Significant clinical impact of recurrent BRCA1 and BRCA2 mutations in Mexico

Abstract

Background

Frequent recurrent BRCA1 and BRCA2 gene (BRCA) mutations among Hispanics, including a large rearrangement Mexican founder mutation (BRCA1 ex9-12del), suggest that an ancestry-informed BRCA-testing strategy could reduce disparities and promote cancer prevention by enabling economical screening for hereditary breast and ovarian cancer in Mexico.

Methods

In a multistage approach, 188 cancer cases unselected for family cancer history (92 ovarian cancer and 96 breast cancer) were screened for BRCA mutations using a Hispanic mutation panel (HISPANEL®) of 115 recurrent mutations in a multiplex assay (114 on a mass spectroscopy platform, and a PCR assay for the BRCA1 ex9-12del mutation), followed by sequencing of all BRCA exons and adjacent intronic regions, and BRCA1 multiplex ligation-dependent probe amplification assay (MLPA) for HISPANEL negative cases. BRCA mutation prevalence was calculated and correlated with histology and tumor receptor status, and HISPANEL sensitivity was estimated.

Results

BRCA mutations were detected in 28% (26/92) of ovarian cancer cases and 15% (14/96) of breast cancer cases overall and 27% (9/33) of triple negative breast cancer. Most breast cancer cases were diagnosed with locally advanced disease. The Mexican founder mutation (BRCA1 ex9-12del) accounted for 35% of the BRCA-associated ovarian cancer cases and 29% of the BRCA-associated breast cancer cases. At 2% of the sequencing and MLPA cost, the HISPANEL detected 68% of all BRCA mutations.

Conclusion

In this study, we found a remarkably high prevalence of BRCA mutations among ovarian and breast cases not selected for family history, and BRCA1 ex9-12del explained one third of the total. The remarkable frequency of BRCA1 ex9-12del in Mexico City supports a nearby origin of this Mexican founder mutation and may constitute a regional public health problem. The HISPANEL presents a translational opportunity for cost-effective genetic testing to enable breast and ovarian cancer prevention.

Introduction

Approximately 5% of breast cancer cases and 10–15% of ovarian cancer cases are attributable to germline BRCA1 or BRCA2 (BRCA) mutations among non-Hispanic white women.^1–3 However, a lack of access to clinical BRCA gene analysis in some countries like Mexico has limited the implementation of prevention efforts and the scope of comparative studies of genetic factors that influence breast and ovarian cancer risk within Hispanic populations. Breast cancer patients in Mexico have shown an earlier age at onset of disease (<50 years)^4–6 and a high prevalence of triple negative breast cancer,⁵ both features suggestive of hereditary etiology and the possibility that deleterious BRCA mutations may account for a higher proportion of breast cancers in this population. Our studies demonstrating that BRCA mutations are prevalent among high-risk Hispanics in the United States support this hypothesis.⁷ A pattern of multiple recurrent mutations was observed in a large study (n=746) of Mexican-Americans,⁸ and this led to the development of an inexpensive Hispanic mutation panel (HISPANEL) assay as a screening strategy.

The purpose of this study was to use the HISPANEL and additional analyses to characterize the pattern and frequency of BRCA mutations in Mexican breast and ovarian cancer patients, unselected for family history, and to examine the association between BRCA mutations and breast/ovarian cancer phenotypes.

Materials & Methods

A blood sample was obtained after written informed consent from participants in two IRB-approved prospective clinical trials: one designed to assess response to cisplatinum in newly diagnosed advanced ovarian cancer patients between January 2008 and December 2012, and the other was a neoadjuvant treatment trial for patients with triple negative breast cancer or hormone receptor positive cases < 50 years old, between January 2005 and March 2008. None of the cases were selected for family cancer history. Histology was analyzed in ovarian cancer cases, and tumor receptor status was analyzed in breast cancer cases. The participants’ age at diagnosis and state of origin in Mexico were obtained, as was any family history of breast or ovarian cancer that was recorded in the medical record. DNA was extracted from each sample. A 114 BRCA mutation panel (HISPANEL) was developed based on our data from U.S. Hispanics,^{7, 8} other published data on BRCA mutations among Spanish, Hispanic or South American populations,^9–16 and entries citing Hispanic ancestry in the Breast Cancer Information Core (http://research.nhgri.nih.gov/projects/bic/Member/index.shtml). The assay was designed to detect insertions/deletions and point mutations on the Sequenom® (San Diego, CA 92121) MassARRAY platform (MALDI-TOF MS). Mutation positive samples cluster on the center axis of the Call Cluster Plot, while wild-type samples cluster at the corners. Quality control criteria were developed, including a requirement that >97% of the specified loci provided an unambiguous reading. Comprised of 5 multiplex assays and run in duplicate, the HISPANEL has a capacity of 154 samples per run, and robotics is used to load samples on the SpectroCHIP. DNA extraction, amplification and HISPANEL analysis can be completed within 72 hours from sample collection (data not shown). All samples were analyzed by PCR for the presence of the BRCA1 ex9-12del Mexican founder mutation, as described.¹⁷

In addition, complete pyrosequencing of all BRCA1 and BRCA2 translated exons and adjacent intronic regions was performed on all breast cancer cases and all HISPANEL-negative ovarian cancer cases using the BRCA MASTR Dx kit from Multiplicom (Belgium) on the Roche 454 GS Junior. Finally, all HISPANEL-negative and sequencing-negative cases were analyzed for large deletions/duplications in BRCA1 by multiplex ligation-dependent probe amplification (MLPA) (MRC-Holland, Amsterdam).

Results

The sample consisted of 92 ovarian cancer and 96 breast cancer cases from the Instituto Nacional de Cancerologia (INCan). The mean age of diagnosis of ovarian cancer was 53 years (range 23 to 83 years) and of breast cancer was 40 years (range 26 to 83) (Table 1). At least 67% of breast cancer cases were diagnosed with locally advanced disease (stage III). Most cases (30%) originated in the Distrito Federal (DF) or nearby regional states (92% in Central States, including the DF). A family history of breast or ovarian cancer was reported in the medical record in less than 10% of cases.

Table 1.

Demographics

Breast Cancer n=96			Ovarian Cancer n=92
Age
Range		26–63	Range		23–83
Mean		40	Mean		54
Stage
I		1%	I		0%
II		27%	II		0%
III		67%	III		53%
IV		2%	IV		42%
Unknown		3%	Unknown		5%
Histology
Ductal		81%	Serous		53%
Lobular		12%	Adenocarcinoma-nos		25%
Ductal/Lobular		3%	Mucinous		1%
Other		2%	Carcinosarcoma		1%
Unknown		1%	Mixed/Other		16%
Cancer Subtype
Hormone receptor+, HER2-		26%	-		-
HER2 positive		39%	-		-
Triple negative		34%	-		-
Unknown		1%	-		-
State of Origin
Chiapas		2%	Chiapas		3%
Distrito Federal		39%	Distrito Federal		21%
Guanajuato		25%	Guanajuato		1%
Guerrero		1%	Guerrero		0%
Hidalgo		10%	Hidalgo		10%
Jalisco		0%	Jalisco		1%
Michoacán		2%	Michoacán		5%
Morelos		7%	Morelos		7%
Oaxaca		3%	Oaxaca		1%
Puebla		1%	Puebla		9%
Queretaro		2%	Queretaro		0%
State of Mexico		1%	State of Mexico		3%
Tlaxcala		4%	Tlaxcala		28%
Veracruz		0%	Veracruz		1%
Zacatecas		1%	Zacatecas		3%
Unknown		1%	Unknown		1%
Region in Mexico
Central		92%	Central		85%
North		1%	North		3%
South		6%	South		5%
Unknown		1%	Unknown		1%
Family History
Breast cancer	Yes	6%	Breast cancer	Yes	7%
	No	94%		No	83%
Ovarian cancer	Yes	1%	Ovarian cancer	Yes	0%
	No	99%		No	90%
Other cancer	Yes	1%	Other cancer	Yes	2%
	No	99%		No	88%

Overall, a BRCA mutation was detected in 21% (40/188) of the cases (Table 2). BRCA mutations were detected in 28% (26/92) of ovarian cancer cases and 15% (14/96) of breast cancer cases. Representing 33% (13/40) of BRCA mutations overall (35% and 29% for ovarian and breast cancer, respectively), the Mexican founder mutation BRCA1 ex9-12del was detected in 7% (13/188) of these unselected breast and ovarian cancer patients.

Table 2.

BRCA mutations detected among 188 Ovarian/Breast Cancer Patients.

Mutation		Ovarian cancer (n=92)	Breast cancer (n=96)	Total (n=188)
BRCA1 (85%)	ex9-12del	9 (35%)	4 (29%)	13 (33%)
	IVS5+1G>A	2	0	2
	3977del4*	0	1	1
	R1699W*	1	0	1
	803delA*	1	0	1
	70insAG*	1	0	1
	A1708E	1	1	2
	4184del4	1	0	1
	R71G	1	0	1
	917delTT	0	1	1
	943ins10	1	0	1
	2925del4	0	1	1
	3878delTA	0	1	1
	185delAG	0	1	1
	R1443X	0	1	1
BRCA1 Large Rearrangements (BRCA1)**	ex8-9dup	2	0	2
	ex18-19del	2	0	2
	ex8-10del	1	0	1
BRCA2 (15%)	9463delG	1	0	1
	6244delG*	1	0	1
	2900delCT*	1	0	1
	6714del4*	0	1	1
	1803insA*	0	1	1
	6252insG	0	1	1
Total		26 (28%)	14 (15%)	40 (21%)

^*Mutations detected by pyrosequencing;

^**detected by MLPA

Papillary serous carcinoma (n=49) was the most common ovarian cancer histology and had the highest prevalence (31%) of BRCA mutations, as shown in Table 3. None of the 3 cases with unknown histology had a BRCA mutation. The majority of mutations were BRCA1 (88%) and just three (12%) were BRCA2. BRCA1 ex9-12del accounted for 9 of 26 (35%) BRCA-associated ovarian cancers. Including the 5 BRCA1 mutations detected by MLPA, 14/26 (54%) mutations detected in the ovarian cancer cases were large rearrangements, not detectable by sequencing. Most breast cancer cases, (88/96; 92%) were younger than 50 years old, and 11/88 (13%) cases harbored a BRCA mutation (8 BRCA1 mutations and 3 BRCA2 mutations). Estrogen receptor/progesterone receptor/Her2 neu-negative breast cancer (triple negative breast cancer) represented 33/96 (34%) of the breast cancer cases, and 9/33 (27%) carried a BRCA1 mutation (Table 4). BRCA1 ex9-12del accounted for 3/9 (33%) BRCA-associated triple negative breast cancers.

Table 3.

Frequency of BRCA mutations in patients with ovarian cancer by tumor histology.

Category Tumor Histology	No. BRCA Positive			Total No. of cases*
	BRCA1	BRCA2	Total
Serous	15	0	15	49
Carcinosarcoma	0	1	1	1
Adenocarcinoma-nos	4	1	5	23
Mucinous	1	0	1	1
Mixed/Other	3	1	4	15
All tumors	23	3	26	89

^*Excluded 3 cases with unknown tumor histology

Table 4.

Frequency of BRCA gene mutations according to triple negative breast cancer (TNBC) status.

ER/PR/Her2 TNBC vs. non (any+)	No. (%) BRCA positive			Total No. of cases*
	BRCA1	BRCA2	Total
TNBC cases	9	0	9 (27%)	33
Non-TNBC	2	3	5 (8%)	62
All cases	11	3	14 (15%)	95

ER: estrogen receptor, PR: progesterone receptor, TNBC: triple negative breast cancer.

^*Excluded one case with unknown tumor receptor status.

The HISPANEL detected 77% (27/35) of BRCA mutations compared to sequencing, and 68% overall (including the 5 additional large rearrangement mutations detected by MLPA among the ovarian cancer cases).

Discussion

Few studies have assessed the prevalence of BRCA mutations among Mexican cancer patients, and all included a limited number of high-risk cases.^18–20 The current study demonstrates a remarkably high prevalence of BRCA mutations among breast and ovarian cancer cases not selected for family history (15% and 28%, respectively) in Mexico, which is higher than expected compared to previous data obtained from a population-based breast cancer registry series in the U.S., where the BRCA1 prevalence was 3.5% among Hispanics (n=393) aged <65, and 8.9% for patients <35 years.²¹ Only our recently published high-risk clinic-based population (n=746) study had a higher prevalence (25%).⁸

Similarly, our observation of BRCA mutations in 28% of the ovarian cancer cases not selected for family history is much greater than the 18% prevalence reported for a series of 360 unselected largely Caucasian ovarian cancer cases from the University of Washington, Seattle,²² or the 13% observed in an Ontario population-based study.¹ Similar to other reports, papillary serous carcinoma was the most common ovarian cancer histology associated with BRCA mutations.²³

BRCA mutations accounted for 13% of the breast cancers < 50 years in our study, which represented 92% of the cases. This is particularly relevant in Mexico where the mean age of breast cancer diagnosis is 50 years.^{24, 25} Moreover, 23% of incident breast cancer cases present at an age younger than 45 years,²⁶ which is the threshold for recommending BRCA testing in the National Comprehensive Cancer Center Network Guidelines.²⁷

We found that a higher proportion (27%) of triple negative breast cancer patients carried a BRCA1 mutation than previous reports (11–20%).^28–30 In a highly selected series of hereditary cancer clinic patients, BRCA1 prevalence was 20% among Hispanic triple negative breast cancer patients.³¹ The observed prevalence of BRCA mutations in this study may explain in part the higher proportion of triple negative cancer in Mexican patients compared to Caucasians.⁵

It is noteworthy that 67% of the breast cancer patients in our study had locally advanced breast cancer at diagnosis. While the prevalence of younger triple negative breast cancer cases in our study might have influenced the stage distribution, it has been previously reported that 56 to 80% of breast cancer cases present at advanced stages in Mexico.³² The high prevalence of BRCA-associated disease has significant clinical implications. It is conceivable that the use of genetic cancer risk assessment with BRCA testing and application of risk-appropriate screening and prevention interventions could result in shift toward more limited stage disease and reduced breast and ovarian cancer incidence. There was a remarkable prevalence of BRCA1 ex9-12del, which constituted 29% and 35% of all mutations for breast cancer and ovarian cancer, respectively, in this Mexico City regional sample. If the founder mutation is omitted, the prevalence of BRCA mutations in the breast and ovarian cancer cases was 10% and 18%, respectively, which are closer to prevalence estimates for BRCA-associated disease in other populations.^{1, 22} This Mexican founder mutation (BRCA1 ex9-12del) comprised ~10% of BRCA1 mutations in a Mexican American sample.^{8, 17} An independent report from a commercial vendor in North America noted that it comprises 1/3 of all large rearrangements in Latin American/Caribbean patients.³³

BRCA1 ex9-12del is clinically significant and one of the most frequent population-specific large rearrangement mutations in the world and is not detectable by standard BRCA gene sequencing approaches. We have speculated that the Mexican founder mutation, estimated to have first arisen 74 generations or 1,480 years ago (95% CI, 920 to 2,260 years),⁸ may have originated around Puebla. Puebla is in close proximity to the study center in Mexico City, which serves as the main referral center from patients from Central and Southern Mexico. Supporting this hypothesis is our observation that the Mexican founder mutation represented 33% of the BRCA-positive cases at the study center, and just 10% of the BRCA1 mutations among a heterogeneous Mexican-American population.⁸ We believe this mutation is a significant public health issue and that Mexican high-risk patients should be screened for its presence. Interestingly, the Jewish founder mutation, BRCA1 185delAG, was only detected once (2.5% of BRCA mutations) in the current study, compared to 10% of all BRCA mutations detected in the Mexican-American high-risk clinic study.⁸ Since Mexico is a setting with wide variability in dietary intake, lifestyle and genetic admixture (large contrast between northern and southern parts of Mexico),^{34, 35} further studies of the genetic epidemiology of breast cancer in different areas of Mexico are needed. Cost has been a barrier to genetic cancer risk assessment access among Hispanics, with access particularly limited in low- and middle-income countries such as Mexico. The observation of recurrent BRCA mutations among Hispanics led to the development of the HISPANEL as an economical BRCA screening platform. The HISPANEL detected 27 of 40 BRCA mutations (68%) at a cost of ~$25 USD/assay; in contrast with pyrosequencing, which detected an additional 8 mutations (Table 2) at a cost of $1,500/assay, and MLPA with the identification of 5 additional mutations at a cost of $50/assay. Thus, the HISPANEL detected 27 mutations for $4,700, while full sequencing cost $241,500 to detect an additional 8 mutations. We demonstrate here that the HISPANEL, which includes a specific assay for the Mexican founder mutation, is 77% sensitive compared to full sequencing (68% compared to sequencing and BRCA1 large rearrangement screening). Although this level of sensitivity is less than optimal for a screening tool, this is by far better than not having access at all for genetic testing, since at the moment testing is not available or covered within the public health system in Mexico, which provides care to more than 85% of the population. Unless and until next generation sequencing(NGS)-mediated clinical grade testing, with complete sequencing and rearrangement studies, costs substantially less than the currently cheapest available commercial testing (~$900), it will not be affordable in the context of competing priorities and needs in the Mexican health care system. We are optimistic that the rapidly evolving NGS technologies will drive the price low enough to obviate even the economical HISPANEL. However, as stated by Voltaire, “The perfect is the enemy of the good,”³⁶ so in the interval, implementing testing with the HISPANEL has the potential to enable the prevention of many cancers and save the lives of women in these families now, while catalyzing the integration of genetic cancer risk assessment in clinical practice in Mexico. We believe the benefits of its implementation in current practice will rapidly enable preventive and therapeutic interventions for patients and families. The patients identified as BRCA mutation carriers in the current study and in other ongoing projects will receive risk-appropriate follow-up care, including risk reduction salpingo-oophorectomies (RRSO) for cancer patients who are identified to have high-risk for additional primary breast and ovarian cancer diagnoses. RRSO is a relatively economical procedure³⁷ that has been documented to decrease new primary breast and ovarian cancer risk, and decrease all cause mortality.^{38, 39}

Limitations of the current study include the possibility of potential ascertainment bias based on the fact that INCan is a National referral center, though 92% of the cases resided in the central states, including the DF. Median age of breast cancer diagnosis in our study was 40 years, in comparison with the general age of diagnosis among Mexican women (50 years), which may partially explain the mutation prevalence.⁴⁰ BRCA2 large genomic rearrangement screening was not performed, so some mutations could have been missed. However, most studies indicate that >80% of all large genomic rearrangements are in BRCA1, given the high concentration of Alu sequences.^{33, 41, 42} Another limitation is that family history was abstracted from the medical record in this study, and this is generally an inadequate source of family history information, typical of a non-genetic service consultation.^{2, 43} Lack of detailed family history did not permit the application of mutation probability models. However, the high prevalence of BRCA mutations found in this study, regardless of family history, suggests that genetic testing should be guided more by early age at onset and/or cancer type (e.g., triple negative breast cancer or ovarian cancer) rather than by family history, and that risk assessment should be established for other identified risk factors. The frequency of BRCA mutations might be even higher among high-risk populations selected for family history of cancer.

Conclusion

In this study, we found a remarkably high prevalence (28%) of BRCA mutations among ovarian and breast cases not selected for family history, and the BRCA1 ex9-12del Mexican founder mutation explained one third of the total. Both in the US Hispanic populations and in Mexico, the Mexican founder mutation prevalence makes it a significant public health issue.

The HISPANEL, which includes recurrent BRCA mutations found in women of Hispanic ancestry, appears to have high sensitivity and thus is likely to have clinical utility while dramatically reducing overall genotyping cost among underserved women in Mexico and presents an opportunity for cost-effective genetic testing strategies to enable breast and ovarian cancer prevention. Further studies are needed to validate the sensitivity of the HISPANEL in different areas in Mexico and among U.S. Hispanics, and to explore the geographic population distribution of BRCA1 ex9-12del.