Validation of a Blood-Based Gene Expression Test for the Detection of Breast Cancer Using an Independent European Cohort

We report the development of a gene expression based blood test to detect breast cancer (BC). The test measures RNA species that undergo regulation as a response to the presence of a breast cancer lesion. We have previously presented data from a European/US cohort with accuracies from 75% - 82%. We now report findings from a multi-centre study where a calibration set of 282 subjects was used to create a model for the classification of breast cancer vs non-breast cancer using a European/US cohort. Validation of this model was performed using an independent set of 109 samples collected in 5 sites one of which was independent from the calibration set.MethodsIn a multicentre-study, blood samples were collected from women recruited in 4 groups 1) early stage BC (stage 0-1), 2) late stage BC (stage2+), 3) benign breast lesions, 4) without abnormal mammographic findings. Samples were collected in PAXgene™ tubes and shipped to a central laboratory where RNA extraction and quality control were performed. Gene expression analysis was performed using real time-RT PCR on an AB Prism 7900HT Fast instrument with a microfluidic card containing a BC-specific gene signature in a 96-gene assay format. Modelling was performed using Partial Least Square Regression providing an algorithm for application to gene expression data. Application of the algorithm to subjects in an independent cohort was used to obtain a test score for each subject. A positive test score classified a subject as positive for breast cancer, whilst a negative score classified a subject as negative for breast cancer. A total of 248 samples were used to develop the algorithm and 109 used as an independent validation set to describe it performance.ResultsThe model correctly predicted the class of 78/109 validation samples, resulting in an overall accuracy of 72%. Performance was similar for early and late stage cancer with a sensitivity of 74% for stage 0 and stage 1 breast cancer. No significant difference in diagnostic performance was seen between pre- (71%) and post-menopausal (70%) women indicating the clinical value of our test in younger women. When assessing the efficacy of ductal and lobular cancers only the test showed a sensitivity of73%. The effect of medications were assessed from the information gathered from the subject report forms. The majority of medications were for hypocholesterolaemia, hypertension, diabetes, and asthma. None of the medication groups noted appeared to have a detrimental effect on the ability for the test.ConclusionWe report the validation of the blood-based gene expression test in a European cohort from a multi-centre study. The test shows efficacy for the detection of early breast cancer for both lobular and ductal breast cancer in both pre- and postmenopausal women. The test may be of clinical benefit as an adjuvant to mammography particularly for pre-menopausal women where mammography is known to have limitations.

Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 3025.

TP53 mutation is an early event in breast cancer progression

Abstract #1047

Background: In breast cancer, previous studies have shown that somatic TP53 mutations cause a more aggressive disease with poor clinical outcome and may impact treatment response. Although TP53 mutation is considered to be an early event in breast cancer, the timing of TP53 mutations is not known, and there are controversies regarding the cellular origin and linear model of breast cancer. The purpose of this study was to investigate whether TP53 mutations are early events in breast cancer progression.
 Methods: From a population-based cohort of women diagnosed between 1986 and 2004 either with a pure ductal carcinoma in situ (DCIS), a pure invasive cancer (<15mm) or a mixed lesion (i.e. invasive cancer with a DCIS component), we included 118 women with stored frozen tissue. Mixed lesions were microdissected using LCM (laser capture microdissection) on a PALM slide to separate in situ and invasive tumor cells. DNA was isolated using phenol-chloroform extraction. The entire coding sequence of TP53 was analyzed for mutations by direct sequencing on a 3730 DNA analyzer.
 Results: Of 118 tumor samples, 19 were detected with a TP53 mutation; five 5 of 32 (15.6%) pure DCIS, 4 of 38 (10.5%) pure invasive cancers and 10 of 48 (20.8%) mixed lesions. In the mixed lesions, both the invasive and the DCIS component showed the same mutation in all 5 cases where we successfully could microdissect the two components separately. Pure DCIS demonstrated missense mutations (4/5, 80%) more frequently than pure invasive cancers (2/4, 50%) and mixed lesions (4/10, 40%), although this difference was not statistically significant (p=0.3). Also, the frequency of missense mutations in the DNA binding domain was not statistically different between the three groups.
 Conclusion: TP53 mutation is likely an early event in breast cancer, occurring previous to or in the in situ stage. Presence of the same mutation in both DCIS and invasive component from the same tumor indicates same cellular origin. The role of mutant TP53 in the progression into invasive cancer is less clear and may vary between subtypes of breast cancer.

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 1047.

A functional SNP in the leader sequence of the manganese superoxide dimutase gene is a predictor of breast cancer patient survival and response to chemotherapy

Abstract #6067

Background: A functional single nucleotide polymorphism (SNP) in the Manganese Superoxide Dismutase (SOD2) gene, characterized by a T-C transition located at codon 16, leads to a Val to Ala (Val16Ala) switch. The Val (T) allele disrupts the alpha-helix structure of SOD2 and causes the protein to be retained at the level of the mitochondrial inner membrane, with 30 to 40% lower activity and increased susceptibility to oxidative stress. We investigated whether the Val16Ala SOD2 polymorphism was associated with breast cancer survival.
 Methods: Two independent populations, one from the United States (n=244) and from Norway (n=329) were genotyped for the Val16Ala SOD2 polymorphism. Kaplan-Meier survival and multivariate Cox proportional hazards regression analyses were used to examine the relationship between the polymorphism and disease-specific survival, and response to chemotherapy.
 Results: We observed an association between Ala (C) allele and an increased risk of disease specific mortality. Carriers of the Ala/Ala genotype had an increased risk of poor survival in the combined analysis (Greater Baltimore and Norwegian datasets combined) when compared to carriers of Val/Val genotype using multivariate Cox regression analysis [Hazard ratio (HR) = 1.89; 95% confidence interval (CI), 1.16–3.07; p=0.011]. In the stratification analysis, the association was significant in those who underwent chemotherapy (HR = 2.54; 95% CI, 1.44–4.47; p=0.001), compared to those who did not (HR = 0.55; 95% CI, 0.16–1.94; p=0.353). Analysis by type chemotherapy received, showed that the Ala/Ala genotype had the strongest effect in patients receiving cyclophosphamide containing chemotherapy (HR = 18.88; 95% CI, 4.17–85.37; p<0.001). Other drugs with an effect included 5-fluorouracil and doxorubicin. These results suggest that the Val16Ala SOD2 polymorphism may be a predictor of response to chemotherapy.
 Conclusion: We found that the Val16Ala SOD2 polymorphism, was associated with breast cancer survival, and is a potential predictor of response to chemotherapy.

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 6067.

Development of a blood based breast cancer test for Indian population

Abstract #5013

Background: The incidence of breast cancer (BC) in India is increasing as the society becomes more westernized. No national screening program exists and there is a high mortality rate which may be linked to the late detection of the disease. Here we report the development of a gene expression based blood test to detect BC. We have previously presented data from European/US cohort with accuracies ranging between 75% - 82% suggesting that test can be developed to early detect breast cancer. We now report the findings from a multi-centre study initiated to test the efficacy of the test for early detection of breast cancer (BC) in an Indian population.
 Methods: A multicentre-study was initiated where blood samples were collected from women recruited in 5 different groups 1) early stage BC, 2) late stage BC, 3) women at high risk - without BC, 4) benign breast lesions, 5) without abnormal mammographic findings. The women recruited in group 3 is a part of ongoing prospective study. Recruitment was balanced between pre- and post-menopausal women among the remaining groups. Samples were collected in PAXgene tubes and shipped to a central laboratory where RNA extraction and quality control. Gene expression analysis was performed using TaqMan® low density arrays (LDA's) containing a BC-specific gene signature in a 96-gene assay format. A total of 442 samples from groups 1, 2, 4 and 5 were used to develop the India model and estimate its prediction efficacy. The 442 samples were divided between a training cohort (N=292) and a test cohort (N=150). Gene expression data of the training cohort were used to develop the breast cancer specific diagnostic model and its performance was determined by predicting the class of the independent test cohort.
 Results: The developed model correctly predicted the class of 110/150 test samples, resulting in an overall accuracy of 73%. Prediction performance was similar for early and late stage cancer and for benign and healthy (no mammographic findings). No significant difference in diagnostic performance was seen between pre- (57/75 correctly predicted) and post-menopausal (53/75 correctly predicted) women indicating the clinical value of our test in younger women.
 Conclusion: The Indian multi-centre study demonstrates the utility of the BC-specific 96 assay signature in detecting BC in an Indian population with similar diagnostic performance seen between women with late stage (stage 2+) and early stage BC. We have previously presented data from European and US cohorts with similar diagnostic performance suggesting the broad applicability of our test. The Breast Cancer test showed the same high level of accuracy in pre-menopausal as in post-menopausal women indicating its clinical applicability in younger women where mammography is of less value due to dense breast tissue.

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 5013.

Employing a blood based gene expression signature to detect early stage breast cancer

21117

Background: Existing methods to detect breast cancer (BC) in asymptomatic patients have limitations, and there is a need to develop more accurate and convenient methods. We recently demonstrated the potential use of gene expression profiling in peripheral blood cells (PBC) for early detection of BC (1) and repeated this with a larger study using the Agilent platform with an accuracy of 75± 7%. Objective: 2 studies are presented that investigate: i) whether effective normalization of experimental conditions can improve diagnostic accuracy, ii) whether a blood based signature developed for BC can discriminate other forms of cancer, and iii) whether an expression signature developed using stage 0 patients can be used to predict BC in stage I disease, and vice versa. Material and Methods: Study I enrolled 60 females with BC and 60 healthy females. Study II enrolled 20 females with early stage BC (10 stage 0 and 10 stage I), 20 healthy females, and 8 females with colon cancer. Gene expression analysis was conducted using the ABI HGSM v2.0 with 32,878 oligo probes. Expression data were analyzed by PLSR for model building and results validated using cross-validation and test set validation. Results: Effective normalization of the data led to improved diagnostic performance. The signature developed using 20 BC and 20 non-BC samples classified 7/8 colon cancer patients as non-BC. The signature developed using stage 0 vs non-BC detected cancer in stage I patients, and the signature developed for stage 1 detected cancer in stage 0 patients. Conclusion: A blood-based gene expression signature can be developed for early stage breast cancer, which is specific and able to distinguish between other forms of malignancy such as colon cancer. The gene expression pattern is systemically affected in early stage BC patients in which there is typically no direct contact of blood cells with cancer cells. References: 1. Sharma P et al. (2005) Breast Cancer Res. 7 (5): R 634–44 2. Aaroe J et al (2006), poster no:125, 97th AACR Annual Meeting, Washington DC, USA Some RT-PCR analyses were performed by Marion Hirt IMGM Laboratories, Martinsried, Germany.

[Table: see text]

Genetic susceptibility and environmental estrogen-like compounds

Environmental chemicals with estrogenic activities have been suggested to be able to interact with the endocrine system. Endogenous estrogen is synthesized in the ovarian theca cells of premenopausal women or in the stromal adipose cells of the breast of postmenopausal women and minor quantities in peripheral tissue. These cells, as well as breast tissue, express all the necessary enzymes for this synthesis, CYP17, CYP11a, CYP19, 17-beta-hydroxysteroid hydrogenase, steroid sulfatase as well as enzymes further hydroxylating estradiol, such as CYP1A1, CYP3A4, CYP1B1, catechol-o-methyltransferase (COMT). Polymorphisms in these enzymes may have a possible role in the link between environmental estrogens and hormone-like substances and the interindividual risk of breast cancer.