Proteomics: SLC27A2 Predicts Pyrotinib Response in HER2+ BC
Pyrotinib clears HER2-positive breast tumours in a good share of patients, and nothing in the pre-treatment workup says which share a given patient falls into. A group at the Fourth Hospital of Hebei Medical University went after that gap with tumour proteomics, profiling pre-therapy biopsies from six women who later reached a pathological complete response (pCR) and six who did not. The 617 proteins separating those groups were pushed through co-expression networks, 127 machine-learning strategies, and cell and xenograft work. What survived was SLC27A2, a long-chain fatty acid transporter.
Key Takeaways
High pre-treatment SLC27A2 protein cut the odds of a complete pathological response to pyrotinib roughly tenfold across 103 independently assessed patients.
Discovery proteomics on twelve tumour biopsies can nominate a drug-response candidate, but the shortlist earns trust only once an untouched cohort confirms it.
Silencing SLC27A2 made breast cancer cells more sensitive to pyrotinib, so the protein reads as a possible combination target, not a passive marker.
From Discovery Proteomics to One Candidate Protein
The discovery design was small and openly so: six pCR and six non-pCR biopsies, matched one to one on outcome rather than sampled at random. Protein profiling ran at a commercial facility, with differential abundance called at an absolute log2 fold change of 1.5 or greater and adjusted p below 0.05. Co-expression analysis isolated a module tracking response almost perfectly (r = 0.99), its overlap with the differential list came to 84 proteins, and topological filtering left nine hub candidates. Ranking those nine across three public cohorts with 127 modelling strategies put a glmBoost plus random forest pairing on top.
Key Findings
617 differentially expressed proteins split responders from non-responders: enrichment pointed to peroxisomal and mitochondrial compartments alongside T cell receptor signalling and autophagy.
SLC27A2 topped the machine-learning shortlist: the winning model reached a mean AUC of 0.678, and SLC27A2 ranked first on SHAP importance in GSE48390 and second in GSE16446.
Knockdown sensitised cells to the drug: silencing SLC27A2 in SKBR3 and MDA-MB-453 lines shifted pyrotinib dose-response curves downward, and a factorial xenograft study showed an interaction with treatment on endpoint tumour weight.
Clinical confirmation in 103 patients: high pre-treatment SLC27A2 staining independently predicted failure to reach pCR after pyrotinib-containing neoadjuvant therapy (OR 0.10, 95% CI 0.03 to 0.31, p < 0.001).
Figure 1. Discovery proteomics in pre-treatment HER2-positive breast tumours. Panel A is a volcano plot of proteins differing between pCR and non-pCR biopsies, Panel B a heatmap of representative markers, and Panel C their Gene Ontology and KEGG enrichment. Panels D and E map the same proteins onto enriched terms and pathways. Panels F through H cover soft-threshold selection, the module dendrogram, and module-trait correlations, and Panel I the 84-protein overlap carried forward. Adapted from Zhang et al. (2026), Cancers.
Why a Fatty Acid Transporter Turns Up in a HER2 Story
SLC27A2 ferries long-chain fatty acids across membranes, which is not where most HER2 resistance work has been looking. Enrichment analysis in TCGA-BRCA tied its expression to fatty acid metabolism, and knockdown cells showed weaker Oil Red O staining plus lower intracellular triglycerides. A PPAR-alpha agonist partly reversed that sensitisation. The authors call the mechanism preliminary, which is the honest read: partial rescue in two cell lines is suggestive, not evidence of direct regulation.
Reading the Result Against Its Own Limits
One constraint deserves attention before anyone designs a trial around this protein. The public cohorts used for model ranking carried no pyrotinib treatment data, so that step measured prognostic signal rather than drug-specific prediction; the 103-patient immunohistochemistry cohort is what carries the predictive claim, and it came from one institution over twelve months of retrospective collection. Set against the phase 3 neoadjuvant evidence from Wu and colleagues (2022), an odds ratio of 0.10 is large enough to justify prospective multicentre replication.
Implications for Response-Biomarker Programs
The practical lesson for drug-discovery teams sits in how the work was sequenced: a twelve-sample discovery run stays cheap enough to be exploratory, and it pays off only because an untouched 103-patient cohort was held back for the one protein that survived. Teams at Dalton see the same pattern in client programs, where quantitative protein profiling by mass spectrometry is planned alongside the orthogonal validation assay from day one, not after the discovery list lands. That habit matters most in multi-omics biomarker discovery, where a candidate that doesn't hold up in fresh samples isn't yet a biomarker.
Frequently Asked Questions
What is proteomics used for in drug development?
Proteomics measures which proteins are present in a sample and at what abundance, most often by mass spectrometry. Drug developers use it to find response and resistance markers, check target engagement, and explain why a subset of patients fails to benefit.
Can proteomics predict which patients respond to HER2-targeted therapy?
This study suggests it can help. Pre-treatment SLC27A2 levels were independently associated with pathological complete response after pyrotinib-containing neoadjuvant therapy, at an odds ratio of 0.10. Prospective multicentre replication is still needed.
How many samples does a proteomics discovery study need?
Fewer for discovery than most teams expect, and more for validation. This work began with twelve biopsies in a balanced six-versus-six design, then tested one protein in 103 independent patients.
Conclusion
The association holds up: tumours staining high for SLC27A2 before treatment were far less likely to clear on pyrotinib. Mechanism is another matter, resting on partial rescue in two cell lines and one xenograft experiment. The cheap next move for a HER2 neoadjuvant trial is to stain banked biopsies and see whether an odds ratio near 0.10 survives a second institution.
Related Reading
See how we run these analyses in one lab: Dalton's multi-omics CRO services.
Citation
Zhang, S., Yang, X., Zhang, Y., Cheng, S., Niu, H., Liao, X., Li, Y., & Ma, L. (2026). Integrative proteomics and machine learning identify SLC27A2 as a candidate biomarker and potential mediator of pyrotinib response in HER2-positive breast cancer. Cancers, 18(16), 2702. https://doi.org/10.3390/cancers18162702
Note
This blog post summarizes findings from the above-cited research. Figures are adapted from the original publication. For full details, please refer to the source article.
By Seungjun Yeo, CEO at Dalton Bioanalytics. Specializing in multi-omics mass spectrometry for drug discovery and biomarker research.