17	40	57	72.8 (8.2)	NA
Total
Patients	31,644	42,209	73,853	66.8 (11.3)	NA
Casesb	35,073	49,016	84,089	66.7 (11.5)	9,849 (11.7)
Control	16,786	21,502	38,288	60.9 (12.3)	NA

Abbreviations: NA, not applicable; SD, standard deviation.

^a Patients with first-degree relatives who had the same cancer.

^b The numbers of patients and cases are described separately as 8,959 patients (11.9%) had more than one type of cancer.

Variant Interpretation and Functional Analysis

After quality control, of 112,141 participants, 111,571 (73,308 patients and 38,263 controls) were included for further analysis. Among the 244 CHEK2 germline variants (Data Supplement, Table S1), 41 were categorized as gDVs before functional analysis, including 33 pathogenic/likely pathogenic variants classified in ClinVar and eight protein-truncating variants (Fig 1). Additionally, 36 of the 139 (25.9%) missense variants were categorized as damaging by functional analysis, including 34 functionally impaired variants (Fig 2A) and two variants with impaired localization to the nucleus (Fig 2B). We validated clinical utility of functional analysis by assessing the comparable OR in female breast cancer between patients with functionally impaired missense variants (OR, 1.7 [95% CI, 1.1 to 2.7]) and those with damaging variants as classified according to ClinVar or truncating variants (OR, 2.0 [95% CI, 1.1 to 3.7]; Data Supplement, Table S2). We further compared our classifications with classifications on the basis of the ACMG/AMP guidelines^30,31 to investigate their applicability. Using female breast cancer as a representative phenotype, eight variants classified as damaging by our classifications were variants of uncertain significance (VUS) under the ACMG/AMP guidelines (Data Supplement, Table S3). To evaluate our assessment in favor of pathogenicity for these VUS, we performed the association analysis showing that their carriers had significantly increased breast cancer risk (OR, 1.6 [95% CI, 1.0 to 2.4]), which was comparable with that in carriers of ClinVar-classified pathogenic variants (OR, 1.8 [95% CI, 1.0 to 3.5]). We did not apply the ACMG/AMP guidelines systematically in this study. Consequently, 77 variants were classified into gDVs. Among the gDVs, p.Arg521Trp, a missense variant with impaired nuclear localization, was the most frequently observed, accounting for almost half of the gDVs carriers among patients, followed by the nonsense variant p.Arg519Ter, classified as pathogenic/likely pathogenic in ClinVar and accounting for an additional 11% (Fig 3).

FIG 1.

Schematic representation of CHEK2 variant classification. ^aWe excluded variants that affected the first/last two coding nucleotides in an exon, which we cannot exclude that their functional consequence may result in aberrant splicing, following the previous report.^{5 b}The following were defined as damaging variants: pathogenic or likely pathogenic reported by ClinVar, protein truncating, or functionally impaired variants determined by functional analysis.⁹ Conflicting, conflicting interpretations of pathogenicity or risk factors; VUS, variant of uncertain significance.

FIG 2.

Functional analysis using KAP1 and CHK2 kinase assays (A) A total of 115 variants showed concordant results between the two kinase assays, while 24 variants including two variants that do not localize to the nucleus showed discordant results. Two variants failed the functional analysis and are not shown in the figure. (B) In the DAPI image, the blue signals indicate the location of the nucleus inside the cell. In the EGFP image, green signals indicate the localization of the expressed CHK2 isoforms. The merge shows a merged image of DAPI and EGFP. In the KAP1-pS473KAP1 p.S473 image, red signals indicate autophosphorylation of CHK2-pS516. CHK2-WT: cells without CHEK2 germline variant as a negative control; CHK2 p.L204V: wild-type–like variant; CHK2 p.K249E: functionally impaired variant; p.R521W and p.R521Q: variants that do not localize to the nucleus. The detailed methods were described in the previous report.⁹

FIG 3.

Locations of amino acids coded by gDVs in patients. Locations of amino acids coded by the gDVs in CHEK2 among Biobank Japan patients with 23 cancers are shown by lollipop structures with the variant type indicated by color. Red indicates protein-truncating variant and yellow indicates missense variants categorized as functionally impaired by functional analysis. Conflicting, conflicting interpretations of pathogenicity or risk factors; FHA, forkhead-associated; gDVs, germline damaging variants; NLS, nuclear localization signal; SQ, serine-glutamine; TQ, threonine-glutamine; VUS, variant of uncertain significance.

Risk of Cancer Type Associated With CHEK2 gDVs

We performed association analyses between 23 cancer types and CHEK2 gDV carrier status. In addition to female breast cancer (OR, 1.8 [95% CI, 1.3 to 2.6], P = 1.2 × 10⁻³), we observed a significant association for prostate cancer (OR, 1.8 [95% CI, 1.2 to 2.6]; P = 1.7 × 10⁻³) after Bonferroni correction (P < 2.08 × 10⁻³ = 0.05/24; Table 2). The carrier frequencies among patients with breast and prostate cancers were 0.6% and 0.5%, respectively. Male breast cancer and sarcoma showed a nominal association (P < .05). No significant association was observed between CHEK2 gDVs and the other cancers, including colorectal, kidney, and thyroid cancers. A further exploratory analysis suggested that the risk of female breast or prostate cancer tended to be higher among CHEK2 gDV carriers with a family history of the same cancer compared with that in noncarriers without a family history of the same cancer (female breast cancer: OR, 3.8 [95% CI, 1.0 to 14.7], prostate cancer: OR, 14.0 [95% CI, 1.7 to 114.8]; Data Supplement, Table S4).

TABLE 2.

Association Between Damaging Variants of CHEK2 and Risk of 23 Cancer Types

Cancer Type	No. of Case		No. of Control		OR (95% CI)a	P b
	All	Carriers of Damaging Variants (%)	All	Carriers of Damaging Variants (%)
Biliary tract	1,018	3 (0.3)	38,263	115 (0.3)	1.0 (0.3 to 3.1)	.98
Bladder	1,105	5 (0.5)	38,263	115 (0.3)	1.6 (0.6 to 3.9)	.32
Bone	78	1 (1.3)	38,263	115 (0.3)	4.5 (0.6 to 32.6)	.14
Brain	197	1 (0.5)	38,263	115 (0.3)	1.7 (0.2 to 12.5)	.58
Breast (female)	11,530	68 (0.6)	16,780	54 (0.3)	1.8 (1.3 to 2.6)	1.2 × 10 − 3
Breast (male)	88	2 (2.3)	21,483	61 (0.3)	8.2 (2.0 to 34.1)	4.0 × 10 − 3
Cervical	2,527	8 (0.3)	16,780	54 (0.3)	1.0 (0.5 to 2.2)	.93
Colorectal	16,385	64 (0.4)	38,263	115 (0.3)	1.3 (1.0 to 1.8)	.10
Endometrial	2,261	8 (0.4)	16,780	54 (0.3)	1.1 (0.5 to 2.3)	.82
Esophageal	2,442	2 (0.1)	38,263	115 (0.3)	0.3 (0.1 to 1.2)	.08
Gastric	12,872	40 (0.3)	38,263	115 (0.3)	1.0 (0.7 to 1.5)	.91
Head neck	889	2 (0.2)	38,263	115 (0.3)	0.8 (0.2 to 3.2)	.74
Liver	4,278	10 (0.2)	38,263	115 (0.3)	0.8 (0.4 to 1.5)	.48
Lung	8,207	35 (0.4)	38,263	115 (0.3)	1.4 (1.0 to 2.1)	.07
Lymphoma	2,010	7 (0.3)	38,263	115 (0.3)	1.2 (0.5 to 2.5)	.70
Ovarian	1,701	8 (0.5)	16,780	54 (0.3)	1.4 (0.6 to 2.9)	.40
Pancreatic	1,136	2 (0.2)	38,263	115 (0.3)	0.6 (0.1 to 2.4)	.47
Prostate	12,161	61 (0.5)	21,483	61 (0.3)	1.8 (1.2 to 2.6)	1.7 × 10 − 3
Renal	1,385	8 (0.6)	38,263	115 (0.3)	2.0 (1.0 to 4.1)	.06
Sarcoma	41	1 (2.4)	38,263	115 (0.3)	8.6 (1.2 to 63.4)	.03
Skin	321	0 (0.0)	38,263	115 (0.3)	NA	NA
Testis	77	1 (1.3)	21,483	61 (0.3)	3.7 (0.5 to 28.6)	.21
Thyroid	721	4 (0.6)	38,263	115 (0.3)	1.8 (0.7 to 5.0)	.25
Ureteral	55	1 (1.8)	38,263	115 (0.3)	6.4 (0.9 to 46.7)	.07

Abbreviations: NA, not applicable; OR, odds ratio.

^a The associations between damaging variants of CHEK2 and the risk of each cancer were evaluated using a logistic regression model adjusted for age at registration, sex, and the area recorded in BioBank Japan.

^b On the basis of a Bonferroni-corrected threshold of significance, cancers with a P value <2.08 × 10⁻³ (ie, 0.05/24) were defined as CHEK2-associated cancers. P values are represented as follows: P < .05 is in bold, and P < 2.08 × 10⁻³ is in bold italics.

Comparison of Clinical Characteristics of Carriers and Noncarriers Diagnosed With CHEK2 gDV-Associated Cancers

To identify the demographic and clinical characteristics unique to CHEK2 gDV carriers and propose clinical recommendations, we compared the clinicopathologic characteristics between carriers and noncarriers with female breast or prostate cancers. Since breast and prostate cancers were associated with BRCA1/2 and BRCA2, respectively, we referred the data from our previous study on BRCA1 or/and BRCA2.¹² As shown in Figure 4, no significant differences were observed in age at diagnosis (CHEK2 gDV carriers v CHEK2 gDV noncarriers with breast cancer, 56.0 v 56.0 years, P = .99; prostate cancer, 70.0 v 71.0 years, P = .80) and tumor pathologic characteristics (proportion of triple-negative breast cancer [10.0% (4/40) v 11.4% (636/5,592), P = .79] and total Gleason score of 8-10 for prostate cancer [28.3% (13/46) v 31.2% (2,881/9,230), P = 0.67]), which was in contrast to our previous report on BRCA1 and BRCA2 carriers.¹² We also did not observe a statistically significant difference in proportions of estrogen receptor–positive breast cancer between CHEK2 gDV noncarriers and carriers (85.0% [34/40] v 78.3% [4,378/5,592], P = .31). In the survival analysis, 2,073 and 1,879 patients who were newly diagnosed with breast and prostate cancers, respectively, and were enrolled in BBJ within 6 months (≤184 days) of diagnosis were included with median follow-up of 10.1 years and 9.1 years, respectively. Overall survival did not significantly differ between CHEK2 gDV carriers and noncarriers with breast (log-rank P = .19, HR, 1.67 [95% CI, 0.62 to 4.47]; P = .31) or prostate (log-rank P = .92, HR, 1.04 [95% CI, 0.39 to 2.79], P = .93) cancers (Data Supplement, Figs S1A and S1B).

FIG 4.

Comparison of clinical characteristics between carriers and noncarriers of CHEK2 germline damaging variant carriers and carriers of BRCA1 and BRCA2 germline pathogenic variants. ^aThe data on carriers of BRCA1 and BRCA2 were cited from our previous report.¹² (A) Age at diagnosis of breast cancer. Data on age at diagnosis were unavailable for 16.4% (1,892/11,530) of patients with breast cancer (four patients with diagnosis age <20 years were treated as missing value because of the possibility of data entry error). Differences were evaluated using the Mann-Whitney U test. (B) Age at diagnosis of prostate cancer. Data on age at diagnosis were unavailable for 10.3% (1,250/12,161) of patients with prostate cancer. Differences were evaluated using the Mann-Whitney U test. (C) Proportion of triple-negative breast cancer. Information on ER/HER2 status in breast cancer was unavailable for 51.2% (5,898/11,530). Differences were evaluated using the chi-square test. (D) Proportion of 8-10 Gleason score in prostate cancer. Information on Gleason scores in prostate cancers was unavailable for 23.7% (2,885/12,161). Differences were evaluated using the chi-square test. ER, estrogen receptor; HER2, human epidermal growth factor receptor 2.

Somatic Patterns for Carriers of Germline Variants in CHEK2

Regarding somatic analysis, 94 unique PCAWG samples with 44 germline variants of CHEK2 remained after quality control. Of these, 15 samples carried CHEK2 gDVs, and the remaining 79 carried variants that were not categorized as damaging. The patterns of somatic mutations, including loss of heterozygosity, HRD, and mutational signatures of SBS3 and ID6,³² which are related to HRD, were not specific to the tumor tissues of CHEK2 gDV carriers (Data Supplement, Table S5), suggesting that CHEK2 gDVs are not associated with HRD in tumor tissues.

DISCUSSION

To our knowledge, this is the largest sequencing analysis among the East Asian population of CHEK2 variant. We did not observe the carriers of c.1100delC variant, which is common among the European population.³ However, using functional analysis, we identified a common damaging missense variant, almost half of the carriers among patients, p.Arg521Trp. Case-control analyses among unselected individuals demonstrated that CHEK2 gDVs were associated with prostate cancer, besides female breast cancer. By contrast, we did not observe an association with the risk of cancer types with higher incidence rates in East Asian countries than in Western countries, such as gastric cancer and esophageal cancer.³³ Also, no differences were observed in age at diagnosis, proportion of triple-negative breast cancer or prostate cancer with high Gleason scores, or overall survival when the clinicopathologic characteristics and prognosis of carriers and noncarriers of CHEK2 gDVs were compared.

We found that 26.4% of the CHEK2 missense VUS tested in the functional analysis were functionally impaired, with ORs comparable with those of the ClinVar-classified pathogenic variants or the protein-truncating variants. Consequently, the number of identified CHEK2 gDVs increased 1.9-fold, which is much higher than the number of gPVs identified by functional assays of BRCA1/2.³⁴ This suggests that the interpretation of VUS should vary depending on the gene, particularly between those with high and moderate penetrance. This would aid clinicians and carriers in interpreting the results of genetic testing. Importantly, the functional categorization of CHEK2 missense variants improved the estimation of disease risk. Indeed, our previous study²¹ on prostate cancer did not find a similar significant association for CHEK2 gPVs identified in this study, but observed a significant difference in the frequency of VUS between cases and controls. This indicates that the improved annotation using functional categorization could partially explain the inconsistency in CHEK2-associated cancers,^2,6,9 highlighting the importance of functional analyses of VUS, especially for moderate-risk genes. In addition to female breast cancer, our association analysis revealed that CHEK2 gDVs were significantly associated with prostate cancer. The association between CHEK2 gPVs and prostate cancer risk has been suggested in lethal type or metastatic state with limited number of samples previously.^35-37 Accordingly, the National Comprehensive Cancer Network (NCCN) guideline has deemed this association limited evidence.² Our current largest case-control evaluation with functional analyses confirmed the impact of CHEK2 gDVs for the risk of prostate cancer regardless of patients' characteristics.

However, there are some inconsistencies between the results of our study and those of previous studies concerning cancer types other than breast and prostate cancers. No significant association was observed between CHEK2 gDVs and colorectal cancers in our case-control analysis (OR, 1.3 [95% CI, 1.0 to 1.8]; P = .10). By contrast, the association between CHEK2 gPVs and colorectal cancer remains inconsistent in literature, suggesting a possible association^38,39 or not.^37,40 Our results, with a larger number of colorectal cancer cases, supplemented the nonsignificant association from recent cohort studies including a large number of carriers.^37,40 Moreover, renal and thyroid cancers were also previously suggested to be associated with CHEK2 gPVs (OR, 2.6-3.0 for renal cancer and 1.6 for thyroid cancer).^40,41 Although there was no significant association with CHEK2 gDVs in this study, the point estimate of OR for these cancers (OR, 2.0, renal cancer and 1.8, thyroid cancer) was moderate and comparable with those of previous reports.^40,41 The results of this study suggest that, even if there is an association between CHEK2 gPVs and these cancers, considering the relative risks, CHEK2 gPVs may not have a high impact on the risk of these cancers, similar to the European population. Further data are warranted to accumulate evidence of CHEK2 involvement in these cancers.

Clinical management of gPV carriers could be changed from clarification of carrier-specific characteristics, such as age at diagnosis, family history, and prognosis. However, recommendations for the clinical management of CHEK2 gPV carriers is unconclusive, even in the European population. For breast cancer, a previous study claimed that additional MRI for CHEK2 carriers was beneficial regardless of individual risk factors.⁴² However, there is little evidence supporting additional systematic clinical management for CHEK2 gPV carriers before diagnosis of diseases. In addition, regarding prostate cancer, because of lack of clear evidence for surveillance of CHEK2 gPV carriers, it is unavoidably suggested that this issue be addressed on the basis of evidence on BRCA2 gPV carriers.⁶ Our study identified no remarkable differences in clinical characteristics between CHEK2 gDV carriers and noncarriers among breast or prostate cancer. This may indicate that the expected benefits of additional systematic clinical management are limited. Currently, the NCCN guidelines recommend early screening for breast and prostate cancers on the basis of specific individual risk factors.^2,43 Screening is usually aimed at improving prognosis or quality of life⁴; however, excessive screening, especially when based on clinical management strategies for high-penetrance genes, may sometimes cause harm and psychological burden.^44,45 In contrast to BRCA1/2 gPV carriers, for whom the same specific clinical management is recommended,² our findings suggest that all CHEK2 gDV carriers should not be strongly considered for earlier cancer screening and that additional clinical management should be restricted to individuals according to their additional individual and familial risk profiles. Especially with the relatively low prevalence of gDV carriers among the East Asian population, its concern would be more considerable.

This study has some limitations. First, this evaluation included hospital-based biobanks, and therefore, generalizing the results to the general population should be considered under caution. However, BBJ included many general hospitals and recruited individuals without suspected hereditary diseases.^13-15 Although there might be a deviation from the general population, this would not be remarkable. Second, although the ACMG/AMP guidelines were recommended for evaluating the pathogenicity of variants in association analysis,^30,46 we did not apply them in this study because gene-specific criteria for CHEK2 were still under development. To clearly distinguish our classification from ACMG/AMP-based interpretation, we used the term gDVs. Further evaluation may be warranted once such criteria are established. Third, the statistical power was limited for certain cancer types because of the limited sample size in some cancers. Therefore, cancer types that did not show a significant association in the analyses would require further data for confirmation. Finally, this study did not conduct long-term prospective observations of CHEK2 gDV carriers to confirm that no additional clinical management is required.

In conclusion, our case-control study of the unselected East Asian population, the largest to our knowledge, showed an association between CHEK2 gDVs and prostate cancer, in addition to female breast cancer, using the improved annotation by functional analysis of missense VUS. Importantly, there were no differences between CHEK2 gDV carriers and noncarriers with breast or prostate cancers with respect to various clinicopathologic characteristics. Besides, there was also no evidence that special care is required for cancer types with high incidence rate in East Asian countries. These suggest that the necessity of additional systematic clinical management for all CHEK2 gDV carriers should be carefully discussed, and standard cancer screening is recommended unless no other clinical features suggestive of cancer predisposition are noted in East Asia.

DATA SHARING STATEMENT

Primary sequencing data used in this study will be available from the NBDC human database (https://humandbs.dbcls.jp/hum0014) under NBDC Data Sharing Policy (controlled-access data Type-1) with publication.

AUTHOR CONTRIBUTIONS

Conception and design: Yuri Takehara, Yoshiaki Usui, Petra Kleiblova, Teruhiko Yoshida, Amanda B. Spurdle, Zdenek Kleibl, Yukihide Momozawa

Financial support: Petra Kleiblova, Kouya Shiraishi, Koichi Matsuda, Libor Macurek, Yukihide Momozawa

Administrative support: Makoto Hirata, Yoichiro Kamatani, Kokichi Sugano, Koichi Matsuda, Yukihide Momozawa

Provision of study materials or patients: Yoshiaki Usui, Yoichiro Kamatani, Yoshinori Murakami, Hidewaki Nakagawa, Yukihide Momozawa

Collection and assembly of data: Yuri Takehara, Yoshiaki Usui, Petra Kleiblova, Yoichiro Kamatani, Yoshinori Murakami, Mikiko Endo, Kouya Shiraishi, Koichi Matsuda, Libor Macurek, Zdenek Kleibl, Yukihide Momozawa

Data analysis and interpretation: Yuri Takehara, Yoshiaki Usui, Lenka Stolařová, Petra Kleiblova, Yusuke Iwasaki, Todd A. Johnson, Makoto Hirata, Yoichiro Kamatani, Takashi Kohno, Kokichi Sugano, Teruhiko Yoshida, Hidewaki Nakagawa, Libor Macurek, Zdenek Kleibl, Yukihide Momozawa

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).