Synthesis: Burstein four-subtype framework for TNBC

Evidence grades (GRADE-adapted): A high — multiple well-conducted RCTs or systematic reviews converge. B moderate — single pivotal RCT or consistent observational evidence. C limited — single observational study, mechanistic, or expert consensus. D preclinical / hypothesis-generating.

Why a separate TNBC subtyping system

The intrinsic-subtype framework introduced by Perou and Sørlie classifies approximately 80% of TNBC into a single category — basal-like (see the PAM50 history synthesis). For investigators interested in within-TNBC biological heterogeneity, this lumping is unhelpful: a single label covering 80% of the population doesn't help stratify response to therapy or identify subgroups with distinct biology. Multiple groups have proposed TNBC-specific subtyping systems to subdivide the basal-like majority and capture finer biological structure.

Two principal frameworks compete in the literature:

• The Lehmann/Pietenpol six-subtype (2011) / four-subtype refined (2016) framework, derived from gene-expression-only clustering of 587 TNBC tumors. See the Lehmann/Pietenpol subtypes synthesis.
• The Burstein four-subtype framework, derived from integrated multi-omic (expression + methylation + copy number + mutation) clustering of 198 TNBC tumors. This page.

The Burstein derivation

Burstein and colleagues at Baylor College of Medicine analyzed 198 TNBC tumors using integrated multi-omic clustering — gene expression, DNA methylation, somatic copy-number alteration, and exome sequencing — rather than gene expression alone^[1]A. The methodology contrasts with Lehmann's expression-only approach and was designed to capture biological structure that wouldn't appear in expression data alone (e.g., methylation-silenced tumor suppressor pathways, copy-number-driven oncogene activation that does not require transcriptional upregulation).

The integrated clustering produced four reproducible subtypes:

• BLIA — Basal-Like Immune Activated
• BLIS — Basal-Like Immune Suppressed
• MES — Mesenchymal
• LAR — Luminal Androgen Receptor

BLIA — basal-like immune-activated

Defining features:

• STAT signal-transduction pathway activation (STAT1, STAT3, STAT4 expression and downstream gene induction)
• Immune-cell signature enrichment — high interferon-gamma response genes, T-cell and NK-cell signatures, immune checkpoint molecule expression (PD-L1, PD-L2, IDO1)
• Antigen presentation machinery upregulation (HLA-A, HLA-B, B2M)
• Frequent tumor-infiltrating lymphocyte presence by histology

Reported prognostic association: BLIA had the best overall and disease-free survival of the four subtypes in the original Burstein cohort. This pattern has been replicated in multiple subsequent cohorts^[2]B.

The biological message — that an immune-activated TNBC subset has favorable prognosis driven by anti-tumor immunity — prefigured the immunotherapy-era insight that high tumor-infiltrating lymphocyte (TIL) content predicts both prognosis and response to immune checkpoint inhibitors. See the TME and TILs synthesis for the operationalization of this biology via the TILs Working Group standard.

BLIS — basal-like immune-suppressed

Defining features:

• SOX-family transcription factor expression (SOX4, SOX8, SOX10)
• Cell-cycle and proliferation gene enrichment (similar to Lehmann BL1)
• Down-regulation of immune-effector and antigen-presentation gene sets
• Frequent biallelic loss of immune-checkpoint and chemokine genes

Reported prognostic association: BLIS had the worst overall survival of the four subtypes in the original Burstein cohort. Replication in subsequent series has been consistent but with variable magnitude^[2].

The BLIA vs BLIS dichotomy is the Burstein framework's most distinctive contribution. Lehmann's BL1 and BL2 subtypes are split on DDR-pathway vs growth-factor signaling axes; Burstein's BLIA and BLIS are split on the immune-active vs immune-suppressed axis. Both splits are valid; they capture different aspects of the same basal-like population.

MES — mesenchymal

Defining features:

• Epithelial-to-mesenchymal transition (EMT) signatures — SNAI1/2, ZEB1/2, TWIST1 expression
• Cytoskeletal and ECM-receptor gene enrichment
• Growth-factor pathway activity (TGF-beta, WNT)
• Lower proliferation than BLIS but high motility

Reported prognostic association: intermediate between BLIA (best) and BLIS (worst); some series show MES with higher distant-metastasis rates despite lower proliferation, consistent with the EMT-mediated motility phenotype.

Burstein MES overlaps substantially with Lehmann's M (mesenchymal) subtype. Both frameworks identify a mesenchymal-axis subset with similar biology.

LAR — luminal androgen receptor

Defining features:

• Androgen receptor (AR) signaling pathway activation — AR target genes (KLK3, FKBP5, NDRG1, DHCR24) at high expression
• Luminal differentiation markers (CK8/18, GATA3)
• Frequent PIK3CA mutations (typically activating)
• Lower proliferation rates than basal-like subtypes

LAR is the most stable subtype across all TNBC molecular frameworks — Lehmann/Pietenpol, Burstein, and FUSCC (Jiang 2019) all identify a clearly delineated LAR subset, typically representing 10–20% of TNBC. AR antagonism is the LAR-specific clinical lead; bicalutamide (TBCRC 011), enzalutamide (Traina 2018), and combinations (TBCRC 032 enzalutamide + PI3K inhibitor) have shown proof-of-concept activity in AR-positive TNBC. See the LAR and AR-targeting synthesis for clinical-trial detail.

Reconciliation with Lehmann/Pietenpol

The Lehmann/Pietenpol and Burstein frameworks differ in derivation methodology and in the number/definition of subtypes, but they converge on common biological structures^[3]B:

Inter-classifier concordance on individual patient samples (Burstein vs Lehmann TNBCtype-4) typically runs 60–75% across cohorts — reasonable agreement on the LAR and mesenchymal categories, less consistent on the basal subdivisions^[4]B.

Subsequent integrative work

The Jiang 2019 FUSCC integrative analysis of 465 Chinese TNBC tumors proposed a four-subtype framework (LAR, IM, BLIS, MES) explicitly designed to reconcile Lehmann and Burstein^[5]B. The FUSCC framework is the basis for the FUTURE-series biomarker-stratified clinical trials in China; outcomes in those trials suggest subtype-matched targeted therapy can produce response rates substantially higher than historical controls in pre-treated metastatic TNBC, with replication in non-Chinese cohorts pending.

Bareche 2018 analyzed 550 TNBC tumors from METABRIC and TCGA, applying integrated multi-omic clustering similar to Burstein's, and produced a four-cluster solution mapping cleanly to BL1/BL2/M/LAR (with TIL signatures as a separate immune axis)^[6]B. The recurring conclusion across these subsequent studies: LAR is the most reproducible discrete entity; mesenchymal axis is reproducible but with cohort-dependent boundaries; the basal-like majority is more variable in how it subdivides.

Clinical applicability today

Burstein-subtype testing in routine TNBC clinical practice is not currently standard. The framework appears in:

• Clinical-trial stratification. Several trials use Burstein subtype calls (or aligned LAR / immune-rich definitions) to enrich for biomarker-positive populations or to stratify analyses.
• Retrospective subgroup analyses of immunotherapy and chemotherapy trials, attempting to identify subtype-by-treatment interactions. Results have been suggestive but underpowered for definitive conclusions.
• Mechanism-of-resistance research, where subtype assignment helps interpret heterogeneity in single-cell or longitudinal cohorts.

Three Burstein-aligned clinical-utility leads have advanced:

1. BLIA / immune-rich TNBC → IO benefit. The TILs Working Group standard provides a practical histologic biomarker that captures the BLIA biology without requiring transcriptomic assays. See the TME and TILs synthesis.
2. LAR → AR-targeted therapy. Bicalutamide, enzalutamide, and combination trials have shown activity; definitive phase III pending. See the LAR synthesis.
3. BLIS → DDR-targeted therapy. Because BLIS retains the DDR-enriched biology of Lehmann's BL1 alongside immune suppression, it may be the right subset for cell-cycle checkpoint inhibitors (CHK1, WEE1, ATR) currently in phase II. Not yet established.

Evidence table

Open questions and active investigation

• Will Burstein-subtype assignment ever enter routine clinical use? Requires a clinically actionable subtype-specific therapy that doesn't have an easier-to-deploy biomarker proxy. The BLIA / IO benefit case is well-served by TIL scoring; the LAR / AR antagonist case is well-served by AR IHC. BLIS / DDR-inhibitor would be a candidate if those trials are positive.
• Sample-composition contamination. Like the original Lehmann framework, the Burstein subtypes can be affected by stromal and immune-cell content in bulk tumor samples. Whether the BLIA/BLIS distinction is tumor-cell-intrinsic or reflects microenvironment differences is still debated; single-cell decomposition would help.
• Burstein vs Lehmann inter-framework concordance. Approximately 25–40% of patients receive discordant subtype calls between the two systems. Whether either is "right" is unclear; perhaps both capture meaningful but non-identical axes of biological variation.
• Prognostic stability across treatment. Whether subtype changes after KEYNOTE-522 chemo+IO is unknown; serial-biopsy studies are technically demanding but would inform sequential-therapy decisions.
• Integration with HER2-low. Now that HER2-low status sub-divides TNBC into therapeutically relevant categories, the intersection of HER2-low status with Burstein subtype hasn't been characterized. Whether HER2-low BLIA differs in T-DXd response from HER2-low BLIS is the kind of question only large-N retrospective cohorts can address.

For the broader intrinsic-subtype framework, see the intrinsic subtypes and PAM50 synthesis. For the competing TNBC-specific framework, see the Lehmann/Pietenpol subtypes synthesis. For the TIL biology that BLIA captures, see the TME and TILs synthesis.

References

Each citation links to the original publication via DOI. The same records are searchable in the evidence library by title or DOI.

1. Burstein MD, Tsimelzon A, Poage GM, et al. Comprehensive genomic analysis identifies novel subtypes and targets of triple-negative breast cancer. Clin Cancer Res. 2015;21(7):1688–1698. doi:10.1158/1078-0432.CCR-14-0432. ↩
2. Denkert C, von Minckwitz G, Darb-Esfahani S, et al. Tumour-infiltrating lymphocytes and prognosis in different subtypes of breast cancer: a pooled analysis of 3771 patients treated with neoadjuvant therapy. Lancet Oncol. 2018;19(1):40–50. doi:10.1016/S1470-2045(17)30904-X. ↩
3. Lehmann BD, Jovanović B, Chen X, et al. Refinement of triple-negative breast cancer molecular subtypes: implications for neoadjuvant chemotherapy selection. PLOS One. 2016;11(6):e0157368. doi:10.1371/journal.pone.0157368. ↩
4. Bareche Y, Venet D, Ignatiadis M, et al. Unravelling triple-negative breast cancer molecular heterogeneity using an integrative multi-omic analysis. Ann Oncol. 2018;29(4):895–902. doi:10.1093/annonc/mdy024. ↩
5. Jiang YZ, Ma D, Suo C, et al. Genomic and transcriptomic landscape of triple-negative breast cancers: subtypes and treatment strategies. Cancer Cell. 2019;35(3):428–440.e5. doi:10.1016/j.ccell.2019.02.001. ↩
6. Bareche Y, Venet D, Ignatiadis M, et al. Unravelling triple-negative breast cancer molecular heterogeneity using an integrative multiomic analysis. Ann Oncol. 2018;29(4):895–902. doi:10.1093/annonc/mdy024. ↩

Last reviewed: 2026-06-04. Researcher-layer synthesis page. Evidence grades follow the GRADE-adapted rubric defined at the top of this page.