Blog

The biomarker challenges faced by ADC developers and clinicians, OD-based Quantitative IHC scoring and AI-driven computational pathology move a step closer to the clinic, and immunotherapy continues to deliver outstanding results but only in specific biomarker-positive patient subgroups.

The American Association for Cancer Research (AACR) Annual Meeting 2025 recently concluded, bringing together leading researchers, clinicians, and industry professionals to share the latest advances in cancer research. We noted the growing importance of spatial biology and AI-powered pathology in precision oncology. Below are three critical takeaways from the conference with significant implications for oncology research and development. We look forward to continued discussion of all three points when we reconvene at the American Society for Clinical Oncology (ASCO) meeting in Chicago later this month.

1. We Need Improved Biomarker Scoring Strategies for Antibody-Drug Conjugates (ADCs)

One of the key topics discussed at AACR 2025 was the urgent need for more reliable biomarker scoring strategies for antibody-drug conjugates (ADCs).

Clinical studies of ADCs have been complicated by two categories of challenges:

The first challenge facing ADCs is that many ADCs share similar targets and payloads, and this has resulted in multiple ADCs being approved for overlapping conditions. Clinicians face difficult decisions about treatment sequencing without adequate biomarker guidance. In her AACR presentation, Dr. Sara Hurwitz from the Fred Hutchinson Cancer Center described some examples of this challenge and its impacts on treatment decisions. The National Comprehensive Cancer Network (NCCN) published breast cancer treatment guidelines last month to reflect approved ADCs — Daiichi Sankyo’s trastuzumab deruxtecan (T-DXd), which targets HER2, is approved in the first line setting for hormone receptor (HR)-positive breast cancer. T-DXd is also approved in the second line setting, but so is Gilead’s sacituzumab govitecan and Daiichi Sankyo’s datopotamab deruxtecan (Dato-DXd). The latter two drugs both target TROP2 and all three have similar cytotoxic payloads. Given these overlaps, further biomarker development is needed for clinicians to select the right patients for each drug.

A second challenge facing ADC development is that molecular mechanisms of action and resistance appear to be sufficiently complex such that linear biomarker strategies are not predictive of response. For some ADCs, therapeutic response correlates with target antigen expression, but other ADCs show no such correlation.

Examples of ADC targets showing variable correlation between ADC efficacy and target antigen expression in tumors:

• HER2: Therapeutic efficacy and resistance shows correlation to expression levels and heterogeneity in breast cancer in the case of Trastuzumab emtansine (T-DM1) presumably because it doesn’t have a cleavable linker and no demonstrable bystander effect has been established¹. However, in the case of T-Dxd, where there’s been established evidence of the bystander effect due to its cleavable linker, the drug shows benefit across both high expressors (IHC 3+) and low expressors (IHC 2+, 1+ and 0+)², ³. By contrast, in gastric and colorectal cancers, response to T-Dxd is more tightly correlated to HER2 expression, and only high expressors (IHC 3+) respond⁴.
• TROP2: Various clinical studies of large-scale Phase III trials have shown that there is no clear correlation between response to TROP2-targeted ADCs and TROP2 antigen expression in the patient’s tumor, as measured by traditional IHC scoring methods (e.g., H-score)^5,6; however, AstraZeneca has developed a novel approach to scoring target antigen expression and have demonstrated in an exploratory analysis of the TROPION-01 Phase III trial patients that this novel biomarker score is correlated with response for a specific subtype of non-small cell lung. We’ll cover the latest updates on this approach in the next section on AI-augmented scoring.
• Folate receptor alpha (FRɑ): The approval of the ADC mirvetuximab soravtansine was limited to ovarian cancers expressing the target antigen^{7, 8}.
• Nectin-4: Some data have shown that urothelial cancer patients with amplification of the NECTIN-4 gene or high membranous expression of the NECTIN-4 antigen (e.g., H-scores of 100 to 300) show better response to the Nectin-4-targeted ADC, enfortumab vedotin, and that tumors exhibiting low Nectin-4 expression may be more likely to develop resistance to the drug⁹.
• c-MET: And finally, in recent news, Abbvie’s c-MET-targeted ADC, telisotuzumab vedotin-tllv, received approval to treat patients with previously treated advanced Non-Small Cell Lung Cancer (NSCLC) that have high c-Met protein overexpression (defined as ≥ 50% of tumor cells with strong IHC 3+ staining)¹⁰.

As Dr. Hurwitz described in her presentation, some of these conclusions drawn may be confounded by the limitations of current immunohistochemistry (IHC)-based methods. Variability in antigen expression might be artifactual due to antibody variation, IHC readout variability, and fixation issues; in fact, a study asking 18 pathologists to distinguish between HER2 0 versus HER2 1+ showed only 26% concordance.

These challenges underscore the need for more sophisticated biomarker assessment strategies that can accurately characterize tumor biology by revealing antigen heterogeneity, membrane localization, and overall expression patterns. The hope is that multimodal approaches and deep spatial proteomic analysis may prove to be more predictive of ADC efficacy and help distinguish between ADCs with overlapping targets, payloads, and indications.

The Nucleai team presented their breakthrough work towards better ADC biomarker strategies at AACR 2025 with their poster, “AI-powered spatial inference platform for qualifying ADCs and multispecific therapeutic targets.” Using data from multiplexed immunofluorescence (mIF), the team used Nucleai’s  AI-augmented platform to elucidate ADC-specific mechanisms of action and spatial biomarker signatures. Specifically, the data revealed the spatially resolved expression levels of ADC targets like HER2, B7H3, CEACAM5, and CDH3 across different indications. The study also showed that interaction between tumor and stromal markers and the colocalizations of tumor markers could be used to better inform decisions during the development of multispecific therapeutics.

Download the poster for a first-hand look at the data.

2. AI-Augmented Biomarker Scoring Has Entered Prime Time

Our second major takeaway from AACR 2025 was that AI-powered computational pathology could be on the verge of entering clinical practice.

AstraZeneca recently made headlines as the first pharmaceutical company to implement an AI-enabled biomarker platform, Quantitative Continuous Scoring (QCS), in a prospective Phase III trial setting. Their approach focused on a novel computational algorithm to measure TROP2 expression for Dato-DXd, addressing the challenges in traditional TROP2 quantitation highlighted above¹¹. To scale this biomarker strategy, AstraZeneca partnered with Roche to develop a companion diagnostic (CDx) test. During AACR this year, Roche announced that the FDA had granted it Breakthrough Device Designation making it the first AI-driven CDx for non-small cell lung cancer.

Some of the key advantages of AI-augmented biomarker scoring include:

• Standardized, quantitative assessment that eliminates inter-observer variability
• Capacity to detect subtle patterns of expression that may predict response to a drug
• Ability to analyze spatial relationships between tumor cells and their microenvironment, thereby modeling the potential bystander effect of a drug

At Nucleai, we’ve enabled even richer insights by developing a suite of deep learning and computer vision models for quantitative scoring of standard pathological and novel spatial biomarkers and co-expression/colocalization analysis from IHC and multiplex IHC data, coupled with an easy-to-use, interactive analysis studio for rapid interpretation of ADC datasets.

See the platform in action at the American Society for Clinical Oncology (ASCO) Meeting, Booth #IH02 — contact us to book a demo.

3. New Immunotherapy Approaches Show Pre-Surgery Efficacy While Highlighting Need for Better Response Prediction

The third key insight from AACR 2025 revolved around remarkable advances in neoadjuvant immunotherapy. Researchers presented updated data on dostarlimab showing unprecedented complete response rates in mismatch repair–deficient (MMRd) solid tumors when administered before surgery. In the dostarlimab study, 92% of patients, including those with rectal and non-rectal cancers, showed no evidence of disease after two years, with 64% of non-rectal cancer patients remaining cancer-free after one year¹². The treatment offers a promising alternative to traditional invasive methods like surgery, chemotherapy, and radiation, with patients experiencing fewer and more manageable side effects compared to conventional treatments.

The consensus among researchers is that integrated approaches combining genomics, proteomics, and AI-powered spatial analysis may identify additional patient subsets who could benefit from these treatments. Tumor spatial context, including immune cell distribution, stromal interactions, and cellular neighborhoods, plays a crucial role in treatment response. This represents a significant opportunity for spatial biology platforms that can analyze these complex patterns.

At AACR, we presented data from our ongoing work with Dr. Arutha Kulasinghe on tumor microenvironment profiling in head and neck cancer using Nucleai’s spatial biomarker platform. Our studies revealed that both the composition of the immune microenvironment and tumor-intrinsic metabolic states played critical roles in modulating response to PD-1 blockade.

Download the poster for a first-hand look at the data.

Looking Ahead: The Need for Better Biomarkers in Advancing Precision Oncology

These three takeaways from AACR 2025 converge on a single imperative: the need for more sophisticated biomarker strategies in oncology. As ADCs, bispecifics and immunotherapies continue to transform cancer treatment, the limitations of traditional biomarker assessment have become increasingly apparent.

Nucleai’s AI-powered spatial biomarker platform addresses these challenges head-on by providing standardized, quantitative analysis of the tumor microenvironment and biomarker expression patterns. Our technology enables researchers and clinicians to move beyond simple presence/absence or H-score assessments to granular, quantitative OD-based IHC scoring and spatial proximity analyses to understand the complex biomolecular and cellular interactions that influence treatment response.

As we continue to partner with leading pharmaceutical companies and research institutions, we remain committed to advancing the field of precision oncology through cutting-edge spatial analysis. The insights from AACR 2025 confirm that we’re on the right path—developing solutions that address the most pressing challenges in cancer treatment today.

Book a meeting with us at ASCO 2025 to learn more.

References:

1. Filho, Otto Metzger, et al. “Impact of HER2 Heterogeneity on Treatment Response of Early-Stage HER2-Positive Breast Cancer: Phase II Neoadjuvant Clinical Trial of T-DM1 Combined with Pertuzumab.” Cancer Discovery, vol. 11, no. 10, Oct. 2021, pp. 2474–87. PubMed, https://doi.org/10.1158/2159-8290.CD-20-1557.

2. Modi, Shanu, et al. “Trastuzumab Deruxtecan in Previously Treated HER2-Low Advanced Breast Cancer.” The New England Journal of Medicine, vol. 387, no. 1, July 2022, pp. 9–20. PubMed, https://doi.org/10.1056/NEJMoa2203690.

3. Bardia, Aditya, et al. “Trastuzumab Deruxtecan after Endocrine Therapy in Metastatic Breast Cancer.” The New England Journal of Medicine, vol. 391, no. 22, Dec. 2024, pp. 2110–22. PubMed, https://doi.org/10.1056/NEJMoa2407086.

4. Yoshino, Takayuki, et al. “Final Results of DESTINY-CRC01 Investigating Trastuzumab Deruxtecan in Patients with HER2-Expressing Metastatic Colorectal Cancer.” Nature Communications, vol. 14, no. 1, June 2023, p. 3332. PubMed, https://doi.org/10.1038/s41467-023-38032-4.

5. Bardia, A., et al. “Biomarker Analyses in the Phase III ASCENT Study of Sacituzumab Govitecan versus Chemotherapy in Patients with Metastatic Triple-Negative Breast Cancer.” Annals of Oncology: Official Journal of the European Society for Medical Oncology, vol. 32, no. 9, Sept. 2021, pp. 1148–56. PubMed, https://doi.org/10.1016/j.annonc.2021.06.002.

6. Rugo HS, Bardia A, Marmé F, et al: Sacituzumab govitecan vs treatment of physician’s choice: Efficacy by Trop-2 expression in the TROPiCS-02 study of patients with HR+/HER2– metastatic breast cancer. 2022 San Antonio Breast Cancer Symposium. Abstract GS1-11. Presented December 6, 2022.

7. Martin, Lainie P., et al. “Characterization of Folate Receptor Alpha (FRα) Expression in Archival Tumor and Biopsy Samples from Relapsed Epithelial Ovarian Cancer Patients: A Phase I Expansion Study of the FRα-Targeting Antibody-Drug Conjugate Mirvetuximab Soravtansine.” Gynecologic Oncology, vol. 147, no. 2, Nov. 2017, pp. 402–07. PubMed, https://doi.org/10.1016/j.ygyno.2017.08.015.

8. Moore, K. N., et al. “Phase III, Randomized Trial of Mirvetuximab Soravtansine versus Chemotherapy in Patients with Platinum-Resistant Ovarian Cancer: Primary Analysis of FORWARD I.” Annals of Oncology: Official Journal of the European Society for Medical Oncology, vol. 32, no. 6, June 2021, pp. 757–65. PubMed, https://doi.org/10.1016/j.annonc.2021.02.017.

9. Klümper, Niklas, et al. “Membranous NECTIN-4 Expression Frequently Decreases during Metastatic Spread of Urothelial Carcinoma and Is Associated with Enfortumab Vedotin Resistance.” Clinical Cancer Research: An Official Journal of the American Association for Cancer Research, vol. 29, no. 8, Apr. 2023, pp. 1496–505. PubMed, https://doi.org/10.1158/1078-0432.CCR-22-1764.

10. Camidge, D. Ross, et al. “Telisotuzumab Vedotin Monotherapy in Patients With Previously Treated C-Met Protein-Overexpressing Advanced Nonsquamous EGFR-Wildtype Non-Small Cell Lung Cancer in the Phase II LUMINOSITY Trial.” Journal of Clinical Oncology: Official Journal of the American Society of Clinical Oncology, vol. 42, no. 25, Sept. 2024, pp. 3000–11. PubMed, https://doi.org/10.1200/JCO.24.00720.

11. 11 Normalized Membrane Ratio of TROP2 by Quantitative Continuous Scoring is Predictive of Clinical Outcomes in TROPION-Lung 01 Garassino, M.C. et al. Journal of Thoracic Oncology, Volume 19, Issue 10, S2 – S3

12. Cercek, Andrea, et al. “Nonoperative Management of Mismatch Repair-Deficient Tumors.” The New England Journal of Medicine, Apr. 2025. PubMed, https://doi.org/10.1056/NEJMoa2404512.