CoChat

Inferring tumour purity and stromal and immune cell admixture from expression data

Open access
Yoshihara, Kosuke, Shahmoradgoli, Maria, Martínez, Emmanuel, Vegesna, Rahulsimham, Kim, Hoon, Torres-Garcia, Wandaliz, Treviño, Victor, Shen, Hui, Laird, Peter W., Levine, Douglas A., Carter, Scott L., Getz, Gad, Stemke-Hale, Katherine, Mills, Gordon B., Verhaak, Roel G.W. · 2013 · Nature Communications

The DOI matches its Crossref record; 64 of the 64 references it cites were found in a registry or an index.

Abstract

Infiltrating stromal and immune cells form the major fraction of normal cells in tumour tissue and not only perturb the tumour signal in molecular studies but also have an important role in cancer biology. Here we describe ‘Estimation of STromal and Immune cells in MAlignant Tumours using Expression data’ (ESTIMATE)—a method that uses gene expression signatures to infer the fraction of stromal and immune cells in tumour samples. ESTIMATE scores correlate with DNA copy number-based tumour purity across samples from 11 different tumour types, profiled on Agilent, Affymetrix platforms or based on RNA sequencing and available through The Cancer Genome Atlas. The prediction accuracy is further corroborated using 3,809 transcriptional profiles available elsewhere in the public domain. The ESTIMATE method allows consideration of tumour-associated normal cells in genomic and transcriptomic studies. An R-library is available on https://sourceforge.net/projects/estimateproject/ . Tumour biopsies contain contaminating normal cells and these can influence the analysis of tumour samples. In this study, Yoshihara et al.develop an algorithm based on gene expression profiles from The Cancer Genome Atlas to estimate the number of contaminating normal cells in tumour samples.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/ncomms3612

Cites

64 of 64 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 An overview of the ESTIMATE algorithm. The ESTIMATE algorithm uses gene expression data to output the estimated levels of infiltrating stromal and immune cells and estimated tumour purity. Infiltrating stromal- and immune cell-related genes were identified by five gene filterings.
Figure 1 An overview of the ESTIMATE algorithm. The ESTIMATE algorithm uses gene expression data to output the estimated levels of infiltrating stromal and immune cells and estimated tumour purity. Infiltrating stromal- and immune cell-related genes were identified by five gene filterings.CC BY-NC-SA · PubMed Central
Figure 2 Stromal and immune scores for tumour cell and stromal fractions of tumour samples. Stromal and immune scores were generated using expression data sets obtained from tumour cell or stromal cell-enriched samples. (a,b) Heatmaps display stromal (upper row) and immune score (lower row) per sample (each column) using ovarian cancer samples after (a) microbead-based cell sorting and (b) laser-capture microdissection (red=high, blue=low score). (c,d) Box and whisker plots display reduced (c) stromal and (d) immune scores for the tumour cell-enriched samples (tumour part) after laser-capture microdissection compared with matched stromal cell-enriched (ovary, breast) or bulk tumour samples (lung). Box represents the median (thick line) and the quartiles (line). Whisker expresses 1.5 interquartile range (IQR) of the lower or the upper quartile.
Figure 2 Stromal and immune scores for tumour cell and stromal fractions of tumour samples. Stromal and immune scores were generated using expression data sets obtained from tumour cell or stromal cell-enriched samples. (a,b) Heatmaps display stromal (upper row) and immune score (lower row) per sample (each column) using ovarian cancer samples after (a) microbead-based cell sorting and (b) laser-capture microdissection (red=high, blue=low score). (c,d) Box and whisker plots display reduced (c) stromal and (d) immune scores for the tumour cell-enriched samples (tumour part) after laser-capture microdissection compared with matched stromal cell-enriched (ovary, breast) or bulk tumour samples (lung). Box represents the median (thick line) and the quartiles (line). Whisker expresses 1.5 interquartile range (IQR) of the lower or the upper quartile.CC BY-NC-SA · PubMed Central
Figure 3 The association between tumour purity variables in TCGA’s ovarian cancer data set. (a–d) Scatterplots between tumour purity and (a) stromal, (b) immune, (c) ESTIMATE scores and between (d) stromal and immune scores in the TCGA ovarian cancer data set. TCGA ovarian cancer samples used in the gene selection (n=28) were not included in the figure. Dash lines denote each median value for stromal and immune scores. (e) The association between tumour purity and stromal- or immune-dominant pattern. Four subgroups were divided based on the median of stromal and immune scores. (f) The ROC curves for four cutoff values in TCGA ovarian cancer data set. N=417.
Figure 3 The association between tumour purity variables in TCGA’s ovarian cancer data set. (a–d) Scatterplots between tumour purity and (a) stromal, (b) immune, (c) ESTIMATE scores and between (d) stromal and immune scores in the TCGA ovarian cancer data set. TCGA ovarian cancer samples used in the gene selection (n=28) were not included in the figure. Dash lines denote each median value for stromal and immune scores. (e) The association between tumour purity and stromal- or immune-dominant pattern. Four subgroups were divided based on the median of stromal and immune scores. (f) The ROC curves for four cutoff values in TCGA ovarian cancer data set. N=417.CC BY-NC-SA · PubMed Central
Figure 4 Evaluation of ESTIMATE algorithm. The accuracy of the ESTIMATE algorithm was evaluated by the AUC when tumour samples were divided into high- and low-purity groups on the basis of DNA copy number-based tumour purity. (a,b) The ROC curves for four cutoff values in (a) the Agilent data set, the Affymetrix data set, and the RNAseq data set, the RNAseqV2 data set, and (b) the validation data set. (c) An example of ESTIMATE for new Affymetrix sample, with an ESTIMATE-predicted tumour purity of 0.58. Black dot and grey dash lines show ESTIMATE tumour purity and 95% prediction interval, respectively. The grey dots represent the background distribution based on 955 samples from the TCGA Affymetrix data set.
Figure 4 Evaluation of ESTIMATE algorithm. The accuracy of the ESTIMATE algorithm was evaluated by the AUC when tumour samples were divided into high- and low-purity groups on the basis of DNA copy number-based tumour purity. (a,b) The ROC curves for four cutoff values in (a) the Agilent data set, the Affymetrix data set, and the RNAseq data set, the RNAseqV2 data set, and (b) the validation data set. (c) An example of ESTIMATE for new Affymetrix sample, with an ESTIMATE-predicted tumour purity of 0.58. Black dot and grey dash lines show ESTIMATE tumour purity and 95% prediction interval, respectively. The grey dots represent the background distribution based on 955 samples from the TCGA Affymetrix data set.CC BY-NC-SA · PubMed Central
Figure 5 Correlation of scores with histological findings. Scatterplots between stromal, immune, ESTIMATE scores and ABSOLUTE-based tumour purity versus the following histological findings: percentage of stromal cells (left upper corner), percentage of infiltrating lymphocytes (right upper corner), and percentage of tumour cells (bottoms panels). Twenty-eight TCGA ovarian cancer samples used in the gene selection were excluded from this analysis.
Figure 5 Correlation of scores with histological findings. Scatterplots between stromal, immune, ESTIMATE scores and ABSOLUTE-based tumour purity versus the following histological findings: percentage of stromal cells (left upper corner), percentage of infiltrating lymphocytes (right upper corner), and percentage of tumour cells (bottoms panels). Twenty-eight TCGA ovarian cancer samples used in the gene selection were excluded from this analysis.CC BY-NC-SA · PubMed Central
Figure 6 Unique distribution of stromal and immune scores. (a,b) Distinct distributions of (a) stromal and (b) immune scores across different tumour types were observed in RNAseqV2Affymetrix platform data sets. The number of parenthesis means sample size per data sets.
Figure 6 Unique distribution of stromal and immune scores. (a,b) Distinct distributions of (a) stromal and (b) immune scores across different tumour types were observed in RNAseqV2Affymetrix platform data sets. The number of parenthesis means sample size per data sets.CC BY-NC-SA · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 64 references

Cited by

Items 701–800 of 1,021

Metadata

Authors
Yoshihara, K., Shahmoradgoli, M., Martínez, E., Vegesna, R., Kim, H., Torres-Garcia, W., Treviño, V., Shen, H., Laird, P. W., Levine, D. A., Carter, S. L., Getz, G., Stemke-Hale, K., Mills, G. B., & Verhaak, R. G. W.
Affiliations
  • The University of Texas MD Anderson Cancer Center (US)
  • University of Southern California (US)
  • Memorial Sloan Kettering Cancer Center (US)
  • Broad Institute (US)
Corresponding author
Verhaak, Roel G.W.
Venue
Nature Communications
Published
2013-10-11
Volume
4
Issue
1
Pages
2612-2612
ISSN
2041-1723 (ISSN-L)
OpenAlex
W2130430382
Cited by
9,778
References
63
Citations by year
2026: 694 · 2025: 1176 · 2024: 1602 · 2023: 1755 · 2022: 1988 · 2021: 1316 · 2020: 683 · 2019: 223 · 2018: 121 · 2017: 93 · 2016: 53 · 2015: 39 · 2014: 20 · 2013: 2
FWCI
17.01
Type
journal article
Language
en
Open access
Locations
  • Nature Communications (publishedVersion, cc-by-nc-sa)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by-nc-sa)
  • PubMed Central (publishedVersion, cc-by-nc-sa)
  • PubMed (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
Funders
  • University of Texas MD Anderson Cancer Center
  • National Cancer Institute (P50CA098258, U24CA143882, CA143883, U24CA143883, P30CA008748, P30CA016672)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Gene expression and cancer classification
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Cancer Research › Cancer Genomics and Diagnostics
Keywords
tumor purity, tumor stroma, tumor-infiltrating immune cells, gene expression signatures, tumor microenvironment, DNA copy number, cancer transcriptomics
MeSH terms
Algorithms, Cell Separation, Female, Humans, Leukocytes, Neoplasms, Research Design, Sensitivity and Specificity, Software, Gene Library, Genome, Human, Stromal Cells, Oligonucleotide Array Sequence Analysis, Gene Expression Profiling, DNA Copy Number Variations, High-Throughput Nucleotide Sequencing, Transcriptome

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.