S100–RAGE Signaling and Enhertu Resistance: A New Target in Metastatic Breast Cancer?
In this article
S100–RAGE Signaling in Breast Cancer Not Responding to Enhertu: A New Resistance Target?
Trastuzumab deruxtecan (T-DXd, Enhertu) is one of the most widely used antibody-drug conjugates in metastatic breast cancer, yet in some patients it does not work from the very first assessment. A study published in Clinical Cancer Research and led by Houston Methodist reported that S100 proteins (especially S100A7 and S100A9) were higher in metastatic lesions that did not respond to T-DXd, and that TTP488, a substance that blocks the RAGE receptor, increased the effect of T-DXd in cell culture and in mice. This is a promising hypothesis for patients receiving T-DXd treatment, but it rests on a small patient cohort, four spatial-analysis samples and immunodeficient mice. The combination has not yet been tried in humans.
- What was found: In 109 metastatic lesions from 47 patients, HER2 level, hormone receptor status, Ki-67 and the organ where the lesion was located did not distinguish T-DXd response; prior T-DM1 use was associated with a weaker response. In spatial profiling, S100A7/S100A9 were higher in non-responding tumors.
- RAGE blockade: TTP488 lowered the IC50 of T-DXd from 13 nM to 0.27 nM in SKBR3 cells and from 28 nM to 0.04 nM in BT474 cells. In a mouse lung metastasis model, the combination reduced the number of metastatic foci compared with T-DXd alone (P = 0.01).
- Caution: The spatial analysis rests on 4 samples (two from bone and two from brain); the experiments were done in immunodeficient mice; the combination of a RAGE blocker with T-DXd was not tested in patients.
- In practice: No finding here should change a treatment plan. TTP488 (azeliragon) is not approved for breast cancer; patients taking T-DXd should neither stop the drug nor add another substance on their own.
Why does this topic matter?
T-DXd is an antibody-drug conjugate that reaches a HER2-bearing cell and releases a chemotherapy payload (a topoisomerase I inhibitor) inside it. It has been approved in many settings, such as HER2-low and HER2-ultralow breast cancer, first-line HER2-positive disease and the combination with pertuzumab. The introduction of the paper cites the DESTINY-Breast03, 04, 06 and 09 trials; the authors estimate that the drug is used in more than 90% of patients with metastatic breast cancer in the United States (this figure is the authors' own estimate and could not be verified independently).
Even so, not everyone responds. According to the paper, about 30% of HER2 IHC 3+ tumors did not respond in the DAISY trial. Knowing in advance which patient will not respond matters. Today the choice rests on HER2 status; yet in this study HER2 level alone did not separate responders from non-responders.
The authors' hypothesis is that T-DXd resistance may depend less on the organ in which the lesion sits than on the tumor cell's own signaling program and on the architecture of tumor and immune cells. To test this, they compared bone and brain metastases at the spatial level. For general background, see our articles on breast cancer and metastasis.
No. Resistance has been examined from different angles by several teams in recent years; the contribution of this study is that it combines spatial profiling of the same patient lesions with experiments on the S100–RAGE pathway and a RAGE inhibitor.
- 2023: The phase 2 DAISY trial examined T-DXd response in metastatic breast cancer with variable HER2 expression; the paper cites the non-response rate in that trial.
- January 2026: A team from Caris Life Sciences reported in npj Breast Cancer the molecular data of 2,799 breast cancer patients treated with T-DXd: high ERBB2 expression was linked to better outcomes and high ABCC1 (a drug efflux pump) expression to worse outcomes; post-treatment samples more often carried ERBB2, NFE2L2, KEAP1 and TOP1 mutations.
- 2026 (this study): Response was increased by adding a RAGE inhibitor, targeting S100–RAGE signaling (at cell and mouse level). We did not find this link between S100–RAGE and T-DXd elsewhere; however, we did not run an exhaustive literature search.
Can S100–RAGE signaling cause T-DXd resistance, and what happens if RAGE is blocked?
According to the study, S100A7 and S100A9 were higher in metastatic breast cancer lesions that did not respond to T-DXd. TTP488, which blocks the RAGE receptor, markedly lowered the effective dose of T-DXd in cell culture and increased its effect in a mouse lung metastasis model. The combination has not been tried in humans; the findings are associative and preclinical.
How was the study done?
The work was carried out at Houston Methodist in the laboratory of Stephen T.C. Wong, together with Emory University. Funding came from NIH grants and foundations; no conflicts of interest were declared. The flow is summarized in the diagram below.
Figure 1 · Flow of the study
Simplified diagram; each step rests on a different level of data.
| Stage | What was done? | What was found? |
|---|---|---|
| Clinical cohort (Emory, October 2022–March 2025) | 47 patients, 109 metastatic lesions; T-DXd 5.4 mg/kg every 3 weeks | Clinical benefit in 31 patients, progression in 16 |
| Spatial transcriptomics (GeoMx DSP) | Two samples each from bone and brain (one responder, one resistant) + a pericardial and a chest-wall sample | S100A7/S100A9 higher in resistant tumors, tumor–immune boundary more compact |
| Cell lines (SKBR3, BT474 and others) | T-DXd ± TTP488; RAGE gene deleted with CRISPR in SKBR3 | IC50 fell markedly |
| Mouse (immunodeficient) | Human SKBR3 cells given into a vein or the heart; 4 weeks of treatment | The combination reduced metastatic foci in the lung |
1. Clinical cohort: what predicts response?
In the cohort, 31 of 47 patients (66.0%) had clinical benefit, defined as complete response, partial response or stable disease at the first RECIST 1.1 assessment; in 16 (34.0%) the disease progressed. The chart below is descriptive; it has no comparison arm.
First assessment result in 47 patients receiving T-DXd
Percent. Colored bar: clinical benefit (complete response, partial response or stable disease); gray bar: disease progression.
Two outcomes of the same cohort; there is no control arm and it is a small single-center series. Because "stable disease" counts as benefit, this proportion should not be read as the rate of tumor shrinkage alone.
| Variable | Relation to response | Value in the paper |
|---|---|---|
| HER2 level (low/high) | Not significant | 81 HER2-low, 18 HER2-high, 10 unknown |
| ER/PR, Ki-67 | Not significant | OR 0.997 for ER/PR percentage |
| Organ of the sampled lesion | Not significant | OR 0.856 (95% CI 0.707–1.037; P = 0.113) |
| Prior T-DM1 use | Associated with weaker response | Univariable OR 4.53 (1.49–13.78; P = 0.008); multivariable OR 6.67 (1.50–29.58; P = 0.013) |
T-DM1 (Kadcyla) had been used before in 16 patients and not in 86 lesions. This does not mean that patients who have had T-DM1 will not benefit; it is an association observed in a small series.
2. Spatial profiling: what do responding and non-responding tumors look like?
The researchers spatially profiled samples from one responding and one non-responding patient for each of bone and brain metastases, plus one pericardial and one chest-wall sample. After quality control, 65 segments and 7,709 genes were analyzed; tumor and immune regions were examined separately. In responding lesions the tumor–immune boundary was longer and interdigitated; in resistant lesions the tumor formed compact masses and immune cells stayed outside (most clearly in the resistant brain sample). The difference in boundary length was P = 5.5 × 10⁻¹² in bone and P = 1.1 × 10⁻¹⁰ in brain; but these values rest on regions from a single patient per group and do not show generalizability across patients.
Figure 2 · Responding and non-responding lesions
Simplified comparison; shows association, not causation.
In gene expression, 314 genes differed in bone and 200 in brain. Shared between the two organs were 22 response-associated genes (such as the GST family, FABP7, HMGCS2, LGALS3BP) and 10 resistance-associated genes (genes related to inflammation and the cytoskeleton, S100A7 and S100A9). S100A7/S100A9 were higher in resistant tumors in all three of the bone, brain and soft-tissue samples.
3. Independent data and proteomics
In an external dataset (KM-Plotter) of 1,936 breast cancer patients treated with chemotherapy, a high S100 score was associated with shorter progression-free survival (HR 1.15–1.58; log-rank P = 0.0002); there was no difference between HER2-positive and negative patients (P = 0.1039). Because these patients received chemotherapy rather than T-DXd, S100 may also be a general marker of poor outcome. In the authors' own spatial cohorts, the correlation between S100 score and progression-free survival was not significant (Spearman R = −0.77; P = 0.103).
In the proteomic analysis, the resistant bone lesion had higher S100P and CD36 and lower topoisomerase I, the target enzyme; responding lesions had higher Hsp27, MX1 and AIF. The resistant brain lesion, in contrast, had higher HER2, S100A8/A9, fibronectin and tuberin, with topoisomerase I unchanged. So in the resistant brain lesion the target (HER2) had not decreased. Positive correlations were found between S100P and phosphorylated HER2 in bone (r = 0.604) and between S100A8/A9 and total HER2 in brain (r = 0.723) (P < 0.001 for both). Total RAGE level did not differ between groups; however, phosphorylated AKT1 (P = 0.0014), ERK1/2 (P = 0.009) and STAT3 (P = 0.0067) were reported higher in resistant lesions. These are associations; they do not show that S100 switches these pathways on.
4. Cell experiments: T-DXd is stronger when RAGE is blocked
HER2-positive SKBR3 and BT474 cells were exposed for 72 hours to T-DXd (0.01–10 µg/mL), with and without 0.5 µM TTP488, which is not toxic on its own. With TTP488, the IC50 fell about 48-fold in SKBR3 and about 700-fold in BT474 (ratios calculated from the reported values). In the discussion the authors say "nearly two orders of magnitude"; the calculated difference for SKBR3 does not match this.
IC50 of T-DXd (nM): without and with TTP488
A lower value means less drug is needed to affect half of the cells. Colored bar: T-DXd + TTP488; gray bar: T-DXd alone.
Two arms of the same experiment in each cell line; cell-culture result. Absolute difference: about 12.7 nM in SKBR3 and about 28 nM in BT474; relative reduction about 48-fold and 700-fold. Bar widths are scaled to 30 nM.
BT474 cells showed marked S100 expression while S100 was very low in SKBR3; even so, SKBR3 was also sensitized by TTP488. This suggests that the effect may not depend only on S100 ligands (this interpretation is ours). When the RAGE gene was deleted by CRISPR in SKBR3, the IC50 fell about 11.7-fold and p65 phosphorylation decreased; adding TTP488 to the deleted cells brought no further sensitization, which supports that the drug acts through RAGE. The combination increased cleaved caspase-3, a marker of apoptosis, and γH2AX, a marker of DNA damage, and reduced BCL-2. Sensitization was also seen in the MB231-HER2 and 4T1.2-HER2 lines. TTP488 was the only RAGE-blocking substance used; no experiment was done in which S100 proteins were added to the cells.
5. Mouse experiment: the combination in lung metastasis
Female immunodeficient (nude) mice were given 1.75 × 10⁵ SKBR3 cells through the left ventricle of the heart or the tail vein. Starting on day 3 and for 4 weeks, they received vehicle, T-DXd (10 mg/kg intravenously), TTP488 (3 mg/kg, orally, daily) or both together. This mouse dose cannot be compared directly with the human dose of 5.4 mg/kg.
| Measure (lung) | Comparison | Result |
|---|---|---|
| Metastatic foci per section | Combination – vehicle | P = 0.0092 |
| Metastatic foci per section | Combination – T-DXd alone | P = 0.01 |
| Metastatic foci per section | TTP488 alone – vehicle | P = 0.09 (not significant) |
| Mean lesion size | Combination – T-DXd alone | P = 0.0013 |
| Total lesion area | Combination – vehicle / T-DXd alone | P = 0.0061 / P = 0.11 (the latter not significant) |
Cleaved caspase-3 in the tissue was about 2.3-fold higher in the combination group than in vehicle (P < 0.001) and also higher than in the T-DXd-alone group (P < 0.001).
Cleaved caspase-3 in lung metastases (marker of apoptosis)
Fold increase relative to the vehicle group. Colored bar: T-DXd + TTP488; gray bar: vehicle (control).
Two arms of the same experiment. The absolute difference is about 1.3 fold-units; the "about 2.3-fold" value is taken from the paper's text, and the number of animals is given differently as n = 3 in the figure legend and n = 10 in the methods. Bar widths are scaled to 3-fold.
Although the abstract mentions "lung and brain metastasis models," the text gives numerical results only for the lung. The immunocompetent mouse model (4T1.2-HER2, implanted in the mammary fat pad) did not work: tumors regressed spontaneously and disappeared completely in 12 of 20 mice by day 14; the authors attribute this to rejection of the human HER2.
The effect of the combination on immunity is therefore unknown. Whether similar pathways apply to other antibody-drug conjugates (sacituzumab govitecan, datopotamab deruxtecan) is also unknown. For the general logic of resistance, see our article on treatment resistance and on PARP inhibitor resistance.
According to the Enhertu article on our site, T-DXd has been licensed in Turkey since May 2024 (license holder Daiichi Sankyo); the summary of product characteristics lists HER2-positive and HER2-low metastatic breast cancer, HER2-positive gastric cancer and HER2-mutant lung cancer, and there is no reimbursement for any indication. Patients pay for the drug themselves; court rulings are no guarantee. The same article gives a price of 44,538.10 TL for a 100 mg vial as of March 2026. We could not independently verify this information against TİTCK and SUT records. TTP488 (also known as azeliragon) is not approved for breast cancer in any country; it has no license or access route in Turkey, and its use is limited to research. For the general framework on unlicensed drugs, see our articles on drugs imported from abroad, the access map and the SUT guide.
Practical advice for patients and their families
- This study does not describe a treatment or test that can be used today. If you are taking T-DXd, do not stop the drug on your own; talk to your oncologist about your response and side effects.
- If you have had T-DM1 before, consider this information together with your oncologist in your treatment plan. The association in the study rests on 16 patients and is not a reason to rule out T-DXd.
- Discuss next-generation sequencing, biopsy and tumor mapping decisions with your physician. There is no routine test for S100 or RAGE today.
- Do not try supplements claimed to "block" RAGE or S100, or ways of obtaining TTP488/azeliragon. During cancer treatment, do not take supplements without asking your oncologist.
- New approaches pass through preclinical and clinical stages; for eligible patients, clinical trials may be an option. In humans the first step is usually a phase 1 trial.
Frequently asked questions
- Why does T-DXd resistance arise, and what does this study add?
- Resistance can have several causes; drug efflux pumps, changes in the target enzyme and tumor signaling programs have each stood out in different studies. This study showed that S100–RAGE signaling stands out in lesions that do not respond. The finding was supported by cell and mouse experiments, but causality has not been proven in humans.
- Can TTP488 (azeliragon) be used together with T-DXd?
- No, this combination has not been tried in patients and is not approved for breast cancer. The paper states that the drug has phase 2/3 safety data in humans; however, this does not show that the combination is safe or effective. Its use can only be considered within a clinical trial.
- Does being HER2-low mean I will not respond?
- No. In this cohort HER2 level did not significantly distinguish response, meaning responses were also seen in HER2-low patients. Even so, T-DXd use is determined by the HER2 result, and your oncologist makes the decision.
- I have had T-DM1 before; will T-DXd not work?
- That is not certain. In the study prior T-DM1 use was associated with a weaker response, but this rests on 16 patients and the confidence interval is wide. The treatment decision is made by the oncologist looking at the disease as a whole.
Glossary
- Antibody-drug conjugate (ADC)
- A type of drug in which an antibody that recognizes a target on the cancer cell carries a chemotherapy payload.
- T-DXd (trastuzumab deruxtecan)
- An ADC that links a topoisomerase I inhibitor to an antibody that recognizes HER2; brand name Enhertu.
- S100 proteins
- A family of calcium-binding proteins that can also act as signals of damage and inflammation (such as S100A7, S100A8, S100A9, S100P).
- RAGE
- A surface receptor to which molecules such as S100 proteins can bind and which can influence pathways inside the cell such as NF-κB, ERK, AKT and STAT3.
- TTP488 (azeliragon)
- An orally taken investigational substance that blocks RAGE; not approved for breast cancer.
- Spatial transcriptomics (GeoMx DSP)
- A method that measures gene expression separately in selected regions (tumor, immune) of a tissue section.
DROZDOGAN Academy Commentary
This study suggests that explaining T-DXd non-response by HER2 level alone is not enough and that the tumor's own signaling program may also play a role, but it remains a hypothesis-level study.
⚡ Why it matters
T-DXd is widely used and some patients do not respond. That an orally taken substance with human safety data (TTP488) targets a resistance pathway offers a concrete idea for a clinical trial.
✅ Strengths
Real patient lesions, different organs such as bone and brain, spatial and proteomic data combined with cell and mouse experiments, and the demonstration through RAGE gene deletion that the drug acts on its intended target. The authors state their limitations openly.
⚠️ Limitations
The spatial analysis rests on 4 samples (plus 2) and the P values were calculated at the region level rather than the patient level, which can lead to overly optimistic P values. The S100 score was not significantly linked to survival in their own cohorts; the external data come from patients treated with chemotherapy. TTP488 worked even though S100 was very low in SKBR3, which suggests that the S100–RAGE reading may not be the only explanation (our interpretation). The mouse experiments were done in immunodeficient animals, the immunocompetent model failed, brain metastasis results were not given numerically in the text, and the 10 mg/kg mouse dose is not equivalent to the clinical dose.
🩺 In practice
No change is recommended. Measuring S100 or RAGE is not in clinical use; the T-DXd decision continues to follow approved indications and the oncologist's assessment.
❓ Open questions
Is S100 elevation specific to T-DXd, or a general marker of poor outcome? Does S100 really act through RAGE? Would the combination work in an immunocompetent model and in patients? What are the right dose and the safety profile?
Conclusion: S100–RAGE is an interesting candidate for T-DXd resistance that looks targetable with a drug; however, the evidence is limited to an association in a small cohort and to effects at cell and mouse level. It is a sufficient reason to propose a clinical trial, but not to change treatment.
- This article was prepared in October 2026.
- The data were taken from the full text in PubMed Central (PMC13495545; publication date in the record: July 17, 2026); the Houston Methodist press release and the ecancer news item (August 3–4, 2026) were used for comparison. Because the full text was read through an automated summarizing tool, we recommend comparing the key numbers with the paper again before publication.
- That TTP488 is azeliragon is stated not in the paper but in company and press sources. The information that azeliragon did not show efficacy in an Alzheimer's phase 3 trial and that phase 2 trials in glioblastoma were ongoing rests on an April 2024 interview with a company executive; the current status could not be verified.
- The license, reimbursement and price information for Turkey was taken from the Enhertu article on our site and has not been independently verified against TİTCK and SUT records. No exhaustive literature search was done for the "first time?" box.
- The paper has the following inconsistencies, and the values were transferred as they are: the mouse T-DXd schedule is every 3 days in the methods and twice weekly in the results; the number of animals is 10 in the methods and 3 in the figure legend; the phrase "nearly two orders of magnitude" does not match the calculated roughly 48-fold fall for SKBR3; there are unexplained differences between the number of regions (102) and the number of segments after quality control (65), and in the sum of the HER2 categories (99; 81/18/10); the reported confidence interval for phosphorylated AKT1 could not be interpreted. The direction of the reported OR for T-DM1 (non-response or response) should be checked separately against the table.
- The comparisons in the charts each belong to arms of the same experiment; no comparison was made between different studies. The diagram is derived from the text and contains no new data. The Academy commentary is our own.
Sources
- Dong W, Wang S, Chan B, Vasquez M, Cui JZM, Sampson J, Sannes C, Yang W, Puri A, Chang J, Zhao H, Li X, Wong STC. Targetable S100-RAGE signaling mediates intrinsic trastuzumab deruxtecan resistance in metastatic breast cancer. Clin Cancer Res, 2026. doi.org · PMC13495545 · PubMed
- Houston Methodist. Scientists identify promising new target behind treatment resistance in advanced breast cancer (press release, August 2026). houstonmethodist.org
- ecancer. Scientists identify promising new target behind treatment resistance in advanced breast cancer (August 4, 2026). ecancer.org
- Mosele F, et al. Trastuzumab deruxtecan in metastatic breast cancer with variable HER2 expression: the phase 2 DAISY trial. Nat Med 2023. PubMed
- Mechanisms of resistance to trastuzumab deruxtecan in breast cancer elucidated by multi-omic molecular profiling. npj Breast Cancer, 2026 (Caris Life Sciences). PMC12774910
- Clinical Trial Vanguard. Cantex Pharma's azeliragon therapy: a potential cancer breakthrough (company executive interview, 2024). clinicaltrialvanguard.com
Was this article helpful?
Your feedback helps us improve our content.
