The recruitment of Malaysian families with a high risk of breast or ovarian cancer started in 2003 at the University Malaya Medical Centre in Kuala Lumpur. Patients with breast cancer were first approached by clinicians responsible for their care to see if they would participate in a research study to determine the genetic factors which increase the risk of breast cancer in Malaysia. Thereafter, individuals interested in the project were approached by a member of the research team who explained to them the nature and objectives of the research project. A total of 678 index cases were referred to our study between January 2003 and December 2007. Index cases signed a consent form and a blood sample was taken. A family history was recorded and the pedigree analysed. Where possible, pathology reports were requested to confirm all diagnoses of breast and ovarian cancers in the index case.

The study was approved by the ethics committee of University Malaya Medical Centre.

All breast cancer patients in the cohort had: early-onset breast cancer (≤ 40 years) and 1 or more additional cases of breast cancer in first- or second-degree relatives; breast cancer (≥ 40 years) and two or more additional cases of breast cancer in first- or second-degree relatives; bilateral breast cancer; or a personal or family history of ovarian cancer. They were analysed for mutations in the BRCA1 and BRCA2 genes. In addition, approximately 50% of patients with only early-onset breast cancer (≥ 40 years) with no significant family history or breast cancer (≤ 40 years) and one additional case of breast cancer in a first- or second-degree relative were also analysed. Individuals were categorised based on self-reported race or ethnicity. Of the 678 index cases, 187 were analysed for mutations in the BRCA1 and BRCA2 genes (Table 1).

Blood from index cases was separated into two 10 ml EDTA-tubes and DNA was extracted using standard methods. Samples were analysed either at Genetic Technologies Laboratory (Australia) or at Cancer Research Initiatives Foundation (Malaysia) using direct DNA sequencing and multiple ligation dependent probe amplification (MLPA), as previously described 18. Confirmation tests were performed on a second blood sample. Naming and interpretation of sequence analysis were performed as previously described and all patients were classified as having a deleterious mutation if the BRCA1 or BRCA2 protein terminated prematurely at least 10 or 110 amino acids, respectively, from the C terminus. Genetic variants of undetermined clinical significance (unclassified variants) included missense mutations and mutations that occurred in analysed intronic regions whose clinical significance had not yet been determined. If more than one variant was observed in a single analysis, the overall interpretation was that of the most clinically significant mutation observed.

The BOADICEA model allows for a polygenic modifier locus effect in which several low-penetrance genes have joint effects 19. Family history information (including incidence of breast, ovarian and other cancers, age at diagnosis and relationship to the proband) was collected and recorded by a genetic counsellor or researcher. Breast cancer patients were included in the BOADICEA analysis if they had more than one primary cancer or if they had a family history of breast, ovarian, pancreatic and/or prostate cancer in first-, second- or third-degree relatives. Of the 187 individuals who were analysed for mutations in BRCA1 and BRCA2, we excluded 42 individuals (including three individuals with deleterious BRCA2 mutations), all of whom had a single case of breast cancer and no family history of breast, ovarian, pancreatic or prostate cancer in first-, second- or third-degree relatives. The predicted likelihood of carrying either a BRCA1 or BRCA2 mutation was generated for each individual using BOADICEA risk estimation on the internet (BOADICEA WEB APPLICATION v1.0).

We used the Manchester Scoring System to estimate the probability of identifying mutations in BRCA1 and BRCA2 genes 2021. In brief, the scoring system was developed using a combination of results from screening and family history of those with and without mutations, where a combined score of 15 or more was proposed to correlate with a 10% mutation probability. The exclusion criteria for BOADICEA analysis was also applied to the Manchester Scoring System.

A total of 1226 patients with breast cancer were treated at the University Malaya Medical Centre between January 2003 and December 2007, of which 678 individuals with breast cancer from 674 families were recruited to the present study. Despite rigorous efforts to identify women with a family history of breast cancer, only 159 women were found to exhibit a family history in a first- or second-degree relative (129 with one additional affected member, 19 with two additional affected members and 11 with three or more additional affected members).

We conducted full sequence analysis of the BRCA1 and BRCA2 genes in 187 women (Table 1). Of these: 73 had no significant family history of breast or ovarian cancer in first- or second-degree relatives (45 of whom developed breast cancer ≤ 40 years old, nine of whom developed breast cancer between 41 and 55 years old; one had both breast and ovarian cancer, 16 had bilateral breast cancer and two had a family history of prostate cancer); and 114 had two or more cases of breast or ovarian cancer in first- or second-degree relatives.

Mutation detection led to the discovery of 27 deleterious mutations in 28 breast cancer patients (14 in BRCA1 and 13 in BRCA2; Table 2). Notably, 14 of these mutations were novel (five in BRCA1 and eight in BRCA2) and one mutation in BRCA2 was found in two unrelated families. In addition, we identified 47 sequence variants of unknown clinical significance (16 in BRCA1 and 31 in BRCA2), of which eight were found to occur in more than one unrelated family (Table 3). Of these, 26 sequence variants were novel (nine in BRCA1 and 17 in BRCA2). Eleven sequence variants (two in BRCA1 and nine in BRCA2) may be potentially damaging based on sequence conservation and Grantham score, and five sequence variants are unlikely to be clinically relevant.

Table 4 shows that the majority of mutations in the Indian subgroup were BRCA1 mutations, whereas the majority of the mutations in the Malay subgroup were BRCA2 mutations, and there were an equal number of mutations in both genes among the Chinese subgroup. One possible explanation for the higher incidence of BRCA1 mutations in the Indian subgroup is that there was a high proportion of women with both breast and ovarian cancer in this subgroup compared with the Chinese or Malay subgroups (14% of total, compared with 1 to 2% among the Chinese and Malay; Table 1) and it is clear from other studies that BRCA1 mutations confer a higher penetrance to ovarian cancer 22. By contrast, the reason for the higher prevalence of BRCA2 mutations among the Malays is unclear.

The mean age of diagnosis of all women analysed was 43.8 years (range 22 to 78 years), and the mean age of diagnosis was 40.3 years for BRCA1-positive (range 28 to 57 years), 43.6 years for BRCA2-positive (range 34 to 59 years) and 44.2 years for BRCA-negative women (range 22 to 78 years) (Table 5 and 6). This is comparable to data from a large study of 10,000 individuals in the USA 6, where individuals with BRCA1 mutations had a significantly younger age at diagnosis (40 years) than those with BRCA2 (41 years).

Table 5 shows the pathological characteristics of breast cancer arising in BRCA1 and BRCA2 deleterious mutation carriers. Consistent with other studies 23, the majority of BRCA1 tumours are high grade and negative for the oestrogen, progesterone and HER2 receptors. By contrast, the BRCA2 tumours arise with various status for the oestrogen, progesterone and HER2 receptor.

Table 7 shows the likelihood of having a BRCA1 or BRCA2 deleterious mutation, based on the personal and family history of the individual. We found that the best predictive factors for the presence of a BRCA1 or BRCA2 mutation are families with at least two cases of breast cancer, at least one of which occuring before the age of 50 (p = 0.0335), and families with a history of both breast and ovarian cancer (p < 0.0001). Notably, of the 15 breast cancer patients with personal or family history of ovarian cancer, six had mutations in BRCA1 and two in BRCA2. In addition, of the 14 BRCA1 carriers and 14 BRCA2 carriers identified in this study, five had bilateral breast cancer (one synchronous and four metachronous). The mean time interval between surgery for the first primary cancer and the second occurrence in the contralateral breast was 9.8 years (range four to 24 years). However, the association between bilaterality and BRCA mutation status was not statistically significant (p = 0.45). Taken together, the data suggests that the presence of two or more breast cancers with at least one case under the age of 50, and ovarian cancer at any age are significant predictors for the presence of a BRCA1 or BRCA2 mutation.

Using receiver operating curves (Figure 1), we evaluated the accuracy of using the Manchester Scoring System empirical method compared with the BOADICEA risk-prediction models to discriminate between those families that have a BRCA mutation and those that do not (Table 8). The classifications based on Manchester Score System and BOADICEA are similar for BRCA1, but significantly different for BRCA2, which BOADICEA did not predict accurately. In terms of discriminating between those with and without a mutation, the areas under the receiver operating curves, a common measure of the adequacy of a quantitative predictive algorithm, are 0.74 (95% confidence interval [CI] = 0.67 to 0.81) and 0.82 (95% CI = 0.75 to 0.88) for BRCA1, and 0.82 (95% CI = 0.75 to 0.88) and 0.56 (95% CI = 0.48 to 0.64) for BRCA2, for Manchester Score System and BOADICEA, respectively (Table 8).

The probability of identifying a mutation was 66% (96 of 145), 10% (15 of 145), 9.7% (14 of 145), 6.2% (nine of 145) and 7.6% (11 of 145), at ranges of scores 4 to 14, 15 to 16, 17 to 20, 21 to 24 and 25 and above by the Manchester Scoring System. Our analysis also indicated that a cut-off at a combined Manchester score of 15 seems optimal as a threshold (sensitivity of 72%, specificity of 74%, positive-predictive value of 37% and negative-predictive value of 93%). At a combined score of 18, the sensitivity was 56%, specificity was 84%, positive-predictive value was 42% and negative-predictive value was 90%.

For a combined score of 0.10 or above using BOADICEA, we observed sensitivity of 40%, specificity of 85%, positive-predictive value of 36% and negative-predictive value of 87%.

The total number of cases predicted by the Manchester Scoring System was eight BRCA1 mutation carriers, nine BRCA2 mutation carriers and 17 mutation carriers in total (using the probabilities listed 2021), and by BOADICEA was nine BRCA1 mutation carriers, five BRCA2 mutation carriers and 14 mutation carriers in total. This was compared with the observed numbers of 14 BRCA1, 11 BRCA2 and 25 total carriers (three BRCA2 carriers were among the 42 individuals excluded in the analysis because they had a single case of breast cancer and did not have any family history of cancer in the family). This suggests that both the Manchester Scoring System and BOADICEA methods underpredicted the probability of having a mutation in BRCA1 or BRCA2 in our cohort.

Of the 27 deleterious mutations identified, 13 mutations have never been previously reported in any other population. Two groups studying BRCA mutations in Singapore, where the population is also multi-ethnic but in different proportions (Chinese 77%, Malays 14% and Indians 8%), reported seven mutations in BRCA1 that were not observed in our cohort 24252627. One mutation, c.2845insA in BRCA1, was recently reported to have a founder effect among the Malay population in Singapore 252829, but this was not found in 44 Malays in our study.

We did not identify any mutations that occur with high frequency in Asians. Four mutations that have previously been reported in Asians: BRCA1 589delCT in at least four families from Southern China 30, BRCA1 1100delAT in families from Shanghai 16, BRCA2 4265delCT in at least three families from the Philippines 31 and BRCA2 2699del6 in two families from Indonesia 32, were also found in our cohort but these are unlikely to be common enough to warrant specific testing. Given the genetic diversity of the Asian populations, it is unlikely that screening for a panel of founder mutations will be as effective in this population as is the case for the Ashkenazi Jewish 33 or Icelandic 34 populations.

We found that BRCA2 may play an important role in genetic susceptibility among the small numbers of Malays in this cohort. Similar associations have been reported among the Filipinos and Japanese 3135. However, the reason for a higher proportion of BRCA2 mutations over BRCA1 mutations is not known, but may reflect cohort selection, genetic drift, or a possible prevalence of a modifying genetic or environmental factor that modifies the penetrance of BRCA1 or BRCA2 among Malays. Further studies using larger cohorts of Malay individuals are needed to address these possibilities.

In addition to the clearly deleterious protein-truncating mutations, a number of unclassified sequence variants were detected by sequencing. Of these, nine of 16 in BRCA1 (56%) and 17 of 31 in BRCA2 are novel (55%), and several of the remainder have only been described in Asian women. It is likely that the majority of these sequence variants have no clinical relevance, and the few that are likely to be deleterious are unlikely to change the basic conclusions of this study. We are conducting further analyses using established methods 363738 to better understand the clinical significance of these sequence variants in our cohort.

As BRCA1 and BRCA2 mutation testing is expensive, any ability to determine the probability that a specific family may benefit is an important issue. Several tools have been developed to help clinicians in predicting the probability of carrying a BRCA1 or BRCA2 mutation based on the familial history of breast and/or ovarian cancer and all tools have been based on populations where the age-standardised rate for breast cancer has been significantly higher than that reported in Malaysia (80 to 100 per 100,000 compared with 30 to 50 per 100,000 in Malaysia).

The Manchester Scoring System is perhaps the easiest of these prediction tools to use, and to determine whether the likelihood of identifying a mutation in a family reaches the 10% threshold for either BRCA1 or BRCA2. This model is does not require computer implementation. However, given that the Manchester Scoring System was devised using data from the Manchester region in northwest England, it is unclear whether it would be as effective in populations with different demographics, such as that in many Asian countries. Our analysis indicated that a cut-off at a combined Manchester score of 15 seems optimal as a threshold (sensitivity of 72%, specificity of 74%, negative-predictive value of 93%) but with a low positive-predictive value (37%). Notably, for the same cut-off of 15 among the Danish 39, the sensitivity was 84%, but the specificity was lower at 44%. The optimal cut-off appears to be lower for our population compared with studies in Australians 40 and French-Canadians in Canada 41, where comparable sensitivity and specificity (86% and 82% respectively in the Australians, and 72% and 64% respectively in the Canadians) were obtained with a higher score of 18. Taken together, our data suggests that in populations such as ours, where the overall incidence of breast and ovarian cancers is lower than that in Caucasian populations, family history is more significant therefore giving a higher specificity at a lower score.

In contrast, BOADICEA requires computation and is more difficult to implement in a clinical setting that is not familiar with risk prediction and genetic testing. Nevertheless, the strength of the model relies in its ability to compute extensive family history, and mutations in other genes relevant to breast cancer, such as CHEK2, to calculate the probability of carrying a mutation in the BRCA genes. In our analysis, a cut-off probability of 10% or higher yielded comparable specificity albeit with compromised sensitivity (85% and 40% respectively) to that reported in an analysis of 300 non-Ashkenazi Jewish, predominantly Caucasian families 42 (specificity 68% and sensitivity 73% respectively).

We found that both the Manchester Scoring System and BOADICEA underestimated the number of mutation carriers in our cohort. The most likely explanation, which is consistent with other studies, may be that both methods underestimate the number of carriers in families with insignificant family history 42, that is both methods underestimate the exceptional nature of the families, within the context of a lower population rate of breast cancer. In addition, the underestimate may be more significant for BRCA2 in BOADICEA because the allelic frequency of BRCA2 may be different in the ethnic population we have studied compared with Caucasian populations, and may be because BRCA2 may have a different penetrance in this population.

Overall, in this typically Asian cohort where 80% (149 of 187) of the breast cancer patients had no or only a single first- or second-degree relative with breast cancer or ovarian cancer, 15% of women carried deleterious mutations in BRCA1 or BRCA2. This suggests that a family history as reported by an individual who carries a mutation in BRCA1 or BRCA2 may be neither dramatic nor obvious. Indeed, in a study of 10,000 individuals in the USA, up to 13% of individuals with a single first- or second-degree relative with early-onset breast cancer or ovarian cancer, carried deleterious mutations 6. This suggests that careful evaluation of paternal as well as maternal family history is required, especially in women diagnosed with breast cancer before age 50 years or ovarian cancer at any age, to enable the appropriate identification and counselling of individuals at risk of carrying mutations in BRCA1 and BRCA2. We believe that identification of individuals at risk is particularly significant in Asia because the problem of limited family structure is not uncommon, in part because of loss of family information from migration, and there remains significant stigma in talking about cancer 9.