Enrichment of high-grade tumors in breast cancer gene expression studies

M. van Seijen, A. L. Mooyaart, L. Mulder, M. Hoogstraat, C. A. Drukker, C. E. Loo, B. Pouw, G. S. Sonke, J. Wesseling, E. H. Lips

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

Purpose: Gene expression (GE) profiling for breast cancer classification and prognostication has become increasingly used in clinical diagnostics. GE profiling requires a reasonable tumor cell percentage and high-quality RNA. As a consequence, a certain amount of samples drop out. If tumor characteristics are different between samples included and excluded from GE profiling, this can lead to bias. Therefore, we assessed whether patient and tumor characteristics differ between tumors suitable or unsuitable for generating GE profiles in breast cancer. Methods: In a consecutive cohort of 738 breast cancer patients who received neoadjuvant chemotherapy at the Netherlands Cancer Institute, GE profiling was performed. We compared tumor characteristics and treatment outcome between patients included and excluded from GE profiling. Results were validated in an independent cohort of 812 patients treated with primary surgery. Results: GE analysis could be performed in 53% of the samples. Patients with tumor GE profiles more often had high-grade tumors [odds ratio 2.57 (95%CI 1.77–3.72), p < 0.001] and were more often lymph node positive [odds ratio 1.50 (95%CI 1.03–2.19), p = 0.035] compared to the group for which GE profiling was not possible. In the validation cohort, tumors suitable for gene expression analysis were more often high grade. Conclusions: In our gene expression studies, tumors suitable for GE profiling had more often an unfavorable prognostic profile. Due to selection of samples with a high tumor percentage, we automatically select for tumors with specific features, i.e., tumors with a higher grade and lymph node involvement. It is important to be aware of this phenomenon when performing gene expression analysis in a research or clinical context.

Original languageEnglish
Pages (from-to)327-335
Number of pages9
JournalBreast Cancer Research and Treatment
Volume168
Issue number2
DOIs
Publication statusPublished - 1 Apr 2018

Cite this

van Seijen, M., Mooyaart, A. L., Mulder, L., Hoogstraat, M., Drukker, C. A., Loo, C. E., ... Lips, E. H. (2018). Enrichment of high-grade tumors in breast cancer gene expression studies. Breast Cancer Research and Treatment, 168(2), 327-335. https://doi.org/10.1007/s10549-017-4622-9
van Seijen, M. ; Mooyaart, A. L. ; Mulder, L. ; Hoogstraat, M. ; Drukker, C. A. ; Loo, C. E. ; Pouw, B. ; Sonke, G. S. ; Wesseling, J. ; Lips, E. H. / Enrichment of high-grade tumors in breast cancer gene expression studies. In: Breast Cancer Research and Treatment. 2018 ; Vol. 168, No. 2. pp. 327-335.
@article{a222cae9139e41f8a6ded8edbe7cf25f,
title = "Enrichment of high-grade tumors in breast cancer gene expression studies",
abstract = "Purpose: Gene expression (GE) profiling for breast cancer classification and prognostication has become increasingly used in clinical diagnostics. GE profiling requires a reasonable tumor cell percentage and high-quality RNA. As a consequence, a certain amount of samples drop out. If tumor characteristics are different between samples included and excluded from GE profiling, this can lead to bias. Therefore, we assessed whether patient and tumor characteristics differ between tumors suitable or unsuitable for generating GE profiles in breast cancer. Methods: In a consecutive cohort of 738 breast cancer patients who received neoadjuvant chemotherapy at the Netherlands Cancer Institute, GE profiling was performed. We compared tumor characteristics and treatment outcome between patients included and excluded from GE profiling. Results were validated in an independent cohort of 812 patients treated with primary surgery. Results: GE analysis could be performed in 53{\%} of the samples. Patients with tumor GE profiles more often had high-grade tumors [odds ratio 2.57 (95{\%}CI 1.77–3.72), p < 0.001] and were more often lymph node positive [odds ratio 1.50 (95{\%}CI 1.03–2.19), p = 0.035] compared to the group for which GE profiling was not possible. In the validation cohort, tumors suitable for gene expression analysis were more often high grade. Conclusions: In our gene expression studies, tumors suitable for GE profiling had more often an unfavorable prognostic profile. Due to selection of samples with a high tumor percentage, we automatically select for tumors with specific features, i.e., tumors with a higher grade and lymph node involvement. It is important to be aware of this phenomenon when performing gene expression analysis in a research or clinical context.",
keywords = "Breast cancer, Gene expression, Profiling, Selection bias, Tumor percentage",
author = "{van Seijen}, M. and Mooyaart, {A. L.} and L. Mulder and M. Hoogstraat and Drukker, {C. A.} and Loo, {C. E.} and B. Pouw and Sonke, {G. S.} and J. Wesseling and Lips, {E. H.}",
year = "2018",
month = "4",
day = "1",
doi = "10.1007/s10549-017-4622-9",
language = "English",
volume = "168",
pages = "327--335",
journal = "Breast Cancer Research and Treatment",
issn = "0167-6806",
publisher = "Springer New York",
number = "2",

}

van Seijen, M, Mooyaart, AL, Mulder, L, Hoogstraat, M, Drukker, CA, Loo, CE, Pouw, B, Sonke, GS, Wesseling, J & Lips, EH 2018, 'Enrichment of high-grade tumors in breast cancer gene expression studies' Breast Cancer Research and Treatment, vol. 168, no. 2, pp. 327-335. https://doi.org/10.1007/s10549-017-4622-9

Enrichment of high-grade tumors in breast cancer gene expression studies. / van Seijen, M.; Mooyaart, A. L.; Mulder, L.; Hoogstraat, M.; Drukker, C. A.; Loo, C. E.; Pouw, B.; Sonke, G. S.; Wesseling, J.; Lips, E. H.

In: Breast Cancer Research and Treatment, Vol. 168, No. 2, 01.04.2018, p. 327-335.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Enrichment of high-grade tumors in breast cancer gene expression studies

AU - van Seijen, M.

AU - Mooyaart, A. L.

AU - Mulder, L.

AU - Hoogstraat, M.

AU - Drukker, C. A.

AU - Loo, C. E.

AU - Pouw, B.

AU - Sonke, G. S.

AU - Wesseling, J.

AU - Lips, E. H.

PY - 2018/4/1

Y1 - 2018/4/1

N2 - Purpose: Gene expression (GE) profiling for breast cancer classification and prognostication has become increasingly used in clinical diagnostics. GE profiling requires a reasonable tumor cell percentage and high-quality RNA. As a consequence, a certain amount of samples drop out. If tumor characteristics are different between samples included and excluded from GE profiling, this can lead to bias. Therefore, we assessed whether patient and tumor characteristics differ between tumors suitable or unsuitable for generating GE profiles in breast cancer. Methods: In a consecutive cohort of 738 breast cancer patients who received neoadjuvant chemotherapy at the Netherlands Cancer Institute, GE profiling was performed. We compared tumor characteristics and treatment outcome between patients included and excluded from GE profiling. Results were validated in an independent cohort of 812 patients treated with primary surgery. Results: GE analysis could be performed in 53% of the samples. Patients with tumor GE profiles more often had high-grade tumors [odds ratio 2.57 (95%CI 1.77–3.72), p < 0.001] and were more often lymph node positive [odds ratio 1.50 (95%CI 1.03–2.19), p = 0.035] compared to the group for which GE profiling was not possible. In the validation cohort, tumors suitable for gene expression analysis were more often high grade. Conclusions: In our gene expression studies, tumors suitable for GE profiling had more often an unfavorable prognostic profile. Due to selection of samples with a high tumor percentage, we automatically select for tumors with specific features, i.e., tumors with a higher grade and lymph node involvement. It is important to be aware of this phenomenon when performing gene expression analysis in a research or clinical context.

AB - Purpose: Gene expression (GE) profiling for breast cancer classification and prognostication has become increasingly used in clinical diagnostics. GE profiling requires a reasonable tumor cell percentage and high-quality RNA. As a consequence, a certain amount of samples drop out. If tumor characteristics are different between samples included and excluded from GE profiling, this can lead to bias. Therefore, we assessed whether patient and tumor characteristics differ between tumors suitable or unsuitable for generating GE profiles in breast cancer. Methods: In a consecutive cohort of 738 breast cancer patients who received neoadjuvant chemotherapy at the Netherlands Cancer Institute, GE profiling was performed. We compared tumor characteristics and treatment outcome between patients included and excluded from GE profiling. Results were validated in an independent cohort of 812 patients treated with primary surgery. Results: GE analysis could be performed in 53% of the samples. Patients with tumor GE profiles more often had high-grade tumors [odds ratio 2.57 (95%CI 1.77–3.72), p < 0.001] and were more often lymph node positive [odds ratio 1.50 (95%CI 1.03–2.19), p = 0.035] compared to the group for which GE profiling was not possible. In the validation cohort, tumors suitable for gene expression analysis were more often high grade. Conclusions: In our gene expression studies, tumors suitable for GE profiling had more often an unfavorable prognostic profile. Due to selection of samples with a high tumor percentage, we automatically select for tumors with specific features, i.e., tumors with a higher grade and lymph node involvement. It is important to be aware of this phenomenon when performing gene expression analysis in a research or clinical context.

KW - Breast cancer

KW - Gene expression

KW - Profiling

KW - Selection bias

KW - Tumor percentage

UR - http://www.scopus.com/inward/record.url?scp=85038362707&partnerID=8YFLogxK

U2 - 10.1007/s10549-017-4622-9

DO - 10.1007/s10549-017-4622-9

M3 - Article

VL - 168

SP - 327

EP - 335

JO - Breast Cancer Research and Treatment

JF - Breast Cancer Research and Treatment

SN - 0167-6806

IS - 2

ER -