Stemming the Tide of Cancer for BRCA1/2 Mutation Carriers

Advances in molecular genetics during the past decade have resulted in the identification of numerous germline mutations in an increasing number of hereditary cancer syndromes, producing a sea change in the clinical approach to these disorders. The best known of these cancer-causing mutations are BRCA1 and BRCA2, which predispose to the hereditary breast-ovarian cancer syndrome. These mutations account for upwards of 10% of the entire breast cancer (BC) burden and are causal for approximately half of hereditary BCs. The salient clinical and pathologic differences between BRCA1 and BRCA2 BC are discussed at length by Atchley et al in this issue of Journal of Clinical Oncology.¹ In this commentary, we offer diverse thoughts on how recent advances in our understanding of the pathology and molecular functioning of BRCA1 and BRCA2 might point the way toward therapies for the corresponding BCs.

An important concern from a patient's perspective pertains to how a BRCA1 or a BRCA2 (BRCA1/2 or BRCA) mutation may impact their prognosis should they develop cancer. Physicians and researchers have been wrestling with data comparing the overall prognosis for women with BRCA-related BCs with their sporadic BC counterparts and whether treatment-related factors are also BRCA tumor–specific. If prognosis and treatment of BRCA-related BCs are indeed tumor-specific, the implications are considerable. Customizing treatment to prognosis is a key objective of oncologic care, and understanding how tumor profiles influence efficacy of treatment seems to characterize the main thrust of clinical oncology research in recent years.² To take it one step further, if treatment can not only be optimized but also targeted to the underlying biologic flaw, then characterizing tumor profiles at the time of treatment becomes paramount.

Genotypic and phenotypic heterogeneity must be given major consideration when evaluating treatment outcomes in hereditary breast cancer.³ For example, Gonzales-Neira et al⁴ performed a whole genome screen of 19 non-BRCA1/2 BC families using 4,720 genome-wide single-nucleotide polymorphisms with technology from Illumina Inc (San Diego, CA).

Atchley et al¹ discuss the importance of evaluating pathology differences between BRCA1 mutation carriers with BRCA-negative and those with BRCA2 mutations, first described more than a decade ago.⁵ We know that BRCA1 mutation carriers manifest a high frequency of estrogen receptor (ER) –negative and progesterone receptor (PR) –negative BCs, which limits antiestrogen hormonal therapy. Furthermore, when accompanied by HER-2/neu–negative findings, the so-called triple-negative immunophenotype, one frequently finds a uniformly poor response to most conventional forms of BC therapy.

The Atchley et al¹ study is limited by its relatively small size and the lack of central pathology slide review. The authors report only the nuclear grade, but not the tumor histologic type, which can also discriminate BRCA1 and BRCA2 BC phenotypes. Nevertheless, the clinicopathologic parameters that the authors do report are in basic agreement with other studies in the literature.^6,7

The BRCA1 pathophenotype features high-grade, so-called no special-type (NST, or ductal) carcinomas and an excess of medullary and atypical medullary special-type carcinomas.^5,8 BRCA1 BCs have pushing borders and significant lymphocyte and plasma cell infiltration, which are features also seen in medullary carcinomas.⁹ Like medullary carcinomas, BRCA1 BCs are highly proliferative as measured by mitotic grade in almost all studies, and this is also reflected in the high flow cytometric DNA S-phase fractions measured in the Creighton University series.^5,10 DNA cytometry also shows that BRCA1 BCs are more prevalently aneuploid than their sporadic BC counterparts.⁵ With respect to immunohistochemical markers commonly applied in diagnostic pathology laboratories, they are predominantly ER-, PR-, and HER-2 negative, as previously noted. Atchley et al¹ maintain that in their data set, HER-2 overexpression was similar in BRCA1 carriers and noncarriers (one of 38 v 38 of 267; P = .06). Although this P value is not formally significant at the .05 level, the power of the discrimination is limited by the small BRCA1 BC sample size and the low prevalence of HER-2–positive cases in all of the groups in this data set. With a larger BRCA1 BC sample, the trends would probably reach formal statistical significance. The authors’ results should thus be interpreted as consonant with most of the rest of the literature, which reports decreased HER-2 expression in BRCA1 BCs.^6,7 Other immunohistochemical features of BRCA1 BCs include increased p53 expression and dominance of the basal (myoepithelial) BC phenotype,¹¹ which is associated with the expression of markers such as cytokeratin 5/6 and P-cadherin.⁷

The BRCA2 BC phenotype is less well discriminated than the BRCA1 BC phenotype. Most larger studies find a later age of onset in BRCA2 BC compared with BRCA1 BC, but still considerably lower than the average age of onset in sporadic BC. The histologic characteristics of BRCA2 BC generally are reported to be similar to those in sporadic BC. In the Creighton University series, there seems to be an excess of tubular-lobular group special type carcinomas (invasive lobular, tubular, tubulolobular, and cribriform) in the 37 BRCA2 BC cases in the most recent update,⁶ but other studies do not confirm this finding.^6,7 The Breast Cancer Linkage Consortium⁸ found higher grade in BRCA2 BC as compared with sporadic BC, resulting more from poorer propensity to form tubules than to increased nuclear or mitotic grades. However, it should be noted that nearly half (49%) of the Consortium BRCA2 BC cases comprise the Icelandic 999del5 mutation. This mutation is remarkable for its association with very high grades,¹² which may not be typical for non-999del5 BRCA2 mutations and which could skew the data set. Genetic and ethnic correlations also suggest heterogeneity in the BRCA2 BC phenotype: there is more ovarian cancer associated with mutations in the central portion of the gene and with the Ashkenazi Jewish 6174delT mutation, whereas there is less ovarian cancer with BRCA2 carriers of French-Canadian ancestry.¹³ With respect to immunohistochemical markers, ER and PR expression in BRCA2 BC seems to be comparable to that in sporadic BC, whereas HER-2 may be the same or reduced and cyclin D1 increased.^6,7

Repeated observations of high grade in BRCA1 breast tumors parallel clinical observations of rapid tumor growth. Tilanus-Linthorst et al¹⁴ investigated tumor volume doubling time through magnetic resonance imaging or mammography in 100 patients with BC. Thirty-three patients were women with BRCA1 mutations, 16 patients had BRCA2 mutations, and 41 patients were at high risk in the absence of an identified mutation. Growth rate was decreased continuously with increasing age (P = .004); this is not surprising, as it is well recognized that BCs in younger women are more proliferative.¹⁵ However, Tilanus-Linthorst et al¹⁴ found that the growth rate was twice as fast in BRCA1 (P = .003) or BRCA2 (P = .03) mutation carriers as in other high-risk patients of the same age. Tumor size decreased with increasing age (P = .001). Median size was 15 mm for patients younger than 40 years old, compared with 9 mm in older patients (P = .003); tumors were largest in young women with BRCA1 mutations. These authors concluded that tumors grow quickly in women with BRCA1 mutations and in young women. Age and risk should, therefore, be taken into account in screening protocols. These enormous differences in BC's growth rate logically may strongly affect all BC treatment modalities (hormonal, chemotherapy, radiation).

Histopathology already has an important role in screening patients for Lynch's syndrome (hereditary nonpolyposis colorectal cancer). Syndrome colon cancers more frequently have tumor-infiltrating lymphocytes, Crohn's-like lymphocytic reaction, mucinous/signet ring differentiation, and medullary growth pattern. Indeed, the Revised Bethesda Guidelines¹⁶ incorporate these histologic patterns for consideration for testing colorectal tumors for microsatellite instability. The use of histopathology in screening patients with breast cancer for genetic counseling and BRCA1 and BRCA2 mutation testing perhaps is lagging behind its use in Lynch's syndrome, but we believe that the compelling phenotypic features of BRCA1 BCs, their increased frequency in younger women notwithstanding,¹⁵ should be factored into decisions for genetic counseling and testing of patients with BC.

The use of poly(ADP-ribose) polymerase (PARP) inhibitors for BRCA1- and BRCA2-deficient tumors is an elegant approach to targeted therapy, and the results of phase II trials in BRCA carriers with ovarian or BC are awaited. The strategy is based on sensitization of cells already deficient in DNA double-strand break repair, by virtue of an underlying BRCA1 or BRCA2 mutation, to the additional toxicity of base excision repair deficiency (a key pathway in the repair of DNA single-strand breaks) by PARP1 inhibitors. Again, development of this approach may benefit not only BRCA carriers, but the wider group of patients whose tumors are also defective in homologous recombination, including BRCA-like sporadic cancers.¹⁷

However, tumors may not all melt away with PARP1 inhibitors. The effectiveness of targeted therapies will also depend not just on the inciting germline event, but also on the level of resistance conveyed by the panoply of mutations present in cancer cells. PARP1 inhibitors may be most effective when used in combination with antitumor drugs, potentiating their toxicity.¹⁸ For BRCA1 and BRCA2 carriers, PARP1 inhibitors may work best as chemoprevention, before numerous defects have accumulated, exerting an effect at the moment of the second hit.¹⁹

The recent work of Saal et al²⁰ lends insight into the molecular details and characteristic nature of BRCA1 breast tumorigenesis. Their observation of basal-like features of mammary tumors on immunohistochemical staining in mice with Pten mutations spurred a series of PTEN studies in human BRCA1 BCs, which also are usually of the basal-like subtype. Loss of PTEN expression was associated with the basal-like BC (BBC) subtype in both sporadic and BRCA1-associated BCs.

Notably, the type of PTEN mutation in sporadic and BRCA1-associated BCs was qualitatively different; sporadic BBCs displayed coding mutations, whereas in BRCA1 BBCs, no sequence alterations were found. Instead, analysis of patient tumors, cell lines, and xenografts revealed that BRCA1-deficient BCs are associated with gross PTEN mutations involving intragenic chromosome breaks, inversions, deletions, and micro-copy number aberrations. Saal et al²⁰ proposed a theory whereby a BBC progenitor cell with a germline-mutated BRCA1 sustains a TP53 mutation, then undergoes loss of the wild-type BRCA1 allele (which would otherwise be lethal), leading to a double-strand break repair defect and subsequent loss of PTEN function and clonal selection. They hypothesize that this sequence of events may result in dependence on aberrant PTEN-PI3K pathway signaling. Therapy targeted to PTEN-PI3K and other pathways involved in BBC biology may then be an effective way to treat and possibly even prevent some sporadic and hereditary BBCs.²⁰

What is unique about BRCA-related BCs is that loss of function of a single, definable gene seems to not only predispose to a high lifetime risk of cancer but also influences the resulting tumor biology in a characteristic way. Several lines of data—histopathologic, immunohistochemical, gene expression profiling, and array-comparative genomic hybridization—have revealed a distinctive biology for BRCA1 and, to a lesser extent, BRCA2 BCs.^6,7,19 The mechanism by which BRCA germline mutations set off this sequence of events is unclear, but has been proposed to involve defects in DNA repair, DNA recombination, and cell cycle checkpoint control, which channel tumor evolution down a particular pathway.²¹ BRCA1 and BRCA2 defects cause similar clinical patterns that make the underlying mutations difficult to distinguish within individual families, and they play a role in overlapping biologic processes. However, the functions of the respective proteins seem to be distinct, which could result in a unique biologic profile for each gene and consequent treatment strategy.²¹

Variability in genetic modifiers or environmental factors that influence cancer penetrance could influence the cancer phenotype and response to treatment. Antoniou et al²² studied the manner in which BRCA1 and BRCA2 proteins interact with the RAD51 protein, which is also involved in cell cycle control and the G₂-M DNA damage checkpoint, homologous recombination, and double-strand break repair. The RAD51 gene contains an single-nucleotide polymorphism that has been suggested as a possible modifier of BC risk for BRCA1 and BRCA2 mutation carriers. They showed that an increased BC risk was statistically significant only among BRCA2 mutation carriers. Specifically, 135 224 C may modify risk of BC in BRCA2 mutation carriers by altering expression of RAD51. This is believed to be the first gene to be reliably identified as a modifier of risk among BRCA1/2 mutation carriers. Conceivably, these and other risk modifiers may contribute to the finer details of BRCA cancer phenotypes as observed by Atchley et al¹ and in other pathologic and molecular studies.

An intriguing theory that genes predisposing to hereditary cancers play a role in stem-cell fate has been proposed for BRCA1²³ and other²⁴ hereditary cancers. In the case of BRCA1 BC, this theory seems to lend a framework for several clinical and pathologic observations, including the frequently observed basal BC phenotype, also reminiscent of normal breast stem cells. Liu et al²⁵ investigated the role of BRCA1 in mammary stem-cell fate in in vitro and humanized mouse model systems and demonstrated that BRCA1 expression is required for the differentiation of ER-negative progenitor stem cells into ER-positive luminal cells. Loss of heterozygosity of BRCA1 was demonstrated in breast lobules of BRCA1 mutation carriers displaying stem-cell features, but not in adjacent lobules, suggesting that BRCA1 loss contributes to expansion of the undifferentiated progenitor stem-cell pool. This pool may then be prone to further carcinogenic events, both within BRCA1 germline mutated cells and those with somatic BRCA1 loss, resulting in the genetic evolution to aneuploid clones with relatively lower DNA indices, which we have found to be characteristic of BRCA1 cancers.^5,10 Interestingly, these studies may provide a biologic explanation for unresolved clinical paradoxes in BRCA1 carriers, namely, that prophylactic oophorectomy clearly reduces the risk of BC^26,27 and tamoxifen seems to have a chemopreventive effect for BC,^19,27 despite the ER/PR-negative status of BRCA1 BCs. Indeed, Watson et al²⁸ had earlier cautioned that the prevalent ER/PR negativity of BRCA1 BCs should not preclude consideration of hormone or receptor modulators in chemoprevention trials. This was due to the lack of evidence that the pretransformed target intermediate cell lacked receptors.²⁸ That evidence may now be at hand. One study suggests, in potential explanation, that there may be a paracrine-signaling effect of adjacent ER/PR-positive luminal cells.²⁵ Indeed, a parallel role of normal BRCA1 on suppression of aromatase expression, the rate-limiting step in estrogen production, could lead to upregulation of aromatase in BRCA1-deficient cells, providing such a paracrine effect from aromatase-laden stromal cells.²⁹

The results of Atchley et al¹ underscore that not all BRCA1 BCs are triple negative. Although there are many stereotypic features of BRCA1 BCs, and although it is tempting to think that these will pinpoint effective targeted therapies, BRCA1 and BRCA2 patterns and related treatments may not be that simple. For example, Fadare et al³⁰ suggest that although immunohistochemical and expression profiling definitions may be imprecise, there may be several basal BC profiles. In the case of BRCA mutations, the resulting phenotype may depend in part on the stage of differentiation at which BRCA function is lost.²⁵ If there is actually more than one BRCA1 and one BRCA2 profile, it will be important to define these more precisely.

The work of Atchley et al¹ and prior studies suggest that more efficient identification of BRCA carriers might be facilitated by immunohistochemical data, perhaps in addition to family history information. For example, a model incorporating ER status and CK14 and CK5/6 markers resulted in an area under the receiver operating characteristic curve of 0.87.³¹ In an analogous approach, a decision tree using age of BC diagnosis, Ki67, and epidermal growth factor receptor markers added to the level of certainty based on family history for classification of BRCA1 tumors.³² Classification of BRCA2 tumors is more difficult, but was accomplished using RAD51 and CHEK2 markers, effectively distinguishing BRCA2 from non-BRCA1/2 tumors with an estimated probability of more than 76%.³³ Gene expression profiling can differentiate BRCA1, BRCA2, and sporadic BCs, but the sensitivity and specificity of these methods have not been reported, and the ability to distinguish ER-positive BRCA1 BCs remain in question.³⁴ The practical clinical purposes of stratifying patients for gene testing and identifying candidates for targeted therapies require that a technique be applicable to archival materials, which prior gene expression profiling approaches are not, and that methods be relatively inexpensive. Alternatively, if the $1000 genome is imminent, then sequencing may be possible as a screen for patients with cancer (but genetic counseling issues are not obviated). However, there are still advantages to a phenotypic assay because DNA sequencing fails to identify approximately 10% of germline mutations. Furthermore, “BRCAness” will need to be identified phenotypically if targeted therapies become relevant for patients with sporadic cancer.³⁵

Mutation screening in hereditary breast-ovarian cancer syndrome for BRCA1/BRCA2 is becoming an increasingly important part of clinical practice. However, a vexing problem pertains to nontruncating sequence variants in these genes, wherein subtle changes may alter function in cells that predispose to cancer. In an enormous data set of nearly 70,000 full-sequence tests, 1,433 variants of unknown significance (VUSs) in the BRCA genes have been assessed for clinical significance at Myriad Genetics (Salt Lake City, UT).³⁶ A total of 133 VUSs had odds of at least 100:1 in favor of neutrality with respect to cancer risk; 43 VUSs had odds of at least 20:1 in favor of being deleterious. VUSs with evidence in favor of causality were those that were predicted to affect splicing, fell at positions that are highly conserved among BRCA orthologs, and were more likely to be located in specific domains of the proteins. It was concluded that there is abundant utility to improve genetic counseling of patients and their families.³⁶ Therein, the global assessment of these troublesome problems will be invaluable for validation of functional assays, structural models, and in silico analyses. A difficult question in accord with our treatment concerns is: How will VUSs differentially impact treatment of BC?

Spurdle et al³⁷ note that rare substitutions and in-frame deletions of BRCA genes pose a challenge to those harboring their unclassified variants (UVs). Using tumor immunohistochemistry markers, they evaluated the clinical, genetic, and evolutionary knowledge wherein these markers elevated the clinical significance of UVs. Their findings indicated that ER, cytokeratin 5/6, and cytokeratin 14 tumor expression, in concert with clinical relevance of amino acid evolutionary conservation, “may assist genetic counseling of individuals with unclassified sequence variants.”³⁷ Hopefully, these methods will help to resolve this very difficult problem of UVs in the genetic counseling setting.

Studies on the prognosis of BRCA BC versus sporadic BC have been inconsistent, with a worse prognosis found in some, but not all, studies.³⁸ Although there is a considerable amount of published data on the chemosensitivity of BRCA tumors, there is still no definitive word on the best treatment regimens for BRCA BC.³⁹ Unfortunately, the current limitations in knowledge about BRCA BC prognosis and treatment will likely continue until BRCA germline mutation status is prospectively taken into account either in well-powered studies of primarily sporadic BCs or in trials aimed at assessing BRCA tumor-specific treatment effects. A cogent argument has been made in the case of BRCA-associated ovarian cancers, which carry an improved prognosis, to prospectively incorporate germline mutation status in therapeutic trials.⁴⁰ Because approximately 11% to 30% of sporadic BCs have loss of BRCA1 function through promoter methylation^41,42 and approximately 44% to 88% of BRCA1-associated BCs are BBCs,³⁰ information gleaned from the study of BRCA mutation carriers will likely also benefit the much larger group of women with BRCA-like sporadic cancers. To study BRCA-related treatment effects and to guide the successful development of targeted therapies, efficient methods as well as processes to identify BRCA mutation carriers at the time of diagnosis must be devised.

Stemming the tide of cancer for BRCA carriers means not only elucidating the basic biology of these tumors and their relation to stem cells and unique phenotypes, but devising more efficient ways to identify carriers. The referral pipeline for BRCA1/2 testing is a slow drip when considering the number of carriers who do not know their mutation status and who thus remain at high risk of developing cancers that might otherwise be prevented or detected earlier. Moreover, as valuable as a positive family history is in identifying testing candidates, 43%⁴³ to 72%⁴⁴ of women with early-onset breast cancer unselected for family history, tested in British,⁴⁵ Spanish,⁴⁶ Dutch,⁴⁷ Australian,⁴⁴ Norwegian,⁴⁸ and Polish⁴³ populations and found to have germline BRCA1/2 mutations, have no family history of breast or ovarian cancer within one, two, or three degrees of relationship. Phenotype-based assays of incident cancers that identify BRCA1 and BRCA2 BCs, if these can be developed, would help identify candidates for genetic testing whose family histories are unrevealing.

What are the next steps for resolving dilemmas associated with BRCA prognosis and optimizing therapy? We believe such problems can best be resolved through conducting large-scale BRCA1/2 trials, working across institutions and industry. The broad network of genetic counseling professionals and researchers involved with BRCA families can surely support this goal. Focused attention on African American women, who face a disproportionately high rate of early-onset breast cancer with triple-negative features and significantly worse survival, would be a convergent goal.⁴⁹ The yield of such research would not only help improve the care of BRCA families, but moreover, should crystallize a better understanding of the etiology and pathogenesis of so-called sporadic breast cancer and, importantly, reduce the burden of this major public health problem.

AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a “U” are those for which no compensation was received; those relationships marked with a “C” were compensated. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.

Employment or Leadership Position: None Consultant or Advisory Role: None Stock Ownership: None Honoraria: Wendy S. Rubinstein, Myriad Genetics Research Funding: None Expert Testimony: None Other Remuneration: None

AUTHOR CONTRIBUTIONS

Conception and design: Wendy S. Rubinstein

Administrative support: Wendy S. Rubinstein

Data analysis and interpretation: Henry T. Lynch, Joseph N. Marcus, Wendy S. Rubinstein

Manuscript writing: Henry T. Lynch, Joseph N. Marcus, Wendy S. Rubinstein

Final approval of manuscript: Henry T. Lynch, Joseph N. Marcus, Wendy S. Rubinstein