Pathogenic Variants in CHEK2 Are Associated With an Adverse Prognosis in Symptomatic Early-Onset Breast Cancer

Abstract

PURPOSE

Checkpoint kinase 2 (CHEK2) is frequently included in multigene panels. We describe the associated outcomes among carriers of CHEK2 pathogenic variants in young patients with symptomatic breast cancer.

PATIENTS AND METHODS

Participants (N = 2,344) in the Prospective Outcomes in Sporadic Versus Hereditary Breast Cancer study had a diagnosis of primary invasive breast cancer at age ≤ 40 years. Summary statistics were used to compare tumor characteristics among CHEK2+ carriers with those who were CHEK2−. Kaplan-Meier curves were used to demonstrate overall survival (OS) and distant disease-free survival.

RESULTS

Overall, 53 of the 2,344 participants (2.3%) had a pathogenic CHEK2 variant. CHEK2+-associated tumors were significantly more likely to be grade 2, estrogen receptor and progesterone receptor–positive compared with CHEK2− tumors (grade 2, n = 28 of 52 [53.8%] v n = 803 of 2,229 [36.0%]; P = .029). CHEK2-associated tumors were significantly more likely to have nodal involvement (N1, n = 37 of 53 [69.8%] v 1,169 of 2,253 [51.9%]; P = .0098) and demonstrated a trend toward multifocality. A higher proportion of participants with CHEK2+ variants with invasive breast cancer were obese than were those with CHEK2− variant (28.3% v 18.8%; P = .039). Univariate and multivariable analyses revealed that OS and distant disease-free survival were significantly worse in CHEK2+ versus CHEK2− carriers (OS hazard ratio, 1.58; 95% CI, 1.01 to 2.48; P = .043).

CONCLUSION

This work highlights the adverse prognosis associated with breast cancer in carriers of CHEK2 pathogenic variants. It also identifies a potential association among obesity, family history, and breast cancer risk in young CHEK2 gene carriers.

INTRODUCTION

A consensus guideline published by the American Society of Breast Surgeons in 2019 advocates the routine testing for BRCA1, BRCA2, and PALB2 genes, along with other high- and moderate-risk breast cancer susceptibility genes to facilitate practice standardization in the context of hereditary cancer susceptibility.¹ Despite this consensus opinion, there is residual concern regarding the true clinical validity and utility of moderate risk genes in the context of patient care. One moderate risk gene that remains a focus for such evaluation is checkpoint kinase 2 (CHEK2).

Context

• Key Objective

• The aim of this analysis was to determine whether germline CHEK2 gene alterations influence tumor histopathological phenotype and survival in early-onset breast cancer.

• Knowledge Generated

• Breast cancers occurring in the context of a CHEK2 pathogenic variant are predominantly estrogen-receptor positive but patients have significantly worse overall survival and distant disease-free survival compared with individuals without an identifiable gene alteration. Obesity may influence breast cancer risk in women with an inherited CHEK2 pathogenic variant.

• Relevance

• There is increasing interest in the use of germline cancer susceptibility testing for patient risk stratification and precision medicine. This work highlights the potential interaction between obesity and CHEK2-associated cancer risk. The findings also suggest that CHEK2 gene carriers with breast cancer may have a poorer outcome compared with noncarriers. These are potential areas for research that may influence cancer risk management and primary intervention at the time of cancer diagnosis.

CHEK2 is a serine/threonine kinase that functions as a tumor suppressor gene necessary for cell cycle checkpoint regulation, the inhibition of cellular proliferation, and activation of DNA repair pathways or apoptosis.^2-4 Pathogenic variants in CHEK2 abolish the protein kinase activity and confer a moderate increase in breast cancer risk which is two- to threefold above the baseline population level.^4-10 CHEK2 pathogenic variants are most prevalent among individuals of European descent, and several founder mutations exist.^4,11,12 Of these, CHEK2 c.1100delC is the most frequently identified among Northern Europeans, with an observed population frequency of 0.5%-1%.^3,5,6

Presently, there are limited data on the effects of a germline CHEK2 mutation on breast tumor biology. In general, CHEK2-associated tumors are considered more likely to be grade 2, estrogen receptor–positive (ER+) and progesterone receptor–positive (PR+) compared with CHEK2− tumors.^3,5,13-15 A trend toward bilateral disease at presentation with higher levels of nodal involvement has also been suggested, indicating a potentially more aggressive underlying tumor biology that may influence patient outcome.^15,16

The primary aim for this study is to describe the histopathological tumor phenotype of CHEK2-associated breast cancers in a young-onset, symptomatic cohort. The secondary aim is to determine whether a germline CHEK2 pathogenic variant differentially influences breast cancer survival compared with that of sporadic control subjects in early-onset breast cancer.

PATIENTS AND METHODS

Sample Population

The sample population has been obtained from the Prospective Outcomes in Sporadic Versus Hereditary Breast Cancer (POSH) Study (UK Clinical Research Network identifier: 1137).^17,18 POSH was a large, prospective, multicenter, cohort study to which 3,095 women were recruited from 127 UK hospitals between January 1, 2000, and January 31, 2008 (Medical Research Ethics Committee approval no. 00/6/69).^17,18 The study was designed to evaluate which factors influence prognosis and treatment response in women diagnosed with early-onset breast cancer.^17,18

Inclusion required a primary diagnosis of invasive breast cancer at age ≤ 40 years.^17,18 Diagnostic pathology data were collected for all patients, including tumor size; stage; grade; multifocality; and ER, PR, and HER2 receptor status. Genomic DNA was extracted from whole blood and a customized gene panel using a multiplex amplicon-based library preparation system (Fluidigm Access Array; Fluidigm, South San Francisco, CA). Extracted DNA was available for 2,907 recruited participants (96%). Next-generation sequencing (NGS) was performed using an Illumina platform (Illumina Sequencing Technology; Illumina, San Diego, CA).²⁰ Sequencing targeted exonic regions and the exon/intron boundaries for the panel of genes, which included CHEK2, BRCA1, BRCA2, PALB2, ATM, and TP53.

In total, 2,344 patients (78%) were included in the analysis population for this study and 677 (22%) were excluded. Participants were excluded because of missing genotyping data (n = 159), a germline pathogenic variant in BRCA1, BRCA2, PALB2, ATM, or TP53 (n = 400), M1-stage disease (n = 74), age 41-50 years (n = 42), and missing primary tumor data (n = 2). Samples failed NGS due to either inferior quality or low concentration of DNA (Appendix Fig A1).

Analysis

Bioinformatics analysis of sequence data was conducted as previously described.¹⁹ CHEK2 variants were classified using the American College of Medical Genetics and Genomics guidelines,²⁰ which assign pathogenicity on the basis of several factors, including population frequency, segregation, presence in databases of functional significance, and in silico predictions. Filters were applied to identify those variants that were clearly or highly likely to be pathogenic (American College of Medical Genetics and Genomics class 4 and 5) or protein truncating (eg, frameshift, stop-gain or stop-loss) with a minor allele frequency < 1% in the genome aggregation database.^21,22 All filtered variants were manually reviewed and compared with the assigned pathogenicity in ClinVar.²³ Variants of uncertain clinical significance and hypomorphic alleles were classified as mutation negative. Confirmatory testing was completed by Sanger sequencing. For the purpose of this analysis, CHEK2 pathogenic variant carriers were categorized as CHEK2+ and noncarriers as CHEK2−.

Statistical Analysis

Statistical analysis was performed according to a prespecified statistical analysis plan. Summary statistics are used to describe the cohort. The Mann-Whitney test was used for continuous variables and the Pearson χ² test was used for categorical variables to identify any specific differences between CHEK2+ and CHEK2− patients. Kaplan-Meier curves were used to demonstrate overall survival (OS) and distant disease-free survival (DDFS). Differential survival between CHEK2+ and CHEK2− groups was compared using a univariate Cox regression model. Multivariable analysis was also performed using Cox regression.

RESULTS

The complete analyzed cohort comprised 2,344 participants diagnosed with a primary invasive breast cancer when younger than 40 years (Appendix Table A1). A confirmed CHEK2 pathogenic variant was found in 53 of the 2,344 participants (2.3%) in this study cohort (1.9%; n = 53 of 2,744 of the whole POSH cohort).

The Northern European founder CHEK2 c.1100delC was the most frequently identified pathogenic variant, accounting for 36 of 53 (67.9%) of all CHEK2 pathogenic variants. In total, three participants also possessed a germline pathogenic variant in either BRCA2 (n = 2) or PALB2 (n = 1). We identified 28 individuals with variants of uncertain significance in CHEK2. This included 27 missense variants (including six instances of the low-penetrance variant c.470T>C, p.Ile157Thr) and one in-frame deletion.³ In routine clinical practice, individuals with low-penetrance risk alleles or variants of uncertain significance would be treated as mutation negative, so these cases were included in the control group (Appendix Table A2).

Overall, CHEK2+-associated tumors were significantly more likely to be grade 2 at presentation compared with CHEK2− (grade 2: n = 28 of 52 [53.8%] v CHEK2−: n = 803 of 2,229 [36.0%]; P = .029; Table 1). There was no difference in baseline tumor size between CHEK2+ and CHEK2− tumors (P = .73).

TABLE 1.

Baseline Histopathological Characteristics of the CHEK2 Cohort: Comparison of CHEK2 Variant Carriers and Non-CHEK2 Carriers

CHEK2+-associated tumors also had significantly higher levels of nodal involvement compared with CHEK2− tumors. In total, 37 of 53 (69.8%) CHEK2+ patients presented with N1-stage disease versus 1,169 of 2,253 CHEK2− patients (51.9%) (P = .0098). In addition, tumors from CHEK2+ patients demonstrated a trend toward multifocality at presentation compared with CHEK2− (n = 22 of 52 [42.3%] v n = 624 of 2,085 [29.9%]; P = .055). We compared baseline tumor grade, size, and focality between CHEK2 c.1100delC carriers and all other truncating-variant carriers and found no significant difference (Table 1; Fig 1).

FIG 1.

Baseline histopathological characteristics of the CHEK2 cohort: comparison of the tumor focality, grade and lymph node involvement between CHEK2 truncating variant carriers and non-CHEK2 carriers.

Tumors from CHEK2+ patients were significantly more likely to be ER+ and PR+ compared with CHEK2− (Table 1). In total, 47 of 53 (88.7%) tumors in CHEK2+ patients were ER+ compared with 1,557 of 2,279 CHEK2− (68.3%; P = .0016) and 33 of 42 (78.6%) tumors in CHEK2+ patients were PR+ compared with 1,084 of 1,848 CHEK2− (58.7%; P = .0094). CHEK2+ patients were also significantly less likely to have a triple-negative breast cancer (TNBC) compared with CHEK2− patients (P = .0022). In total, one of 53 (1.9%) CHEK2+ patients had TNBC compared with 417 of 2,291 CHEK2− (18.2%; Table 1; Fig 2). There was no significant association with HER2 receptor status and CHEK2 genotype.

FIG 2.

Hormone receptor status of the CHEK2 cohort: comparison of estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 (HER2) receptor status between CHEK2 variant carriers and non-CHEK2 carriers. Values are reported as a percentage of the cohort. Samples were derived from the Prospective Outcomes in Sporadic Versus Hereditary Breast Cancer Cohort.

The median age of patients at cancer onset was 37 years (interquartile range [IQR], 34-39 years) for both groups (CHEK2+ mutation carriers and CHEK2−). Most CHEK2+ carriers were white (n = 50 of 53; 94.3%). There was no association between family history of breast cancer and median Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA) score comparing patients with CHEK2+ and those with CHEK2− tumors (Appendix Table A1). In total, 36 of 51 (70.6%) patients with CHEK2+ tumor had no family history of breast cancer. The median Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm score was 0.03 for both groups (IQR, 0.02-0.07 v 0.02-0.05 for CHEK2+ and CHEK2−, respectively; P = .86). However, we noted that a higher proportion of patients with CHEK2+ tumor with invasive breast cancer were obese compared with the CHEK2− group (28.3% v 18.8%; P = .039).

Most patients received adjuvant chemotherapy. The most frequent regimen included anthracyclines with or without an additional of taxane. There were no significant differences in chemotherapy received between the CHEK2+ and CHEK2− groups (Table 2). The significant difference in tamoxifen use reflects the higher proportion of ER+ cases in the CHEK2+ group. A nonsignificant trend toward mastectomy was identified among patients in the CHEK2+ group. In total, 36 of 53 (67.9%) CHEK2+ patients underwent mastectomy as the primary surgical intervention, versus 1,122 of 2,291 (49.0%) in the CHEK2− group (P = .054).

TABLE 2.

Treatment Characteristics of the CHEK2 Cohort: Comparison of the Treatment Protocol Relative to Genotype

The median duration of follow-up was 8.2 years. Contralateral breast cancers were more frequently observed in the CHEK2+ group compared with patients in the CHEK2− group. A contralateral breast cancer was observed in five of 53 CHEK2+ patients (9.4%) at 10 years compared with 85 of 2,291 patients (3.7%) in the CHEK2− group. Of the five CHEK2+ individuals with contralateral breast cancer, two had bilateral disease at presentation and one participant had a contralateral breast cancer in the same year as their primary breast cancer diagnosis.

Subgroup analysis revealed that the observed increase in contralateral breast cancer risk observed among CHEK2+ carriers occurred in the context of familial breast cancer. In total, a contralateral breast cancer developed in three of 15 CHEK2+ patients (20.0%) with a positive family history of breast cancer, compared with one of 36 patients (2.8%) in the CHEK2+ group without family history. Family history data were missing for one individual with a contralateral breast cancer. This difference was apparent in the first 5 years after breast cancer diagnosis. The contralateral breast cancer rates observed among CHEK2+ patients without a family history of breast cancer were similar to those in the CHEK2− group with or without a family history (Table 3). Furthermore, three of five CHEK2+ individuals with contralateral disease had a family history of breast cancer and were obese.

TABLE 3.

Contralateral Breast Cancer Risk at 5 and 10 Years for CHEK2-Truncating Variant Carriers Versus CHEK2−

By univariable analysis (UVA), we identified significantly worse OS in the CHEK2+ group versus the CHEK2− group (hazard ratio [HR], 1.58; 95% CI, 1.01 to 2.48; P = .043; Fig 3). At 5 years, OS was 75.1% (95% CI, 60.9 to 84.7) among patients with CHEK2+ versus 85.1% (95% CI, 83.5 to 86.5) in the CHEK2− group. At 10 years, OS was 60.7% (95% CI, 42.5 to 74.8) among the CHEK2+ group versus 70.2% (95% CI, 67.8 to 72.5) in the CHEK2− group. The observed difference in OS between CHEK2+ patients and those in the CHEK2− group was maintained after adjustment for known prognostic factors, including age at diagnosis, BMI, grade, maximum invasive size, hormone receptor status, nodal involvement, ethnicity, and taxane treatment, in a multivariable analysis (HR 1.65; 95% CI, 1.05 to 2.59; P = .03).

FIG 3.

Kaplan-Meier plot of overall survival for CHEK2 variant carriers versus noncarriers (CHEK2−) after univariable analysis.

DDFS was also significantly worse in the CHEK2 group+ compared with CHEK2− (UVA HR, 1.62; 95% CI, 1.06 to 2.48; P = .025; Appendix Fig A2). At 5 years, DDFS was 61.8% (95% CI, 47.2 to 73.4) among the CHEK2+ group versus 77.7% (95% CI, 75.9 to 79.4) in the CHEK2− group. At 10 years, DDFS was 56.8% (95% CI, 41.8 to 69.3) among the CHEK2+ group versus 69.0% (95% CI, 66.7 to 71.2) in the CHEK2− group (Appendix Fig A2). The observed difference in DDFS between the CHEK2+ and CHEK2− groups was maintained after adjustment for known prognostic factors in a multivariable analysis (HR, 1.60; 95% CI, 1.04 to 2.46; P = .033).

DISCUSSION

This study identified pathogenic variants in CHEK2 in 1.9% of unselected, early-onset breast cancers within a UK population. CHEK2 c.1100delC was the most frequently identifiable variant, accounting for 67.9% of all mutations. Decker et al⁵ identified a truncating CHEK2 variant in 1.6% of unselected breast cancer cases within a large UK population-based study and CHEK2 c.1100delC accounted for 81% of truncating mutations.

We have determined that patients with a germline CHEK2 pathogenic variant in whom breast cancer develops have an adverse outcome with reduced OS and DDFS compared with those without a CHEK2 pathogenic variant—a relationship that persists after adjustment for known prognostic factors. We also noted that of the 74 individuals who were removed from the analysis because they had M1 disease, four (5.7%) carried a pathogenic CHEK2 variant. This is in comparison with the total eligible cohort within the POSH study, among whom 74 of 3,021 patients (2.4%) had M1 stage disease (Appendix Fig A1). Although the numbers are small, this observation is supportive of future work to better define the relationship between CHEK2 mutational status and outcome in early-onset breast cancer.

Our results are consistent with those of Schmidt et al,⁷ who found that CHEK2, c.1100delC mutation carriers had a worse recurrence-free and breast cancer–specific survival than CHEK2− (HR, 1.7 [95% CI, 1.2 to 2.4], P = .006; and HR, 1.4 [95% CI, 1.0-2.1], P = .072, respectively), but after multivariable analysis the difference was no longer statistically significant. Weischer et al¹⁵ also found that ER+ CHEK2 c.1100delC carriers within the Breast Cancer Association Consortium had a significantly increased risk of breast cancer–specific death, which persisted after multivariable analysis (HR, 1.63; 95% CI, 1.24 to 2.15; P < .001). They also identified a 2.8-fold risk of a second breast cancer (HR, 2.8; 95% CI, 2.00 to 3.83; P < .001).¹⁵

In our study, breast cancers occurring in patients with a CHEK2 pathogenic variant were significantly more likely to be grade 2, ER+ and PR+ compared with age-matched control patients but with no difference in HER2 expression. They were not associated with a TNBC phenotype. These results are consistent with those of other reports of histopathological tumor phenotype associated with germline mutations in CHEK2. We were not able to classify these into intrinsic subtypes, because gene expression data were not available for the POSH study.

Decker et al⁵ found that CHEK2-associated tumors were significantly more likely to be ER+ (odds ratio, 3.42; 95% CI, 2.33 to 5.21; P = 1.5 × 10⁻¹¹). Cybulski et al¹⁴ also found that CHEK2-associated cancers were significantly more likely to be ER+ and PR+ (69.7% v 63.1%; P = .002) and (77% v 68.7%; P < .001, respectively). Weischer et al¹⁵ also found a significantly higher frequency of ER+ and PR+ breast cancers in c.1100delC carriers than in noncarriers (CHEK2−; 63% v 57%, P < .001; and 46% v 43%, P = .01, respectively). Couch et al²⁴ found no CHEK2 pathogenic variants among 1,824 patients with TNBC. The strong association of a germline CHEK2 pathogenic variant with ER+ disease may support the use of ER blockade such as tamoxifen as chemoprophylaxis in patients with a CHEK2-related breast cancer risk.¹⁴

Within the POSH cohort, family history did not predict the presence of a germline CHEK2 pathogenic variant. This is consistent with a low to moderate overall increase in risk compared with the population average.

A higher proportion of patients carrying a CHEK2 pathogenic variant who had invasive breast cancer were obese compared with noncarriers. It is well recognized that obesity is associated with an increased risk of postmenopausal breast cancer, but, to our knowledge, it has not been associated with an increased risk in premenopausal breast cancer.²⁵ In 2017, the Premenopausal Breast Cancer Collaborative Group assessed the BMI-associated breast cancer risk in 758,592 premenopausal women and found an inverse correlation between age and the associated risk. The results of this study suggest a potential association in young patients with breast cancer between CHEK2-related risk and obesity.^26,27 We had previously shown that BMI was associated with an adverse prognosis after breast cancer diagnosis, independent of other known risk factors; therefore, we included BMI in the multivariable analysis.²⁸

Invasive breast cancers occurring in the context of a pathogenic CHEK2 variant demonstrated a trend toward multifocality with significantly higher levels of nodal involvement at presentation. Our study confirmed the findings of Cybulski et al,¹⁴ who found that CHEK2-associated cancers had higher levels of nodal involvement. Multifocal tumor pathology is likely to be one of the key drivers for more frequent mastectomy among CHEK2 carriers.

The contralateral breast cancer rate among CHEK2 pathogenic variant carriers was higher than that of patients in the CHEK2− group at both 5 and 10 years. Although the absolute numbers of cases was small, we noted that CHEK2 carriers with a family history had a contralateral breast cancer rate that was higher than that of noncarriers. Within the POSH cohort, family history was not an independent predictor of outcome.²⁹

In 2004, de Bock et al³⁰ reported that at 5 years, a contralateral breast cancer had developed in 21% of CHEK2 c.1100delC carriers compared with 4% of noncarriers, representing an almost sixfold increase in risk. Decker et al⁵ found that CHEK2-associated tumors were significantly more likely to be bilateral at presentation (OR, 3.27; 95% CI, 1.66 to 5.83; P = .0014). This was supported by Kilpivaara et al¹³ who also noted a strong association with bilateral disease at presentation. In our study, we noted the presentation with bilateral disease, particularly in the context of other risk factors (ie, obesity, family history)

There are several strengths of this study. Participants were ascertained shortly after diagnosis and presented with symptomatic rather than screen-detected breast cancers. The clinical data available are comprehensive and allow multivariable data analysis. There was no systematic bias in selecting patients for the study or for genotyping, and the cohort is representative of the general UK breast cancer population.¹⁷ Furthermore, carriers of additional high-risk breast cancer susceptibility genes were excluded from the analysis, allowing a clean comparison between CHEK2 pathogenic variant carriers and noncarriers.

CHEK2 is commonly included in multigene panel testing and most frequently identified in the context of a patient presenting with breast cancer. Although the numbers are small, we observed a higher rate of a contralateral breast cancer associated with a pathogenic CHEK2 variant. The increased incidence appears to be confined to carriers with a family history of breast cancer. For carriers with no family history, the incidence of contralateral breast cancer is no greater than the incidence in a noncarrier population.

CHEK2 pathogenic variant carriers had significantly poorer OS. This is in contrast to analysis in the same cohort that showed prognosis was not altered in a multivariable analysis of BRCA1 or BRCA2 carriers compared with noncarriers.¹⁹

Including CHEK2 genotyping as part of population risk-stratified approaches to inform targeted screening and improve early diagnosis is aspirational. The current approach for managing moderate breast cancer risk within the United Kingdom is annual mammograms from the age of 40 years.³¹ The use of chemoprophylaxis may be effective, given the high proportion of hormone receptor–positive breast cancers.³¹ However, neither measure has yet been tested in this particular group of patients.

Our study highlights the importance of including effective measures to address lifestyle risk factors, particularly for maintaining a healthy body weight, for premenopausal women at increased breast cancer risk.

In conclusion, this work describes the characteristics and clinical outcomes for patients who present with invasive, early-onset breast cancer and carry a CHEK2 pathogenic variant. Because a pathogenic CHEK2 variant is likely to be identified in approximately 2% of white patients with breast cancer, including those age ≤ 40 years, clinicians should be aware of the adverse prognosis and potential effect of family history on contralateral cancer risk in planning cancer treatment. Our findings also highlight the importance of preventing and managing obesity in the context of breast cancer susceptibility.

Appendix

FIG A1.

Overview of the sample population. Individuals with a confirmed pathogenic/likely pathogenic variant in CHEK2 were classified as CHEK2+. POSH, Prospective Outcomes in Sporadic Versus Hereditary Breast Cancer.

FIG A2.

Kaplan-Meier plot demonstrating distant disease-free survival (DDFS) for CHEK2 variant carriers versus noncarriers after univariate analysis.

TABLE A1.

Baseline Characteristics of the Cohort

TABLE A2.

Summary of the CHEK2 Variants Identified Across the Sample Cohort

AUTHOR CONTRIBUTIONS

Conception and design: Stephanie L. Greville-Heygate, Ellen Copson, Diana M. Eccles

Administrative support: Linda Haywood

Provision of study material or patients: William J. Tapper, Alison M. Dunning, Linda Haywood

Collection and assembly of data: Stephanie L. Greville-Heygate, Ramsey I. Cutress, Alison M. Dunning, Linda Haywood, Diana M. Eccles

Data analysis and interpretation: Stephanie L. Greville-Heygate, Tom Maishman, William J. Tapper, Ramsey I. Cutress, Ellen Copson, Louise J. Jones, Diana M. Eccles

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).

Stephanie L. Greville-Heygate

Employment: Ardens Health Informatics (I)

Leadership: Ardens Health Informatics (I)

Stock and Other Ownership Interests: Ardens Health Informatics

Ramsey I. Cutress

Research Funding: Seca (Inst)

Ellen Copson

Honoraria: Roche, Pfizer, AstraZeneca

Consulting or Advisory Role: Eli Lilly, Nanostring, Pfizer

Speakers' Bureau: Roche, Pfizer, AstraZeneca

Research Funding: SECA

Travel, Accommodations, Expenses: Roche, AstraZeneca

Linda Haywood

Employment: Bupa (I), GE (Finance) (I)

Stock and Other Ownership Interests: GE (Finance) (I)

Diana M. Eccles

Honoraria: AstraZeneca, Pierre Fabre

Consulting or Advisory Role: AstraZeneca

Travel, Accommodations, Expenses: Pierre Fabre

No other potential conflicts of interest were reported.