HER2: the drug landscape across breast, gastric, and lung

HER2 is the target that proved the whole premise of precision oncology: find the molecular lesion, build the drug, treat the patients who carry it. Thirty years on, “HER2-positive” is no longer one thing. It splits into three distinct biomarker states, each with its own drugs, its own resistance story, and its own answer to the question of whether molecular docking even applies.

What HER2 is

HER2 (ERBB2) is a receptor tyrosine kinase in the EGFR/ErbB family. It is the odd one out: it has no known activating ligand of its own. Instead it sits in the membrane in a conformation that is always ready to dimerize, and it is the preferred partner for the other ErbB receptors. When HER2 is overexpressed, that constant readiness becomes constitutive signaling through the PI3K/AKT and RAS/MAPK pathways. Roughly 15-20% of breast cancers carry HER2 gene amplification; it also drives a subset of gastric and gastroesophageal junction tumors, and a smaller slice of lung cancers carry HER2 point mutations instead.

The three biomarker states

• HER2-amplified / overexpressed — IHC 3+ or ISH-positive. Many extra copies of the gene, lots of receptor on the cell surface. This is classic “HER2-positive” disease and the population every first-generation HER2 drug was built for.
• HER2-low — IHC 1+ or 2+/ISH-negative. Not enough receptor to call the tumor HER2-positive by the old binary, but enough for an antibody-drug conjugate to find and kill the cell. This category did not clinically exist until a drug created it.
• HER2-mutant — activating point mutations or small insertions in the gene, mostly independent of amplification. The recurring lesions are exon 20 insertions (the YVMA insertion is the canonical one), plus kinase-domain substitutions like L755S and V777L and the extracellular S310F. These show up in 2-4% of non-small-cell lung cancers and a smaller fraction of breast and other tumors. Crucially, a HER2-mutant tumor is usually not HER2-amplified, so the standard IHC test misses it entirely.

Four drug classes, and which state each one serves

Naked antibodies. Trastuzumab binds extracellular domain IV of HER2; pertuzumab binds domain II and physically blocks HER2 from dimerizing with HER3. Trastuzumab plus chemotherapy was the 1998 approval that started everything, and adding pertuzumab on top (the CLEOPATRA regimen) became the long-running first-line standard in metastatic HER2-positive breast cancer. Both need abundant surface receptor, so they are amplified-disease drugs.

Antibody-drug conjugates. An ADC uses the antibody only as a delivery vehicle for a cytotoxic payload. T-DM1 (trastuzumab emtansine) carries a maytansinoid microtubule poison on a non-cleavable linker. T-DXd (trastuzumab deruxtecan) carries a topoisomerase I inhibitor on a cleavable linker, and because the freed payload can cross into neighboring cells, it kills tumor even where HER2 expression is sparse. That bystander effect is exactly why T-DXd works in HER2-low disease where naked antibodies do nothing. Interstitial lung disease is the toxicity that defines T-DXd’s safety monitoring.

Small-molecule tyrosine kinase inhibitors. These bind the intracellular ATP pocket. Lapatinib is a reversible dual EGFR/HER2 inhibitor; neratinib is an irreversible pan-HER inhibitor whose dose-limiting toxicity is diarrhea; tucatinib is HER2-selective and crosses the blood-brain barrier well enough to matter for patients with brain metastases. This is the only drug class that is a docking problem.

The HER2-mutant lane. HER2 mutations turned out to respond to ADCs rather than to the TKIs designed against amplified disease. T-DXd earned accelerated approval for HER2-mutant NSCLC in 2022, the first targeted therapy for that population, on the strength of the DESTINY-Lung data. The TKIs have had a harder time here: neratinib and tucatinib show activity against some kinase-domain mutants in basket trials, but the exon 20 insertions in particular reshape the pocket in ways that blunt the ATP-competitive drugs.

How HER2-positive tumors escape

Resistance to trastuzumab clusters around the downstream pathway rather than the receptor itself. PIK3CA activating mutations and PTEN loss both reactivate PI3K/AKT signaling underneath a fully blocked receptor. A truncated form of the receptor called p95HER2, which lacks the extracellular domain trastuzumab binds, leaves the kinase active with nothing for the antibody to grab. Each of these is a reason a HER2 program eventually needs either an ADC (deliver a payload regardless of signaling) or a kinase inhibitor (hit the catalytic domain p95HER2 still has).

Try the molecular docking yourself

Only one of the four drug classes is a molecular docking problem. Antibodies and ADCs are engineered against the extracellular domain and, in the ADC case, do their real work through a payload, so there is no small-molecule pose to score. The TKIs are different: lapatinib, neratinib, and tucatinib all sit in the HER2 kinase ATP pocket, and kinase-domain mutations like L755S and V777L change that pocket directly. Open Studio on Liganx, the free molecular docking online platform, and pick HER2 from the target catalog, then choose a kinase-domain mutation from the mutation chips. Run molecular docking on the three TKIs against wild-type and mutant side by side and watch the ΔΔ: the number that tells you whether a mutation is likely to keep, weaken, or break a given inhibitor is the wild-type-to-mutant gap, not any single absolute score. The ADMET panel will also flag the diarrhea-adjacent and hepatic liabilities that have shaped this drug class’s tolerability.

Primary sources

• Slamon DJ, et al. Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. NEJM 344, 783-792 (2001). doi:10.1056/NEJM200103153441101
• Modi S, et al. Trastuzumab deruxtecan in previously treated HER2-low advanced breast cancer. NEJM 387, 9-20 (2022). doi:10.1056/NEJMoa2203690
• Li BT, et al. Trastuzumab deruxtecan in HER2-mutant non-small-cell lung cancer. NEJM 386, 241-251 (2022). doi:10.1056/NEJMoa2112431