Clinician-Reported Impact of Germline Multigene Panel Testing on Cancer Risk Management Recommendations.

BACKGROUND: With increased adoption of multi-gene panel testing (MGPT) for hereditary cancer, management guidelines now include a wider range of predisposition genes. Yet little is known about whether MGPT results prompt changes to clinicians' risk management recommendations and whether those recommendations adhere to guidelines.
METHODS: We assessed cancer risk management recommendations made by clinicians ordering MGPT for hereditary cancer at a diagnostic laboratory using an internet-based survey. We received paired pre- and posttest responses for 2172 patients (response rate = 14.3%). Unpaired posttest responses were received in 168 additional patients with positive results. All tests were 2-sided.
RESULTS: Clinicians reported a change in risk management recommendations for 76.6% of patients who tested positive for a pathogenic or likely pathogenic variant, with changes to surveillance being most common (71.1%), followed by surgical (33.6%), chemoprevention (15.1%), and clinical trial (9.4%) recommendations. Clinicians recommended risk-reducing interventions more often for patients with pathogenic variants in high-risk than moderate-risk genes (P < .001), whereas surveillance recommendations were similar for high-risk and moderate-risk genes. Guideline adherence was high for surveillance (86.3%) and surgical (79.6%) recommendations. Changes to risk management recommendations occurred in 8.8% and 7.6% of patients with uncertain and negative results, respectively.
CONCLUSIONS: Clinicians report frequent changes to cancer risk management recommendations based on positive results in both high-risk and moderate-risk genes. Reported introduction of interventions in patients with inconclusive and negative results is rare and adherence to practice guidelines is high in patients with positive results, suggesting a low probability of harm resulting from MGPT.

Germline genetic testing for hereditary cancer predisposition syndromes provides risk stratification and allows clinicians to personalize patients’ cancer treatment and risk management. In recent years, there has been a shift towards larger gene panel testing rather than a more targeted, gene-specific approach (1,2). The increased adoption of multi-gene panel testing (MGPT) to identify high-risk patients has led to refined cancer estimates for numerous susceptibility genes (2,3). Genes such as ATM, BRIP1, CHEK2, MUTYH, NBN, NF1, PALB2, RAD51C, and RAD51D have all been shown to confer a greater than twofold risk for breast, ovarian, or colorectal cancer. Consequently, cancer risk estimates have been incorporated by the National Comprehensive Cancer Network (NCCN) into the development of management guidelines for these moderate-risk genes along with high-risk genes such as APC, BRCA1/2, and Lynch syndrome genes (4,5).

When genes included on MGPT correspond to management guidelines, the increase in diagnostic yield provided by MGPT theoretically increases the clinical utility of testing. However, few empirical data exist on whether NCCN recommendations are being implemented following receipt of results, especially for patients with pathogenic variants in genes other than BRCA1/2 and Lynch syndrome. Recent studies have indicated that 70%-83% of patients with pathogenic variants (PVs) in moderate-penetrance MGPT genes have had a change in cancer risk management due to their PV result (6-8). Additionally, 79%-82% of interventions that patients reported their physicians recommending were consistent with published guidelines (8). These studies addressed the impact that moderate-penetrance PVs had on physician recommendations, but they did not compare results with the impact of high-penetrance PVs. Therefore, many questions remain about how management changes and guideline adherence are influenced by gene penetrance and intervention type. To address these questions, we present initial findings of an ongoing study aimed to assess changes in cancer risk management recommendations prompted by MGPT results, using clinician-reported data on an individual patient level.

Methods

Survey and Study Population Description

Clinicians were invited to participate in an institutional review board-exempt (Western Institutional Review Board) study using an online survey to assess cancer risk management recommendations before and after MGPT. Management recommendations pertaining to the following 5 areas were assessed: 1) surveillance, 2) risk-reducing surgery, 3) chemoprevention, 4) clinical trial eligibility, and 5) education and counseling. Pretest survey invitations were emailed to clinicians by the testing laboratory, Ambry Genetics (Aliso Viejo, CA), on submission of each MGPT order. Clinicians who completed the pretest survey were then emailed a posttest survey on results disclosure. Because the majority of changes to management were expected to arise in individuals with PVs, we sought to expand the cohort of individuals who tested positive with PVs. Beginning in February 2020, posttest surveys were sent to any clinician who had received a positive MGPT result in any gene even if no pretest survey was completed, provided the clinician had not opted out of the study. The detailed survey submission workflow is shown in Figure 1. Demographic and clinical information, including age, sex, self-reported race or ethnicity, personal history of cancer, and family history of cancer, was curated from the test requestion form and supporting clinical documents (ie, clinic notes, pathology reports, pedigrees) submitted by the ordering provider at the time of test order. Racial or ethnic groups were categorized as African American or Black, Ashkenazi Jewish, Asian, Hispanic, non-Hispanic White, Other (Brazilian, Middle Eastern, Native American, Portuguese, Sephardic Jewish, and multiple races or ethnicities), and unknown or not provided.

Figure 1.

Survey submission and response workflow: A) Paired pre- and posttest surveys are administered to providers as depicted here. Pretest surveys include an opt-out link (Active Opt Out). Providers who complete a pretest survey are sent a posttest survey once results are returned. Providers who do not complete 5 consecutive surveys will be opted out (Auto Opt Out). B) Posttest surveys are also sent to all providers with a positive multigene panel test (MGPT) result regardless of if a pretest survey was received, unless the provider has actively opted out.

MGPT Result Classification

Interpretation of sequence variations was performed according to the American College of Medical Genetics and Genomics guidelines (9). Variants identified by MGPT were classified as pathogenic, likely pathogenic, Variant of Unknown Significance (VUS), likely benign, or benign according to the Ambry 5-tier variant classification protocol (10). Pathogenic and likely pathogenic variants were defined as positive results. Case s with VUS in the absence of a pathogenic or likely pathogenic variant were defined as inconclusive. Cases with likely benign or benign findings in the absence of a pathogenic, likely pathogenic, and VUS finding were defined as negative. For the purposes of this study, MUTYH monoallelic carriers were included in the positive results group due to the existence of NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines), whereas NTHL1 and CFTR monoallelic carriers were excluded from analysis.

New Management Changes

Reported changes in cancer risk management recommendations were compared across overall test result status (positive, inconclusive, negative) and personal history of cancer. Increases in or introduction of interventions and decreases in or discontinuation of interventions were both assessed. Adjustment in the age of onset, frequency, or type of intervention (ie, analog or digital mammogram vs tomosynthesis or 3D mammogram) was also scored as a change. A copy of the pre- and posttest survey is provided in the Supplementary Methods (available online). Differences between recommendations for patients with positive results in high-risk vs moderate-risk genes were tested in the same way. High- and moderate-risk designations were made based on NCCN Guidelines in place during the study period such that genes specifically designated as high risk by NCCN and/or those with surgical recommendations were categorized as high risk, and the remaining genes were categorized as moderate risk (Table 1) (1,2). Table 1 was adapted with permission from the NCCN Guidelines for Breast/Ovarian Genetic/Familial High-Risk Assessment V3.2019 and Colorectal Genetic/Familial High-Risk Assessment V1.2019. The NCCN Guidelines and illustrations herein may not be reproduced in any form for any purpose without the express written permission of the NCCN. To view the most recent and complete version of the NCCN Guidelines, go online to NCCN.org. The NCCN Guidelines are a work in progress that may be refined as often as new data becomes available.

Table 1.

Gene groupings

Gene list	Any cancer			Breast			Colon		Ovarian
	High risk	Moderate risk	Limited evidence	High-risk (mammo/MRI 20–25 y age to initiate varies by gene), RRM	Moderate-risk mammo/MRI 30–40 y (age to initiate varies by gene), no RRM	Limited evidence no breast guidelines	High-risk colonoscopy 10–25 y (age to initiate varies by gene)	Moderate-risk colonoscopy 30–40 y (age to initiate varies by gene)	Increased risk (RRSO/BSO)
APC truncating (n = 6)	Yes	—	—	—	—	—	Yes	—	—
APC p. I1307Kb (n = 17)	—	Yes	—	—	—	—	—	Yes	—
ATM (n = 34)	—	Yes	—	—	Yes	—	—	—	—
BARD1 (n = 3)	—	—	Yes	—	—	Yes	—	—	—
BLM (n = 3)	—	—	Yes	—	—	Yes	—	—	—
BMPR1A (n = 1)	Yes	—	—	—	—	—	Yes	—	—
BRCA1 (n = 45)	Yes	—	—	Yes	—	—	—	—	Yes
BRCA2 (n = 72)	Yes	—	—	Yes	—	—	—	—	Yes
BRIP1 (n = 12)	—	Yes	—	—	—	Yes	—	—	Yes
CDH1 c (n = 2)	—	Yes	—	—	Yes	—	—	—	—
CHEK2 b , d (n = 87)	—	Yes	—	—	Yes	—	—	Yes	—
FANCC (n = 3)	—	—	Yes	—	—	Yes	—	—	—
GREM1 (n = 1)	Yes	—	no	—	—	—	Yes	—	—
MLH1 (n = 8)	Yes	—	—	—	—	—	Yes	—	Yes
MSH2 (n = 9)	Yes	—	—	—	—	—	Yes	—	Yes
MSH6 (n = 16)	Yes	—	—	—	—	—	Yes	—	—
MRE11A (n = 3)	—	—	Yes	—	—	Yes	—	—	—
MUTYH biallelic (n = 5)	Yes	—	—	—	—	—	Yes	—	—
MUTYH monoallelicb (n = 36)	—	Yes	—	—	—	—	—	Yes	—
NBN e (n = 10)	—	Yes	—	—	Yes	—	—	—	—
NF1 f (n = 3)	—	Yes	—	—	Yes	—	—	—	—
PALB2 (n = 18)	—	Yes	—	—	Yes	—	—	—	–
PMS2 (n = 16)	Yes	—	—	—	—	—	Yes	—	—
PTEN (n = 4)	Yes	—	—	Yes	—	—	—	Yes	—
RAD50 (n = 5)	—	—	—	—	—	–	—	—	—
RAD51C (n = 7)	—	Yes	—	—	—	Yes	—	—	Yes
RAD51D (n = 3)	—	Yes	—	—	—	Yes	—	—	Yes
TP53 (n = 4)	Yes	—	—	Yes	—	—	—	—	—
XRCC2 (n = 1)	—	—	Yes	—	—	Yes	—	—	—
Others (n = 48)	—	—	—	—	—	—	—	—	—

Adapted with permission from the NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines) for Breast/Ovarian Genetic/Familial High-Risk Assessment V3.2019 and Colorectal Genetic/Familial High-Risk Assessment V1.2019. BSO = bilateral salpingo-oophorectomy; mammo = mammogram; MRI = magnetic resonance imaging; RRM = risk-reducing mastectomy; RRSO = risk-reducing salpingo-oophorectomy.

The strength of evidence supporting this categorization is not well-established.

CDH1: Strong evidence supporting increased risk of lobular breast cancer.

CHEK2: With predisposition for estrogen receptor+ disease. Risk data are based only on frameshift pathogenic or likely pathogenic variants. The risks for most missense variants are unclear, but for some pathogenic or likely pathogenic variants, such as IIe157Thr, the risk for breast cancer appears to be lower. Management should be based on best estimates of cancer risk for the specific pathogenic or likely pathogenic variant.

NBN: Management recommendations are based on data derived from the 657del5 Slavic truncating pathogenic or likely pathogenic variant. Although risks for other pathogenic or likely pathogenic variants have not been established, it is prudent to manage patients with other truncating pathogenic or likely pathogenic variants similarly to those with 657del5.

At this time, there are no data to suggest an increased breast cancer risk after age 50 years. Screening recommendations only apply to individuals with a clinical diagnosis of neurofibromatosis.

Guideline Adherence

For patients with positive results, recommended interventions (mammogram, breast magnetic resonance imaging [MRI], colonoscopy, risk-reducing mastectomy [RRM], and risk-reducing salpingo-oophorectomy [RRSO]) that were included in pre- or posttest responses were compared with the NCCN Guidelines to determine adherence. Guideline adherence was defined as the proportion of patients eligible for a recommended intervention who 1) were already undergoing the intervention based on the pretest response, or 2) had it newly recommended based on posttest response. Therefore, males were excluded from calculations about breast and ovarian cancer, and patients who had already undergone surgical intervention (RRM, RRSO, colectomy) were excluded from surveillance calculations for that cancer site.

Statistical Analysis

The P values of all comparisons were derived from χ2 test or Fisher’s exact test if any observed cell count was less than 10. The odds ratios (ORs) and 95% confidence intervals (CIs) of recommendations between high-risk and moderate-risk genes were calculated based on unconditional maximum likelihood estimation and normal approximation (Wald) confidence interval, respectively. A 2-sided t test was used to compare mean ages between individuals with CHEK2 PVs and AMT PVs. P values less than .05 were considered statistically significant.

Results

Survey Responses and Test Results

Pretest surveys were sent to clinicians based on 14 987 test orders and responses were received for 3233 of these orders (response rate = 21.6%). Corresponding posttest surveys were sent for 3155 orders, and responses were received for 2172 (response rate = 68.9%). An additional 168 posttest-only surveys (without matched pretest surveys) were received for patients with positive results. These 2340 surveys were completed by 329 unique clinicians. Most respondents were genetic counselors (71.7%), and the median number of completed surveys per clinician was 2 (range = 1-145). Most patients were female (83.8%; 16.2% male), non-Hispanic White (66.7%; 8.0% African American or Black, 5.4% Ashkenazi Jewish, 4.6% Asian, 6.6% Hispanic, 4.0% Other, unknown, or not provided), and had a personal history of cancer (61.9%), consistent with frequencies reported in other hereditary cancer MGPT cohorts (3,11). Among patients about whom surveys were returned, 470 had positive results (20.1%), 562 had inconclusive results (VUS; 24.0%), 1298 had negative results (55.5%), and 10 were carriers (CFTR, NTHL1) (Table 2).

Table 2.

Population description

Characteristic	Study population	Overall lab cohort
Sex, No. (%)
Male	379 (16.2)	12 658 (13.4)
Female	1961 (83.8)	81396 (86.5)
Age at testing, y
Mean (SD)	53.0 (14.5)	53.2 (14.1)
Race and ethnicity, No. (%)
African American/Black	187 (8.0)	7281 (7.7)
Ashkenazi Jewish	127 (5.4)	4066 (4.3)
Asian	108 (4.6)	4787 (5.1)
Hispanic	154 (6.6)	6794 (7.2)
Non-Hispanic White	1561 (66.7)	57 641 (61.3)
Other or unknown	93 (4.0)	13 249 (14.1)
Personal history of cancer, No. (%)
Yes	1449 (61.9)	63 911 (66.6)
No	877 (37.5)	31948 (33.3)
Not provided	14 (0.6)	95 (0.1)
Provider type, No. (%)
Genetic counselor	236 (71.7)	NA
Nurse practitioner/physician assistant	40 (12.1)	NA
Registered nurse	19 (5.8)	NA
Physician	26 (7.9)	NA
Test result, No. (%)
Positive	470 (20.1)	11 618 (12.4)
Positive (paired pre- and posttest responses only)	302 (13.9)	11 618 (12.4)
Inconclusive	562 (24.0)	25 612 (27.2)
Negative	1298 (55.5)	55 958 (59.5)

NA = not applicable.

Clinicians reported making at least 1 results-related change to cancer risk management based on test results for 76.6% of patients with positive results, 8.8% with inconclusive results, and 7.6% with negative results (Figure 1). Thus, recommended changes were statistically significantly more likely to be reported in patients with positive results compared with inconclusive and negative results (OR = 37.99, 95% CI = 28.93 to 49.85, P < .001). There was no statistically significant difference in the frequency of reported changes in patients with inconclusive compared with negative results (OR = 1.20, 95% CI = 0.84 to 1.71, P = .32). There was a decrease in or discontinuation of interventions in 0.8%, 2.9%, and 1.6% of patients with positive, negative, and inconclusive results, respectively. Based on these low numbers, a more detailed assessment of decrease in interventions was not performed.

Among patients with positive results, recommendations about surveillance were the most frequently reported change (71.1%), followed by recommendations about surgery (33.6%), chemoprevention (15.1%), and clinical trials (9.4%) (Figure 2). There were no statistically significant differences in recommendations for surveillance (P = .75), surgery (0.71), chemoprevention (P = .45), or clinical trials (P = .61) between responses from genetic counselors and other provider types. Individuals without a personal history of cancer were slightly more likely to have any change to recommendations reported (OR = 1.6, 95% CI = 1.03 to 2.53, P = .04); however, there were no statistically significant differences in these groups when comparing specific intervention types. Clinicians reported changes in surveillance recommendations at similar rates for patients with pathogenic variants in high-risk compared with moderate-risk genes (OR = 1.22, 95% CI = 0.77 to 1.93, P = .39). More specifically, there was no difference in the proportion of individuals with increases or additions to mammogram or breast MRI recommendations between high- and moderate-risk breast cancer genes or increases or additions to colonoscopy recommendations between high- and moderate-risk colon cancer genes (Table 3). In contrast, clinicians reported higher rates of changes in recommendations for surgery (OR = 6.18, 95% CI = 3.97 to 9.60, P < .001), chemoprevention (OR = 4.12, 95% CI = 2.33 to 7.27, P < .001), and clinical trials (OR = 5.42, 95% CI = 2.43 to 12.08, P < .001) for patients with pathogenic variants in high-risk compared with moderate-risk genes.

Figure 2.

Reported management change by type. *Education includes referral to a specialist, risk counseling, family member identification, and symptom awareness. VUS = variant of unknown significance.

Table 3.

Differences in recommendation introduction between cancer risk groups

Intervention	High-risk cancer genes, %	Moderate-risk cancer genes, %	Difference in recommendation discussion?	P a
Breast cancer
Mammogram	56.0	46.4	No	.11
Breast MRI	67.2	68.2	No	.86
RRM	61.6	20.5	Yes	<.001
Colon cancer
Colonoscopy	79.0	71.1	No	.24

P values derived from χ2 test or Fisher’s exact test if any observed cell count was less than 10. No adjustments were made for multiple comparisons. All tests were 2-sided. MRI = magnetic resonance imaging; RRM = risk-reducing mastectomy.

Changes to cancer management recommendations prompted by positive test results were not always uniform among genes with similar guidelines (Tables 4-6). For example, clinicians reported discussing RRSO more frequently for patients with pathogenic variants in BRCA1/2 than other genes with RRSO recommendations (BRIP1, MLH1, MSH2, RAD51C, and RAD51D) (OR = 2.66, 95% CI = 1.27 to 5.58, P = .01). We also observed that CHEK2 pathogenic variants were less likely to prompt a change in mammogram recommendations compared with other moderate-risk breast cancer genes (P = .01), even when excluding the moderately penetrant p. I157T variant. However, based on pretest survey responses, patients with CHEK2 PVs were more likely to have a mammogram already included in their reported recommendations before receiving their positive result than patients with PVs in other moderate-risk breast genes (54.0% vs 33.8%; P = .01), so receipt of CHEK2 PV results did not require a change to mammography recommendations. Individuals with CHEK2 PVs were not statistically significantly older (P = .64) or more likely to be affected with breast cancer (P = .67) than individuals with PVs in other moderate-risk breast genes, so the reason for higher rates of pretest mammogram in this group may be due to personal or family medical history that was not assessed in this study. In addition, patients with monoallelic MUTYH PVs were less likely to have a change in colonoscopy recommendations compared with other moderate-risk colon cancer genes (P < .001). This observation is likely due to guidelines requiring that a patient have a first-degree relative with colon cancer to recommend increased surveillance, given that 86.4% of those MUTYH carriers without reported colonoscopy recommendations did not have a family history of colon cancer.

Clinicians reported discussing RRM with 20.5% of patients with PVs in moderate-risk breast cancer genes that lacked a guideline recommendation for RRM. A statistically significantly higher proportion of these patients had a personal history of breast cancer compared with those without RRM discussion (OR = 2.92, 95% CI = 1.23 to 76.97, P = .01); thus, treatment of an active cancer diagnosis may have influenced surgical decision making, and this was not a prophylactic surgery only. Furthermore, RRM discussion was introduced to a majority of patients with CDH1 or PALB2 PVs (12 of 20, 60.0%) but rarely in patients with other moderate-risk breast cancer genes (19 of 132, 14.4%). In fact, clinicians’ rates of discussing RRM for patients with PVs in PALB2 (a moderate-risk gene) were more similar to their rates of discussing RRM with high-risk genes (OR = 0.78, 95% CI = 0.28 to 2.31, P = .62) rather than other moderate-risk genes (OR = 6.86, 95% CI = 2.30 to 20.44, P < .001). CDH1 recommendations were not evaluated independently due to small numbers (n = 2).

Guideline adherence calculations were performed in 302 patients with PVs for whom paired pre- and posttest survey responses were returned. The overall adherence rate to guidelines based on survey responses was 86.31% (95% CI = 83.24% to 89.93%) for surveillance recommendations and 79.64% (95% CI = 77.36% to 89.64%) for surgical recommendations. Adherence to mammography guidelines was observed for 89.06% (95% CI = 81.42% to 96.71%) of patients with high-risk pathogenic variants and 90.24% (95% CI = 83.82% to 96.67%) of patients with moderate-risk pathogenic variants, and adherence to breast MRI guidelines was observed with 85.94% (95% CI = 77.42% to 94.45%) of high-risk pathogenic variants and 80.49% (95% CI = 71.91% to 89.07%) of moderate-risk pathogenic variants. Responses were adherent to guidelines for discussion of RRM and RRSO in 79.69% (95% CI = 69.83% to 89.54%) and 8% (95% CI = 73.20% to 90.43%) of eligible pathogenic variants carriers. Colonoscopy was included in recommendations for 96.43% (95% CI = 89.55% to 100.00%) of patients with positive results in high-risk colorectal genes and for 77.57% (95% CI = 69.67% to 85.47%) of patients with positive results in moderate-risk colorectal genes. No differences in adherence rates for mammogram (P = 1.0), breast MRI (P = .57), RRM (0.32), RRSO (P = .22), or colonoscopy (P = .31) were observed between responses from genetic counselors compared with other provider types.

Discussion

This study provides insight into how clinicians are using genetic testing to inform management of inherited cancer risks. In this survey-based study of clinicians ordering MGPT, positive genetic test results frequently prompted changes to cancer risk management recommendations. This was observed for both high- and moderate-risk genes that predispose to a variety of cancer types. Notably, clinicians reported that pathogenic variants in moderate penetrance genes ATM, CHEK2, NBN, NF1, and PALB2 prompted a change in their recommendations as often as did pathogenic variants in BRCA1/2. Appropriately, changes to surgical or chemoprevention recommendations were reported more frequently in patients with pathogenic variants in high- vs moderate-risk genes, and introduction of recommendations not specified in published guidelines were rare.

A slightly lower guideline adherence rate was observed for surgical compared with surveillance recommendations; however, our study did not capture clinical factors such as advanced disease and contraindications to invasive risk-reducing procedures. Despite this, changes in recommendations were largely adherent to published guidelines for both high- and moderate-risk genes. These findings suggest that clinicians are accurately interpreting MGPT results, referring to existing guidelines, and making appropriate recommendations accordingly.

Some results may signal the level of confidence that clinicians have in published guidelines or the strength of evidence supporting cancer risk estimates for a given gene. For example, clinicians were twice as likely to recommend RRSO in patients with PVs in BRCA1/2 vs PVs in BRIP1, MLH1, MSH2, RAD51C, or RAD51D. This could be due to several contributing factors, such as a less comprehensive understanding of ovarian cancer association with non-BRCA1/2 genes and/or ambiguity in guideline language (“recommend” vs “consider” RRSO). Other factors such as personal cancer history, family cancer history, and existing interventions before genetic testing likely influence clinicians’ recommendations, as demonstrated by more RRM discussions in patients with breast cancer and fewer colonoscopy recommendations in MUTYH carriers without family history of colon cancer, and fewer changes to mammogram recommendations in patients with CHEK2 PVs. Clinicians recommended RRM for more than one-half of patients with PALB2 PVs and all patients with CDH1 PVs despite the fact that guidelines cite insufficient evidence to recommend surgery for both of these genes. This may indicate that clinicians augment their interpretation of guidelines with other data sources, because studies have indicated that PALB2 PVs and CDH1 PVs have breast cancer penetrance similar to BRCA2 PVs (12,13) and that detection of lobular breast cancer, which is associated with CDH1 PVs, is challenging with standard imaging techniques such as mammogram (14). Such gene-specific results can identify areas of need for continued evidence accumulation, guideline development, and clinician education.

One potential risk associated with genetic testing is the misinterpretation of VUS results as PVs, which could lead to unwarranted interventions. Previous studies reported frequent use of risk-reducing surgery among patients with VUS results (15). Here, we found that clinicians responded similarly to negative and inconclusive results; reports of inconclusive results prompting surgical or medical intervention were rare. This may be due in part to the high proportion of genetic counselors among survey participants and suggests a benefit from expert oversight of genetic testing and results interpretation.

Our study has limitations. The low response rate and high proportion of genetic counselors among survey respondents limits the generalizability of these findings to the oncology community as a whole. Many patients undergo genetic testing for cancer risk without encountering a genetic counselor, and thus aspects of the current results may not fully translate to less specialized care settings (16). However, we observed that management changes and guideline adherence were similar between genetic counselors and other provider types. Another limitation is self-selection by clinicians who opted into the study (especially given a relatively low response rate), because they may differ in practice patterns from those who did not. However, the patients represented in this survey dataset are representative of the laboratory’s broader population of MGPT recipients with respect to demographics and results distribution (Table 2). Further, potential biases were reduced by controlling for unique respondents, provider type, and patient demographics in multivariable modeling.

These results have implications for clinical practice and health policy. In general, they offer assurance that ordering clinicians report appropriate and guideline-adherent recommendations, which supports the clinical utility and low probability of harm associated with MGPT. However, clinicians’ reported management changes represent a short-term measure in the continuum of care and longer follow-up of health outcomes for patients with hereditary cancer predisposition syndromes is critical. As information on the clinical validity (3,17) and cost-effectiveness (18) of MGPT emerges, so must data on the actionability of such testing. Results of this study can contribute to these efforts and expedite the development of evidence-based management guidelines.

Funding

None applicable.

Notes

Role of the funder: Not applicable.

Disclosures: The following authors are currently employed by Ambry Genetics: CH, ML, VS, HL, JSD. KB was employed by Ambry Genetics during her involvement in this project. AK performed an unrelated research project funded by Myriad Genetics from 2017–2019.

Author contributions: CH: Conceptualization, Methodology, Data Curation, Writing—Original Draft, Writing—Review and Editing, Visualization, Project Administration. KB: Methodology, Investigation, Data curation, Writing—Review and Editing, Visualization. ML: Formal Analysis, Investigation, Writing—Review and Editing, Visualization. VS: Conceptualization, Methodology, Writing—Review and Editing, Visualization. HL: Conceptualization, Methodology, Writing—Review and Editing, Visualization. JSD: Conceptualization, Methodology, Investigation, Writing—Review and Editing, Visualization, Project Administration. AK: Writing—Review and Editing, Visualization, Supervision.

Disclaimers: NCCN makes no warranties of any kind whatsoever regarding their content, use or application and disclaims any responsibility for their application or use in any way.

Prior presentations: Preliminary results of this study were presented at the 2021 American College of Medical Genetics and Genomics (ACMG) Annual Clinical Genetics Meeting.

Data Availability

The data that support the findings of this study are available on request from the corresponding author, CH, with reasonable privacy restrictions.

Click here for additional data file.

Table 4.

Gene-specific differences in breast cancer risk management changes

Gene	Mammogram					Breast MRI					RRM
	Individuals with PV	Compared with high-risk breast genes		Compared with moderate-risk breast genes		Individuals with PV	Compared with high-risk breast genes		Compared with moderate-risk breast genes		Individuals with PV	Compared with high-risk breast genes		Compared with moderate-risk breast genes
	No. intervention recommended/No. with PV per gene (%)	No. intervention recommended/No. with high-risk PV (%)	P b	No. intervention recommended/No. with moderate risk PV (%)	P b	No. intervention recommended/No. with high-risk PV (%)	No. intervention recommended/No. with moderate-risk PV (%)	P b	No. intervention recommended/No. with high-risk PV (%)	P b	No. intervention recommended/No. with moderate-risk PV (%)	No. intervention recommended/No. with moderate-risk PV (%)	P b	No. intervention recommended/No. with moderate-risk PV (%)	P b
BRCA1/BRCA2	67/117 (57.3)	NA		70/151 (46.4)	.08	78/117 (66.7)	NA		103/151 (68.2)	.79	73/117 (62.4)	NA		31/151 (20.5)	<.001
ATM	17/32 (53.1)	70/125 (56.0	.77	51/117c (43.6)	.34	22/32 (68.8)	84/125 (67.2)	.87	79/117c (67.5)	.90	7/32 (21.9)	77/125 (61.6)	<.001	24/117c (20.5)	.87
CHEK2 (including p. I157T)	26/79 (32.9)	70/125 (56.0)	.001	39/65c (60.0)	.01	50/79 (63.3)	84/125 (67.2)	.57	49/65c (75.4)	.12	9/79 (11.4)	77/125 (61.6)	<.001	20/65c (30.8)	.006
CHEK2 (excluding p. I157T)	17/56 (30.4)	70/125 (56.0)	.001	39/65c (60.0)	.001	35/56 (62.5)	84/125 (67.2)	.54	49/65c (75.4)	.13	7/56 (12.5)	77/125 (61.6)	<.001	20/65c (30.8)	.02
NBN	6/9 (66.7)	70/125 (56.0)	.73	63/141c (44.7)	.30	7/9 (77.8)	84/125 (67.2)	.72	95/141c (67.4)	.72	1/9 (11.1)	77/125 (61.6)	.004	29/141c (20.6)	.69
PALB2	10/16 (62.5)	70/125 (56.0)	.79	60/133c (45.1)	.29	12/16 (75.0)	84/125 (67.2)	.78	90/133c (67.7)	.78	9/16 (56.3)	77/125 (61.6)	.79	21/133c (15.8)	<.001

Patients with more than 1 co-occurring pathogenic variant are removed from gene-specific totals. NA represent calculations limited to genes with n greater than 5. MRI = magnetic resonance imaging; PV = pathogenic variant; RRM = risk-reducing mastectomy.

P values derived from χ2 test or Fisher’s exact test if any observed cell count was less than 10. No adjustments were made for multiple comparisons. All tests were 2-sided.

Specified gene removed from analysis.

Table 5.

Gene-specific differences in colorectal cancer risk management changes

Gene	Individuals with PV	Colonoscopy
		Compared with high-risk colorectal genes		Compared with moderate-risk colorectal genes
	No. intervention recommended/No. with PV per gene (%)	No. intervention recommended/No. with high-risk PV (%)	P b	No. intervention recommended/No. with moderate-risk PV (%)	P b
APC truncating	3/5 (60.0)	45/56c (80.4)	.29	101/142 (71.1)	.63
MLH1/MSH2	13/16 (81.3)	35/45c (77.8)	1.0	101/142 (71.1)	.56
MSH6	11/15 (73.3)	37/46c (80.4)	.72	101/142 (71.1)	1.0
PMS2	12/14 (85.7)	35/46c (76.1)	.71	101/142 (71.1)	.35
APC p. I1307K	13/16 (81.3)	49/62 (79.0)	1.0	87/125c (69.6)	.40
CHEK2 (incuding p. I157T)	65/79 (82.3)	49/62 (79.0)	.63	31/57c (54.4)	<.001
CHEK2 (excluding p. I157T)	46/56 (82.1)	49/62 (79.0)	.67	31/57c (54.4)	.002
MUTYH monoallelic	13/34 (38.3)	49/62 (79.0)	<.001	86/104c (82.7)	<.001

Patients with more than 1 co-occurring pathogenic variant were removed from gene-specific totals. Calculations limited to genes with n greater than 5; PV = pathogenic variant.

P values derived from χ2 test or Fisher’s exact test if any observed cell count was less than 10. No adjustments were made for multiple comparisons. All tests were 2-sided.

Specified gene removed from analysis.

Table 6.

Gene-specific differences in ovarian cancer risk management changes

Gene	Individuals with PV	RRSO
		Compared with other increased risk ovarian genes
	No. intervention recommended/No. with PV per gene (%)	No. intervention recommended/No. with PV (%)	P b
BRCA1/BRCA2	78/117 (66.7)	17/39c (43.6)	.01
BRIP1	5/11 (45.5)	89/144c (61.8)	.34
MLH1/MSH2	6/16 (37.5)	89/139c (64.0)	.06
RAD51C/D	5/10 (50.0)	90/146c (61.6)	.51

Patients with more than 1 co-occurring pathogenic variant were removed from gene-specific totals. Calculations limited to genes with n  greater than 5. PV = pathogenic variant; RRSO = risk-reducing saplingo-oophorectomy.

P values derived from χ2 test or Fisher’s exact test if any observed cell count was less than 10. No adjustments were made for multiple comparisons. All tests were 2-sided.

Specified gene removed from analysis.