Clinical management among individuals with variant of uncertain significance in hereditary cancer: A systematic review and meta-analysis

Abstract

Improper medical use of variant of uncertain significance (VUS) remains a concern in hereditary cancer genetic testing. The goal of this study was to assess the association between pathogenic and likely pathogenic (P/LP), VUS, and benign and likely benign (B/LB) genetic test results and cancer-related surgical and screening management. Systematic searches of Medline, Embase, EBSCO CINAHL Plus, and PsycINFO were conducted from 1946 to August 26, 2020. Eligible studies included individuals with cancer genetic test result and surgical or screening management outcomes. We reviewed 885 abstracts and 22 studies that reported relevant surgical and screening outcomes were included. Meta-analysis revealed significantly higher surgical rates among individuals with P/LP than among those with VUS for therapeutic mastectomy with contralateral prophylactic mastectomy (OR = 7.35, 95% CI, 4.14–13.64), prophylactic mastectomy (OR = 3.05, 95% CI, 1.5–6.19), and oophorectomy (OR = 6.46, 95% CI, 3.64–11.44). There were no significant differences in therapeutic mastectomy, or breast conservation or lumpectomy rates between individuals with P/LP and VUS, or in any outcomes between patients with VUS and B/LB. Studies evaluating screening outcomes were limited, and results were conflicting. Comprehensive analysis do not indicate that a significant number of individuals with VUS results undergo inappropriate clinical management.

1 ∣. INTRODUCTION

Up to 10% of all cancers are attributable to germline genetic mutations and identifying individuals with hereditary predispositions to cancer is critical for cancer prevention, early detection, and treatment decisions.¹ Germline genetic testing can identify pathogenic or likely pathogenic variants (P/LP), variants of uncertain significance (VUS), or benign or likely benign variants (B/LB), per the American College of Medical Genetics and American Association for Molecular Pathology (ACMG-AMP) criteria.² Multi-gene panel testing for hereditary cancer, expanded access to genetic testing in non-European populations, and testing of rare and newly described genes have contributed to increased detection of VUS results.³ Concern about misinterpretation of these uncertain genetic test results leading to suboptimal treatment remains and there is a need to empirically assess the overall impact of VUS results on medical management.

Optimizing clinical management of individuals and their relatives is a primary motivation for hereditary cancer (germline) genetic testing in clinical oncology. Individuals identified to have P/LP result are recommended to pursue increased cancer screenings and consider risk-reducing surgeries according to national guidelines. For example, a woman with a P/LP finding in the BRCA1 gene is recommended by National Comprehensive Cancer Network (NCCN) guidelines to initiate breast cancer screening at age 25, to have mammogram and breast MRI annually, and to consider bilateral risk-reducing mastectomy.⁴ In patients with VUS or B/LB results, medical decisions should be based on patients' personal and family histories of cancer and other clinical factors, rather than their uninformative genetic test results. The ACMG-AMP² and the NCCN recommend that VUS results not be used to guide medical decision making.⁴ For example, a woman with a B/LB or VUS finding in the BRCA1 gene is recommended to initiate annual mammograms for breast cancer screening at the age of 40–50, or earlier if she has a family history of early-onset breast cancer, based on USPSTF guidelines.⁵ All else being equal, guideline concordant care should result in rates of cancer screening and risk-reducing surgery among patients with VUS comparable to patients with B/LB results, whereas patients with P/LP results are expected to have higher rates of screening and risk-reducing surgery in accordance with guidelines.

Two VUS-associated treatment patterns have been reported in the literature. The guideline-inconsistent pattern suggests that patients with VUS results undergo inappropriate medical management, such as unnecessary, irreversible surgery and excessive cancer screenings putting them at risk for overscreening and unnecessary follow-up of benign abnormalities. These management decisions are driven in part by patient anxiety, worry, and uncertainty but also inexperienced or misinformed healthcare providers.^6-9 The second clinical management pattern suggests that patients with VUS receive medical management comparable to individuals with negative (B/LB) results or the general population, consistent with recommendations for VUS management, due in part to health care provider familiarity with guidelines.¹⁰ It is expected that as more healthcare providers become familiar with the uninformative nature of VUS and VUS-related clinical management guidelines, they will make more accurate recommendations. Much of the evidence supporting this expectation comes from studies of single cancer types conducted in academic hospitals or comprehensive cancer centers. Therefore, a systematic review and meta-analyses is needed to summarize all relevant individual studies of healthcare management in response to genetic test results to understand which treatment pattern is most prevalent and to make the available evidence readily accessible to decision makers.

The potential for erroneous management based on an increasingly common VUS genetic test result is a cause of concern in the field of clinical genomics. This concern is not trivial as clinical mismanagement and overuse of interventions such as surgery and screening can cause significant harm. In fact, the concern that VUS will compromise medical management is a commonly cited reason against the much debated population genetic screening initiative.¹¹ In addition to the potential for patient harm, assessing medical management trends among patients with VUS results may indicate gaps in healthcare provider and patient knowledge and opportunities for education and training, especially as genetic testing access expands among providers without genetics expertise. It is necessary to understand healthcare utilization patterns among patients with VUS results in comparison to patients with other types of genetic test results to inform debates about the role of VUS results in public health policies and population-scale initiatives.

The aim of this systematic review was to describe clinical management outcomes among patients with a VUS on germline genetic testing for cancer predisposition, and to compare these outcomes to those of patients with P/LP and B/LB results.

2 ∣. MATERIAL AND METHODS

2.1 ∣. Protocol and registration

We registered this review in the International Prospective Register of Systematic Reviews (PROSPERO #CRD42020180552). We reported our results following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis.¹²

2.2 ∣. Eligibility criteria

We included any type of observational study or randomized trial published in English that reported data on surgery or screening rates secondary to germline genetic testing for at least one hereditary cancer predisposition gene. The following types of articles were excluded: (a) absent outcomes from patients and at0risk individuals; (b) ineligible types of genetic tests (e.g., studies that tested patients for somatic, tumor, and other molecular tests only and molecular functional studies); (c) ineligible outcomes (i.e., articles that did not report VUS or in which VUS were reclassified); and (d) ineligible study designs (i.e., case reports, qualitative studies, comments, editorials, letters, reviews, systematic reviews, and meta-analyses).

2.3 ∣. Information sources and search

A research librarian (KJK) searched Medline, Embase and PsycINFO (through Ovid); and EBSCO CINAHL Plus, from 1946 through August 26, 2020. The Medline search strategy is provided in Table S1. Reference lists of pertinent reviews were manually searched by SM.

2.4 ∣. Study selection

Two authors (SM and EMB) screened all citations in a two-step process. Any disagreements were resolved through consensus. Titles and abstracts of 885 studies were first screened using the web app Rayyan.¹³ Next, 394 relevant citations were subjected to independent full-text assessment by SM and EMB followed by resolution of discrepancies through discussion.

2.5 ∣. Data extraction

Data were independently extracted by two researchers (SM and EB) using an extraction form. The authors checked each other's work; discrepancies were resolved through discussion. Items included study participants data (i.e., age, sex, personal history of cancer, number of study participants with cancer, affected organs, histological type, and stage), family history of cancer, reproductive factors (parity and menopausal status), and hormonal factors (oral contraceptive and hormone replacement use), study design, and risk of bias. Primary outcomes were rates of cancer-related surgery (mastectomy, lumpectomy, oophorectomy, hysterectomy, and gastrectomy) and screening (mammography, breast magnetic resonance imaging [MRI], transvaginal ultrasound [TVUS], cancer antigen 125 [CA-125] testing, and colonoscopy).

2.6 ∣. Risk of bias across studies

Risk of bias was assessed in accordance with the Newcastle Ottawa scale (NOS) for observational studies.¹⁴ Studies were scored on three domains: selection (four items), comparability (two items), and outcome (three items) for a maximum score of nine points for cohort studies and six points for cross-sectional studies. The total number of points for each applicable category was summed and categorized as: “low risk of bias” if the study received a point in that category, “unclear risk of bias” if the information was unavailable, or “high risk of bias” if the study received no points in that category.

2.7 ∣. Summary measures

We summarized dichotomous data, calculating odds ratios and their 95% confidence intervals (CI). We compared surgery and screening rates between participants with VUS and P/LP and B/LB using data provided in the article. If multiple indices of a given clinical outcome were reported, we used the most stringent and valid measure of clinical management available (e.g., surgical decisions made specifically in response to a genetic test result rather than surgery in general). When a study had more than one follow-up time point, we used data from the longest follow-up available (e.g., five-year screening rate rather than one-year screening rate). Reported screening rates according to participants' genetic test results were tabulated per anatomical site (i.e., breast, ovarian, and colorectal).

2.8 ∣. Synthesis methods

We performed a meta-analysis using a random-effects model when two or more studies reported surgery rates. We tested heterogeneity with the Chi-squared test and quantified it using the I² statistic, with a value of ≥50% representing substantial heterogeneity. Data analyses were conducted using Review Manager software (version 5.4, Cochrane Collaboration) and R studio software (version 3.5.1).

2.9 ∣. Additional analysis

Sensitivity analyses were used to evaluate heterogeneity by excluding one or more studies in which genes other than BRCA1/2 were included. We evaluated the effect of a genetic test result among participants with a personal history of cancer compared to those who did not however, there were not enough available data to conduct a meta-regression. A funnel plot was performed to assess publication bias and small-study effects in the meta-analyses for a primary surgical outcome analysis comparing VUS and P/LP results.

3 ∣. RESULTS

3.1 ∣. Study selection

Overall, 885 unique references were screened for possible inclusion, of which 394 underwent full-text review. Twenty-two studies met the criteria and were included in the systematic review; 16 were included in the meta-analysis. A flowchart of the study selection process is depicted in Figure 1.

FIGURE 1.

Search and screening process of studies included in the systematic review and meta-analysis

3.2 ∣. Study characteristics

Table 1 provides the characteristics of the included studies. Of the included 22 papers,17 reported surgical outcomes,^10,15-17,^19-21,^23-25,^27-30,³²,³⁴,³⁵ including mastectomy, lumpectomy, and oophorectomy, and six reported cancer screening outcomes,^{16,17,20,21,27,32} including mammography, breast MRI, TVUS, CA-125 testing, and colonoscopy. Studies were published between 2011 and 2020 representing data collected between 1995 and 2019.

TABLE 1.

Characteristics of included studies

Study	Country	Study timing	Setting and participants	Sample size	Study design	Genetic test results	Outcome assessment method
Carr 201815	USA	2013–2016	Clinical sample of women at high risk for breast cancer	163	Retrospective	P/LP, VUS	Chart review
Chang 201916	USA	2015–2018	Clinical sample of individuals who underwent genetic testing for breast cancer	563	Retrospective	P/LP, VUS, B/LB	Chart review
Chern 201910	USA	2006–2012	Clinical sample of women who underwent genetic testing for BRCA1/2	198	Retrospective	P/LP, VUS, B/LB	Chart review
Conley 202017	USA	2009–2015	Clinical sample of black women with breast cancer	143	Retrospective	P/LP, VUS, B/LB	Chart review
Culver 201318	USA	1997–2010	High-risk women who underwent BRCA1/2 genetic testing from a clinic or research registry	785	Prospective	VUS, B/LB	Chart review
Elsayegh 201819	USA	2014–2017	Clinical sample of breast cancer patients who underwent multigene panel testing	314	Retrospective	P/LP, VUS, B/LB	Chart review
Garcia 201420	USA	1995–2012	Women with BRCA mutations recruited from a community health system	374	Retrospective	P/LP, VUS	Chart review
Hamilton 201921	USA	2017	Participants from a research registry tested for CDH1	57	Cross-sectional	P/LP, VUS	Patient Survey
Hermel 201722	USA	2011–2015	Clinical sample of patients who underwent multigene panel testing	227	Retrospective	P/LP, VUS, B/LB	Chart review
Ho 201723	USA	2012–2016	Patients who underwent genetic testing for BRCA1/2 at a cancer center	484	Retrospective	P/LP, VUS, B/LB	Chart review
Howard-McNatt 201124	USA	2005–2009	Clinical sample of women with breast cancer who underwent genetic testing before surgery	110	Retrospective	P/LP, VUS, B/LB	Chart review
Idos 201925	USA	2014–2016	Clinical sample of high-risk individuals who were eligible for cancer genetic testing	1573	Prospective	P/LP, VUS, B/LB	Patient Survey
Kurian 201726	USA	2014–2015	Women with breast cancer identified through population cancer registries	3880	Retrospective	P/LP, VUS, B/LB	Patient Survey
Kurian 201811	USA	2014–2015	Women with breast cancer identified through population cancer registries	1316	Retrospective	P/LP, VUS, B/LB	Patient Survey
Lumish 201727	USA	2013–2015	Clinical sample of patients referred to a cancer genetics clinic for HBOC genetic counseling	232	Cross-sectional	P/LP, VUS, B/LB	Patient Survey
Morgan 201828	USA	2008–2017	Clinical sample of women tested for BRCA1/2 referred for genetic counseling	146	Retrospective	P/LP, VUS	Chart review
Murphy 202029	USA	2014–2019	Women with unilateral breast cancer who underwent genetic testing	1613	Retrospective	P/LP, VUS, B/LB	Chart review
Murray 201130	USA	1998–2009	Women who underwent BRCA genetic testing	107	Prospective	VUS	Chart review
Penderson 201831	USA	2013–2016	Clinical samples of triple-negative breast cancer patients who underwent genetic testing	447	Retrospective	P/LP, VUS	Chart review
Richter 201332	Canada	2008–2011	Clinical sample of women who underwent BRCA1/2 genetic testing	144	Cross-sectional	P/LP, VUS, B/LB	Patient Survey
Vos 200833	Netherlands	1998–2006	Clinical sample of women with breast and ovarian cancer with VUS results in BRCA1/2	49	Cross-sectional	VUS	Patient Survey
Welsh 201734	USA	2004–2016	Clinical sample of women with VUS in BRCA1/2	87	Retrospective	P/LP, VUS, B/LB	Chart review

Abbreviations: B/LB, benign or likely benign; P/LP, pathogenic or likely pathogenic; VUS: variant of uncertain significance.

3.3 ∣. Study participants

The 22 studies involved 13 012 participants (mean number of participants = 591, range: 49–3880) with mean age ranging from 45.0 to 60.3 years. Participants were predominantly adult non-Hispanic white (0%–94%) females (80.7%–100%). Participants' risk factors for hereditary predisposition to cancer were heterogeneous: all studies included patients with a diagnosis of cancer, 13 also included individuals without cancer, and 18 reported the presence or absence of a family history of cancer (family history was specifically defined in 12 of these studies). Most studies (n = 15) did not report hormonal and reproductive factors associated with cancer risk; others reported one or more of the following: parity (n = 6), menopausal status (n = 3), and oral contraceptive use (n = 2).

Most studies included participants who underwent multigene panel testing (n = 14) that often only included the BRCA1 or BRCA2 genes; other studies involved single gene testing (n = 6) or did not report the testing strategy (n = 2). Irrespective of the testing strategy, Figure 2 shows the frequency of genes included in the studies. Of the 22 studies, 14 described clinical outcomes from individuals with P/LP, VUS, and B/LB results, five described outcomes from individuals with P/LP and VUS results, one described outcomes from individuals with VUS and B/LB results, and two described outcomes from individuals with VUS results only. The temporality of genetic testing in relation to clinical management was clearly stated in nine of 22 studies; other studies provided the date or year of genetic testing or whether genetic testing occurred before or after cancer diagnosis.

FIGURE 2.

Frequency of cancer predisposition genes included in the studies

3.4 ∣. Risk of bias within studies

Twenty one of 22 studies were considered to have a high risk of selection, comparability or outcome bias due to insufficient information. As shown in Figure S1, only two studies provided information on loss to follow-up rate. Fifteen studies were judged with inadequate representativeness of the exposed cohorts and 15 of 17 studies did not have an adequate follow-up length. Table 2 shows the complete quality assessments of the studies included in the review.

TABLE 2.

Quality assessment outcomes by studies included in the systematic review

Selection	Selection				Comparability		Outcome
Study	Representativeness of exposed cohort	Selection of non- exposed cohort	Ascertainment of exposure	Demonstration that outcome of interest was not present at start of study	Adjust for most important risk factors	Adjust for other risk factors	Assessment of outcome	Follow-up length	Loss to follow- up rate	Total quality score	Total possible score
Cohort studies
Chern 201910		*	*	*	*	*	*	*	*	8	9
Culver 201318		*	*					*		3	9
Idos 201925		*	*							2	9
Murray 201130			*		*		*	*		4	9
Carr 201815		*	*							2	9
Chang 201916	*	*	*	*		*	*	*		7	9
Conley 202017	*	*	*	*	*	*	*			7	9
Elsayegh 201819		*	*	*		*	*			5	9
Garcia 201420	*	*	*	*		*	*			6	9
Hermel 201722		*	*	*	*	*	*			6	9
Ho 201723		*		*	*					3	9
Howard-McNatt 201124		*	*	*	*		*			5	9
Kurian 201811	*	*	*		*					4	9
Morgan 201828		*	*	*			*	*		5	9
Murphy 202029	*	*	*	*	*	*	*	*	*	9	9
Penderson 201831		*	*		*		*			4	9
Welsh 201734		*	*	*	*	*	*			6	9
Cross-sectional studies
Hamilton 201921		*	*	NA				NA	NA	2	6
Kurian 201726	*	*	*	NA	*			NA	NA	4	6
Lumish 201727				NA	*	*		NA	NA	2	6
Richter 201332		*	*	NA	*			NA	NA	3	6
Vos 200833		*	*	NA			*	NA	NA	3	6

Note: Stars indicate a point in that category.

3.5 ∣. Surgical outcomes

3.5.1 ∣. Therapeutic breast surgery

Five studies^{16,23,28,30,34} reported therapeutic mastectomy rates among breast cancer patients with VUS vs P/LP (Figure 3) and VUS vs B/LB genetic test results (Figure 4). Pooled data from three studies^16,23,28 showed no difference in mastectomy rates between patients with VUS vs P/LP results or among patients with VUS vs B/LB results.

FIGURE 3.

Primary surgical outcome analysis comparing VUS with P/LP results. P/LP, pathogenic or likely pathogenic; VUS, variant of uncertain significance

FIGURE 4.

Primary surgical outcomes analysis comparing VUS with B/LB results.VUS, variant of uncertain significance; B/LB, benign or likely benign

Therapeutic breast conservation or lumpectomy was reported in 5^{16,23,28,30,34} studies. Overall, breast conservation or lumpectomy rates were not significantly different between patients with VUS and P/LP results (OR = 0.47, 95% CI, 0.12–1.83; I² = 83%) or between patients with VUS and B/LB (OR = 0.84, 95% CI, 0.59–1.20; I² = 0%).

Seven^{15,19,24,26,29,31,34} studies reported rates of therapeutic mastectomy with contralateral prophylactic mastectomy (TM + CPM). A pooled analysis revealed that patients with P/LP were 7.5 times more likely to undergo TM + CPM than were patients with VUS results (OR = 7.53, 95% CI, 4.15–13.64; I² = 71%), whereas patients with VUS and B/LB results were not significantly different (OR = 0.94, 95% CI, 0.45–1.94; I² = 84%).

3.5.2 ∣. Prophylactic breast surgery

Nine^{11,15,16,18,20,21,28,30,32} studies reported prophylactic unilateral or bilateral mastectomy rates. A pooled analysis revealed that prophylactic mastectomy rates among individuals with P/LP results was three times higher than that in those with VUS results (OR = 3.05, 95% CI, 1.50–6.19; I² = 65%). There was no significant difference in the rates between individuals with VUS and B/LB (OR = 0.86, 95% CI, 0.56–1.30; I² = 36%).

3.5.3 ∣. Ovarian surgery

Seven^{10,16-18,20,30,32} studies assessed rates of risk-reducing oophorectomy. Pooled data across studies revealed that individuals with P/LP results were more than six times as likely to undergo oophorectomy than those with VUS genetic test results (OR = 6.46, 95% CI, 3.64–11.44; I² = 21%). There was no significant difference in oophorectomy rates among individuals with VUS vs B/LB results (OR = 1.45, 95% CI, 0.80–2.62; I² = 36%).

3.6 ∣. Screening outcomes

3.6.1 ∣. Breast cancer screening

The rate of breast cancer screening with mammography and breast MRI, stratified by genetic test results, were reported in three studies.^17,20,21 There was considerable variation in breast cancer screening rates and characteristics of study participants across studies (Table 3). Mammography rates did not differ regardless of genetic test results in the chart review study by Conley et al,¹⁷ which only included patients with a diagnosis of breast cancer, potentially because all breast cancer patients receive mammograms irrespective of their genetic test results. Mammography rates were higher among individuals with VUS (26.5%) than P/LP results (7.0%) in the chart review study by Garcia et al,²⁰ which included cancer patients and individuals recruited from the community setting. Breast MRI rates reported in these two studies were similar for individuals with P/LP, VUS, and B/LB results (~30%) but varied widely between studies (3% to ~30%). Hamilton et al²¹ presented patient self-reported outcomes, and found the highest reported breast MRI rates, with up to 69.2% among individuals with VUS results and 50.0% among individuals with P/LP results.

TABLE 3.

Cancer screening outcome stratified by P/LP, VUS, and B/LB genetic test results

Breast	P/LP (%)	VUS (%)	B/LB (%)
Conley 202017 (mammogram)	90.0	90.0	94.8
Garcia 201420 (mammogram)	7.0	26.5	NA
Conley 202017 (breast MRI)	30.0	33.3	32.8
Garcia 201420 (breast MRI)	3.0	3.0	NA
Hamilton 201921 (breast MRI)	50.0	69.2	NA
Hamilton 201921 (other)	62.5	56.1	NA
Ovarian
Conley 202017 (CA-125)	25.0	7.1	11.0
Garcia 201420 (CA-125)	2.0	16.2	NA
Conley 202017 (TVUS)	56.2	38.1	26.8
Garcia 201420 (TVUS)	2.3	17.5	NA
Colorectal
Chang 201916 (colonoscopy)	4.5	8.9	9.8

Note: % refers to the proportion of participants who underwent screening.

Abbreviations: B/LB, benign or likely benign; CA-125: cancer antigen-125; P/LP, pathogenic or likely pathogenic; TVUS: transvaginal ultrasound; VUS: variant of uncertain significance.

3.6.2 ∣. Ovarian cancer screening

Only two studies^17,20 reported ovarian cancer screening outcomes, including CA-125 and TVUS (Table 3). Screening rates varied widely between studies and across result types. Conley et al¹⁷ reported higher rates of CA-125 testing and TVUS among individuals with P/LP results than those with VUS, whereas Garcia et al²⁰ reported lower rates among those with P/LP.

3.7 ∣. Other high-risk screening and surgical outcomes

Four^21,25,27,33 studies surveyed participants who underwent genetic testing to evaluate surgery and screening rates. Vos et al³³ reported that no individuals with a history of cancer who perceived their VUS to be pathogenic, underwent prophylactic surgery, whereas 52.6% of those who correctly interpreted their VUS to be non-informative underwent prophylactic surgery. Idos et al²⁵ found that 16%, 3.3%, and 2.4% of individuals with P/LP, VUS, and B/LB results, respectively, reported undergoing “preventive surgery,” which included risk-reducing mastectomy, risk-reducing salpingooophorectomy, or hysterectomy. Lumish et al²⁷ surveyed patients with P/LP, VUS, and B/LB results and found that prophylactic surgery rates were 18.2%, 7.1%, and 10.6%, respectively, among cancer patients and 7.1%, 15%, and 10.1% among individuals with a family history of cancer. Hamilton et al²¹ reported that 12.5% of individuals with P/LP results in CDH1 underwent prophylactic gastrectomy compared with 0% with VUS.

Lumish et al²⁷ found more frequent or additional cancer screening in 9.1%, 21.4%, and 15.4% for with P/LP, VUS, and B/LB in cancer patients and 92.9%, 35%, and 15.9% in individuals with a family history of cancer, respectively. Richter et al³² reported higher rates of self-reported cancer screening (96.1%, 71.4%, and 83.3%). Colonoscopy rates were reported in one study,¹⁶ with no significant difference between VUS vs B/LB or P/LP genetic test results (4.5%, 8.9%, and 9.8%, respectively). Hermel et al²² reported a “minor change in patient management” among 4.5% of individuals with VUS.

3.8 ∣. Additional analysis

No differences were observed when we evaluated the effect of the risk of bias, participant selection, study design, or follow-up duration, except on the rates of breast conservation and lumpectomy. When we removed Chang 2019¹⁶ to focus on a more homogeneous sample of studies that only included BRCA1 and BRCA2, we observed a significantly lower rate of this surgical outcome among VUS than P/LP (OR = 0.24, 95% CI, 0.10–0.54; I² = 0%). Due to the small number of studies, funnel plots could not be performed for most comparisons. For the primary outcome of prophylactic unilateral/bilateral mastectomy the Egger test for the comparison P/LP vs VUS did not show evidence of small study bias (p value 0.88). Trends in clinical management over time could not be assessed as most studies analyzed data from patients seen over wide periods of time and did not report management rates by year.

4 ∣. DISCUSSION

The evidence in this review indicates that P/LP results are associated with higher use of cancer risk-reducing surgeries than are VUS results, whereas the surgical rates between individuals with VUS and B/LB results are not significantly different. There was insufficient evidence to draw a conclusion about the association between genetic test results and cancer screening outcomes. Individuals with VUS results in hereditary cancer predisposition genes undergo risk-reducing surgery at rates comparable to those of individuals with B/LB results, concordant with ACMG-AMP and NCCN guideline recommendations. Our results confirm previous reviews of the clinical impact of genetic test results³⁶ and expands the literature by adding cancer specific clinical management outcomes.

Although we conducted the review using rigorous methods to minimize the risk of bias, our review does have several limitations linked to the limitations of the available evidence. Principally, due to the high risk of bias of the studies included in this review, further research is likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. The studies included data from heterogeneous settings which largely comprised of well-resourced comprehensive cancer centers, with a limited data from cancer registries and none from small community or safety net hospitals. This substantive risk of bias and seemingly poor quality of many of the included studies suggests caution in interpreting the results.

The evidence supporting higher prophylactic surgical rates among individuals with P/LP results was the strongest for TM + CPM; six studies found comparably consistent effects, despite a wide 95% confidence interval. Since breast cancer patients with a BRCA1 or BRCA2 P/LP variant have up to a 60% risk of developing contralateral breast cancer,³⁷ which can be reduced by 90% with CPM³⁸ the finding of more TM + CPM among those with P/LP results than those with VUS is unsurprising. CPM is not recommended for those without a P/LP result because of the risks of associated complications and unclear survival advantage³⁹; however, CPM is also influenced by patient-specific factors, including cancer type and stage and the desire for breast symmetry.⁴⁰

Therapeutic mastectomy and lumpectomy rates were not significantly different by genetic test result. Given that these procedures are performed in individuals with cancer specifically for therapeutic purposes rather than risk-reduction, the decision to undergo these surgery types may be primarily influenced by the cancer diagnoses and medical comorbidities rather than genetic test results. For patients who have been already diagnosed with cancer, information about inherited cancer predisposition derived from germline genetic testing is of lower immediate importance than tumor genetic test results which can be used to guide therapy. Similar rates of therapeutic surgery among patients with different genetic test results could occur if patients delayed genetic testing until after surgery whereby management decisions would be agnostic to genetic test results⁴¹; however, the studies included in this review analyzed data from patients who underwent genetic testing before surgery. Oophorectomy rates were higher among individuals with P/LP than VUS results but not significantly different between individuals with VUS and B/LB results, consistent with guidelines.

Outcomes related to cancer screening strategies were reported in a small number of studies, with varying rates across studies. Overall, the available evidence does not provide support for the feared over-treatment among those with VUS results, although under-treatment among patients with P/LP is possible as not even they were 100% compliant with screening and nor did they have a high uptake of risk-reducing surgery. Future studies should investigate cancer screening among patients with P/LP, VUS, and B/LB results to clarify rates of uptake and identify opportunities to improve guideline-compliant care. Given the overall pattern of the combined evidence, large-scale prospective clinical studies and population-based studies using objective outcome measures are needed (e.g., Surveillance Epidemiology and End Results Program data linked to electronic medical records).

We conducted the review using rigorous methods to minimize the risk of bias and performed a quantitative synthesis using meta-analysis and a systematic assessment of bias risk of included studies and evidence quality by outcome. However, our review does have several limitations. Overall, studies were very heterogeneous with regards to participant characteristics (e.g., inclusion of participants with and without cancer), definition of management outcomes (e.g., therapeutic unilateral mastectomy vs mastectomy), and adjustment for clinical covariates, such as personal and family history of cancer and medical history risk factors which may be partly due to clinical diversity (patient population, setting, and when the study was performed). The substantive risk of bias and seemingly poor quality of many of the included studies, as well as the relative imprecision of the effect estimates, suggest the need for caution in interpreting the results. The heterogeneity of the studies also suggests a need for standard measures and definitions of therapeutic, risk-reducing, and elective surgical procedures and in how baseline cancer risk among the population of interest is measured and reported. The data that were necessary for further analysis, including information on family history of cancer, were sometimes lacking, and there were not enough studies to perform a meta-regression to account for possible methodological and statistical diversity. Publication bias is possible, as well-resourced, academically inclined medical centers with large genetic testing volumes are more likely to publish their clinical outcomes than are smaller community cancer centers or clinics. In addition, because of the lack of translation services, we only included articles published in English. The patient populations of the included studies predominantly comprised White women, who may have greater access to specialty healthcare services (such as genetic testing and preventive surgery) than other non-White patients and who may not be representative of all individuals who are at risk of cancer. All study designs and study qualities were included in the meta-analysis, which may have influenced our findings.

5 ∣. CONCLUSIONS

The available evidence and the results of our review do not indicate that a significant number of individuals with VUS results undergo inappropriate, non-guideline-compliant clinical management. As with any meta-analyses, the results of this study are sensitive to the quality of the base studies included in the review. Nonetheless, results from this study – on the accurate surgical management of VUS result – provides highly compelling evidence that can inform clinical guidelines and policy around medical genetics. Future research is needed to clarify the effects of hereditary cancer genetic testing results on cancer screening strategies. Researchers should define and implement standardized measures of surgical interventions and baseline cancer risks of participant populations to allow greater comparison and synthesis across research studies.

DATA AVAILABILITY STATEMENT

Data is available from the corresponding author upon reasonable request.