1. Genotype-Phenotype Classification of Triple Negative Breast Cancers (TNBC) in Women of African Descent Using the PAM50 NanoString Platform and Genomic Data
Oluwasola A Olayiwola1, Temidayo Ogundiran2, Ashley Hardeman3, Toshio F Yoshimatsu3, Wendy Clayton3, Adewumi Adeoye1, Adeyinka Ademola2, Mustapha Ajani1,
Galina Khramtsova3, Tatyana A Grushko3, Dezheng Huo4, Yonglan Zheng3, Joel Parker5, Charles Perou5, Olufunmilayo I Olopade3
1 Department of Pathology, 2 Department of Surgery, University College Hospital, Ibadan, Nigeria;
3 Center for Clinical Cancer Genetics and Global Health, 4 Department of Public Health Sciences, University of Chicago, Chicago IL, USA;
5 Lineberger Comprehensive Cancer Center, University of North Carolina-Chapel Hill, Chapel Hill NC, USA
Abstract No. P
Introduction
Results 1% 5%
HER2
LumB
LumA
Normal
Basal
Subtype Color Key
Unclassified
It is known that women of African descent are two to three times more likely to develop aggressive forms of breast cancer compared to women of European ancestry, yet wide-scale genomic studies of African and African American breast tumors are limited. To elucidate genotypes and molecular subtypes associated with the most aggressive forms of breast cancer, we used the PAM50 NanoString platform to reclassify Nigerian (NG), African American (AA) and Caucasian (CA) tumors clinically annotated by Immunohistochemistry (IHC) in the past. This work is the first step of our continued effort to correlate our findings to their germline genotype data obtained using high-throughput technologies.
Confidence
Subtype
NG Cohort
AA Cohort
CA Cohort
High Confidence ≥ 0.95
Low Confidence < 0.95
Confidence Color Key
Basal-like
Luminal A
Luminal B
HER2 Enriched
Normal
Subtype Color Key
-3.0 Low Expression
+3.0 High Expression
Expression Level Scale A. C. B.
Materials + Methods
Samples :
Formalin-fixed, paraffin embedded (FFPE) tumor tissues were collected at University College Hospital, Ibadan, Nigeria and the University of Chicago Medical Center, Chicago IL, USA after consenting patients to participate in the study. Blood DNAs were also obtained from each patients where available. 338 samples were prospectively collected on consecutive cases between in Nigeria (NG). 73 samples from women with European ancestry (CA), and 92 samples from African American women (AA) were retrieved from the University of Chicago breast SPORE tumor repository in Chicago. Germline DNA from the cases were included in GWAS using Illumina 2.5 duo chip assay to determine germline genotypes. IRB protocols were approved by the Institutional Review Board of respective institutions prior to consenting patients.
RNA Extraction and Molecular Subtyping by PAM50
RNAs were isolated from formalin-fixed, paraffin embedded (FFPE) tumor tissues using the High Pure Paraffin Kit (Roche) following manufacturer’s protocol. Prior to extraction, all samples were verified for its purity by reviewing a freshly prepared H&E slide. Macrodissection was performed to enrich for tumor cells for samples with less than 100 mm2 tumor surface area. Isolated RNAs were analyzed for their purity and content using NanoDrop. Molecular subtyping of tumors were performed on NanoString nCounter Analysis System using a custom Nano110 (PAM50 + claudin-low & VEGF signatures) probe set following the manufacturer’s protocol. Data quality was checked based on the expression levels of endogenous controls, negative controls, and deviation of global expression level from the median. Intrinsic subtyping were evaluated using custom script written for R statistical software. All study samples were clinically annotated and subtyped by the ER/PR/HER2 IHC classifier.  
Discussion
Figure 1: Distribution of PAM50 subtypes in different population cohorts.
(A) HER2 enriched, Basal-like, and Normal breast-like subtypes are known aggressive subtypes. Majority of NG cohort (52%) belong to these aggressive forms of breast cancer. The US samples was a convenient cohort and does not reflect consecutive cases. (B) Previously published distribution of breast cancer subtypes by IHC across multiple ethnicities (Huo D, et al. JCO 27: , 2009). Concordance of distribution as observed by IHC and PAM50 is consistent for Nigerians. (C) Heat map showing unsupervised clustering of PAM50 gene expression of NG, AA, and CA cohorts. Confidence of subtyping calls and intrinsic molecular subtyping calls by PAM50 is shown using color keys on the top two rows of the figure.
To date, Intrinsic molecular subtyping by Nano110 has been completed on 162 NG, 92 AA and 74 CA tumors.
Concordance between IHC and PAM50 was 54%, which is adequate and comparable to previous studies.
There was a high proportion of basal-like and HER2 subtype and accounted for nearly 35% of NG and 30% of AA cases,
Both IHC and PAM50 subtyping demonstrate population difference in subtype frequencies in this highly selected cohort.
Larger sample size is required to examine association of subtypes with germline genotypes.
PAM50 NanoString assay is reliable and high-throughput for molecular subtyping breast cancer using RNA extracted from FFPE tumors. Ongoing work will correlate PAM50 intrinsic subtypes to genotype data. A. B.
PAM50
PAM50
Subtype
Basal
HER2
LumA
LumB
Normal
Total 17
(58.62%) 1
(3.45%) 7
(24.14%) 3
(10.34%) 29
(100%)
(0.00%)
(25.00%)
(75.00%) 4 2
(11.11%) 13
(72.22%)
(5.56%) 18
Unclassified 19
(33.33%)
(3.51%)
(50.88%) 5
(8.77%) 57
Subtype
Basal
HER2
LumA
LumB
Normal
Total
123
(84.83%) 10
(6.90%) 3
(2.76%) 1
(0.69%) 7
(4.83%)
145
(100%) 6
(16.67%) 29
(80.56%)
(0.00%)
(2.78%) 36 25
(3.61%) 8
(1.15%)
474
(68.40%)
163
(23.52%) 23
(3.32%)
693 5
(4.42%) 21
(18.58%) 49
(43.36%)
(31.86%) 2
(1.77%)
113
159
(16.11%) 68
(6.89%)
526
(53.39%)
200
(20.26%) 33
(3.34%)
986
IHC
IHC
Figure 2. Concordance between PAM50 and IHC subtyping.
Subtyping by IHC is based on interpretation of Estrogen Receptor (ER), Progesterone Receptor (PR), and HER2 expression positivity/negativity. The tumors were subtyped Luminal A (LumA) if ER+/PR+/HER2-; Luminal B (LumB ) if ER+/PR+/HER2+; HER2-enriched (HER2) if ER-/PR-/HER2+; and Basal-like (Basal) if ER-/PR-/HER2-. Percentage inside the parenthesis are raw percentages. (A) IHC data were collected from clinic in Nigeria, and only 57 samples had IHC data available. Concordance of IHC subtype to PAM50 subtype performed on the same samples show overall concordance of 54.4% (Cramer’s V statistics = 0.35 ; Pearson Chi2 = 28.3, P=0.0291). (B) Similar concordance test performed on The Cancer Genome Atlas (TCGA) data set of all population. IHC subtyping were done using the same criteria used in (a), and PAM50 subtyping calls were taken from TCGA data set. Overall concordance between TCGA IHC subtype to PAM50 subtype was 67% (Cramer’s V statistics = 0.58; Pearson Chi2 = 1.0e+03, P<0.0001)
Acknowledgement
This work was funded by NCI 3U01CA S1 awarded to DH and OIO; NCI U01 Supplement Grant awarded to AH and OIO; and Susan G Komen for the Cure grant awarded to OIO.