As antibody-drug conjugates (ADCs) expand across tumor types – and begin to work even at low levels of biomarker expression – the demands on diagnostics are rapidly evolving. Traditional approaches to biomarker assessment are being pushed to their limits, opening the door to computational pathology, multiplex assays, and more quantitative imaging strategies. 

We spoke with Rob Monroe, Chief Medical Officer at Leica Biosystems, about the analytical advances driving this transition and what the next generation of precision diagnostics might look like.

How would you describe the current state of the ADC field, and what factors do you see as most influential in shaping its direction right now?

The ADC field is among the fastest-growing areas of oncology therapeutic development, with hundreds of clinical trials underway targeting dozens of different tumor antigens with antibody-drug conjugate technology (1) This growth trajectory is reflected in broader market projections: the oncology companion diagnostics market is forecast to expand at a compound annual rate of 8.7 percent, reaching $8.4 billion by 2030 (2).

Recent regulatory successes have been a major catalyst, including approvals of several new ADCs – HER2-targeting agents, as well as those directed against Claudin 18.2, folate receptor, and NECTIN4 – spanning multiple tumor types including lung, breast, and bladder cancers (3).

A second major driver has been the dramatic improvement in linker technology, which connects the cytotoxic payload to the monoclonal antibody and enables more precise, site-specific release of the toxin within the tumor (4). These advances have significantly enhanced targeting precision and local drug delivery. While ADCs have existed for more than two decades, these recent technological refinements have produced dramatic gains in efficacy and propelled the field's rapid growth. We are also beginning to see growing interest in combining ADCs with immune checkpoint inhibitors, which is in turn driving the need for diagnostics capable of simultaneously detecting ADC target antigens such as HER2 and TROP2 alongside immune checkpoint biomarkers like PD-L1 and PD-1.

In your view, what features of ADC mechanism or clinical use make accurate biomarker assessment especially important?

The key mechanistic feature of ADCs is their ability to target protein biomarkers – typically surface proteins on tumor cells. Because ADCs target proteins rather than genomic alterations such as mutations or gene amplifications, they can remain effective even when that protein is expressed at relatively low levels (5). This fundamentally changes the standard for biomarker assessment.

Unlike traditional HER2-targeted therapies such as Herceptin, which require high protein expression, ADCs deliver a cytotoxic payload directly into the cell upon internalization. They do not depend on high target expression to be effective. Once internalized, the chemotherapeutic payload kills the tumor cell and can also be released into the surrounding tumor microenvironment, extending its effect to neighboring cells that may not themselves express the ADC target. This "bystander effect" is a major driver of ADC efficacy (6).

Because even low biomarker levels can determine whether an ADC is clinically effective for a given patient, the ability to detect small amounts of target protein with precision becomes critical. Accurate biomarker assessment is therefore not merely a technical requirement – it is a gatekeeper for patient access to potentially life-saving therapies. Improving upon semi-quantitative and qualitative approaches that may overlook patients with low expression is an urgent priority for the field.

How do you view the role of traditional IHC in assessing biomarkers relevant to ADCs, and in what situations might its limitations become more apparent?

IHC has evolved to become the preferred method for assessing ADC biomarkers and guiding patient selection. Many early ADCs were approved without formal companion diagnostics because their target biomarkers were broadly expressed across the relevant tumor type. However, as multiple ADCs are now being developed for the same cancers, it is increasingly important to identify which ADC is most appropriate based on differential protein target expression. IHC is widely available, well-validated, and amenable to automation – Leica Biosystems has invested in automating IHC workflows to make it the most direct and reproducible approach for protein expression assessment in clinical laboratories.

That said, IHC has real limitations, particularly with respect to dynamic range and sensitivity at low expression levels. This has become especially apparent with HER2, where the concept of "HER2-low" and now "HER2-ultralow" expression is pushing the boundaries of assays that were originally designed and optimized for high protein expression. Traditional IHC scoring systems were not built with this sensitivity in mind, and as ADCs continue to expand the therapeutically relevant range of target expression, these limitations become more consequential.

What does computational pathology contribute to the evaluation of ADC targets that may be difficult to achieve through conventional microscopy alone?

Computational pathology, built on digital imaging and AI-powered analysis, provides capabilities that are simply beyond what conventional microscopy can deliver. One important contribution is helping pathologists navigate borderline cases more confidently – for example, applying AI algorithms to assess PD-L1 expression at clinically significant thresholds such as 1 percent or 50 percent tumor cell positivity, where a wrong call can directly affect treatment access (7).

Another key benefit is the ability to distinguish true biomarker staining from background or nonspecific signal – a persistent challenge in IHC, and one that becomes especially problematic at low expression levels where the human eye struggles to make reliable calls. Computational pathology improves sensitivity and specificity in these borderline scenarios and enhances reproducibility across institutions, so that different pathologists reviewing the same slide consistently arrive at the same – and more accurate – conclusion.

Perhaps most significantly, computational pathology enables biomarker assessments that cannot be performed by the human eye at all. A compelling example is AstraZeneca's Quantitative Continuous Scoring (QCS) system for the ADC target TROP2, which received FDA Breakthrough Device Designation and uses a computational approach to measure a normalized membrane-to-cytoplasm staining ratio as a surrogate for target internalization (8). This ratio cannot be determined through visual microscopy. If an ADC is approved using a QCS-based cutoff, the companion diagnostic would rely entirely on computational output – a paradigm shift illustrating how digital pathology is moving from support tool to essential infrastructure.

How are newer staining and imaging approaches influencing the way researchers assess biomarkers related to ADC development?

Beyond quantitative computational scoring systems, fluorescence-based imaging is emerging as an important alternative to conventional chromogenic staining for ADC biomarker research. Fluorescence imaging offers a substantially greater dynamic range, enabling finer discrimination of protein expression levels and supporting patient stratification at much lower expression thresholds than traditional chromogenic IHC.

Fluorescent approaches can also be used to correlate signal intensity with defined protein quantities, enabling estimation of actual protein concentrations within tumor cells – not just relative staining patterns. This is particularly valuable for ADC development, where understanding the absolute level of target expression in a tumor may help predict whether payload delivery will be sufficient for a therapeutic effect.

In parallel, there is growing interest in chromogenic dyes that offer some of the same advantages as fluorescent stains – including more linear relationships between staining intensity and actual protein levels – while remaining more compatible with existing brightfield imaging infrastructure in clinical laboratories. Both approaches are active areas of translational ADC research aimed at resolving low-level expression with greater precision than is achievable with conventional methods.

What developments in analytical technologies – whether in imaging, assay chemistry, or spatial methods – are having the greatest impact on how ADC targets are measured and interpreted?

Across imaging, assay chemistry, and spatial biology, several converging developments are reshaping how ADC biomarkers are measured and interpreted.

In imaging and computational analysis, AI algorithms applied to digitized IHC images can now identify subtle differences in the pattern, spatial distribution, and overall quantity of protein expression between patient populations that respond to an ADC and those that do not. These computational approaches open up biomarker development possibilities that were inaccessible through conventional microscopy and traditional IHC scoring. The TROP2 QCS method is the leading clinical example of this paradigm, and its regulatory success is likely to catalyze similar computational approaches for other ADC targets.

In assay chemistry, fluorescent staining and advanced chromogenic dyes both offer improvements in quantitative precision that are driving translational research. The ability to generate more linear correlations between staining intensity and actual protein levels is critical for developing robust, analytically sensitive companion diagnostics.

In spatial biology, multiplex assays capable of assessing several biomarkers simultaneously – and capturing information about the tumor microenvironment – are among the most impactful emerging tools. Evaluating the spatial arrangement of immune cells relative to tumor cells can inform predictions of ADC efficacy: for instance, understanding whether ADC target-expressing cells are spatially clustered or intermixed with non-expressing cells helps predict whether the bystander effect will meaningfully extend cytotoxic activity. Parallel assessment of H&E images alongside IHC can also provide inferred genomic and transcriptomic information that complements biomarker interpretation. Together, spatial approaches represent an important frontier for refining patient selection beyond what protein expression alone can provide.

How is artificial intelligence being used to analyze biomarker expression in the context of ADCs, and what kinds of insights can it help uncover? Are there considerations around validation or interpretability?

AI is being applied to ADC biomarker analysis in two primary and complementary ways. First, it delivers more accurate and reproducible assessment of known biomarkers – improving tumor cell scoring, quantifying percentages of positive cells, and reducing inter-pathologist and inter-laboratory variability that can otherwise lead to inconsistent patient selection. Studies have shown that pathologists' interpretations can vary substantially from one reader to another and from one lab to the next, with real consequences for patients who may or may not qualify for a given therapy.

Second, AI enables entirely novel forms of biomarker analysis that go beyond what human reviewers can extract from a slide. AstraZeneca's QCS system for TROP2 is the defining example: AI measures the relative distribution of biomarker expression across the membrane and cytoplasm, generating a normalized membrane ratio that predicts patient response – a biological insight that simply cannot be captured through manual scoring. The FDA's granting of Breakthrough Device Designation for the TROP2 QCS assay validates this approach and signals the opening of a new chapter for computational pathology companion diagnostics.

With regard to validation, any AI-generated biomarker result used for clinical decision-making requires rigorous analytical and clinical validation. Interpretability is also an active consideration – as AI-derived scores become increasingly abstract (ratios, spatial interaction metrics, and multimodal features), it is critical that they are mechanistically linked to the underlying biology and clinically validated against treatment outcomes. Regulatory agencies including the FDA are increasingly engaged as partners in defining these validation frameworks, a development that is accelerating the field's progress.

What potential do you see for approaches that assess multiple ADC targets within the same assay, and what would be necessary to make this practical in research or clinical settings?

The potential for multiplex ADC biomarker assays is substantial – and increasingly urgent. We are rapidly approaching a situation where multiple ADCs, potentially five or more, may be approved for a single tumor type such as lung, breast, bladder, or colorectal cancer. Testing each target individually with a separate IHC assay is neither tissue-efficient nor cost-effective, and it will simply be impractical at scale. Multiplex methods that can evaluate several ADC targets simultaneously on the same tissue section represent an essential future solution.

In research settings, multiplexing is already well established. Fluorescent technologies, validated antibodies, and imaging platforms capable of multi-target assessment are widely used in translational and discovery contexts. The challenge lies in clinical translation. For broad clinical deployment, assays must be simplified: panels should ideally contain no more than three to five biomarkers, staining reagents must be compatible with standard automated IHC platforms available in most labs, and the full workflow from staining through imaging to computational analysis must be seamlessly integrated.

Both chromogenic and fluorescent multiplexing formats have roles to play, with chromogenic approaches likely more accessible to clinical laboratories in the near term given their compatibility with existing brightfield infrastructure. In either format, computational pathology is an absolute prerequisite for interpretation – once assays move beyond a single chromogen or fluorescent channel, the complexity exceeds what conventional microscopy can handle reliably. Close collaboration within diagnostics companies between antibody development, staining platform development, digital pathology, and AI software teams will be essential to bring these integrated solutions to the clinic.

How might digital pathology fit alongside other analytical approaches – such as genomic profiling, flow-based methods, or spatial omics – to build a more comprehensive understanding of ADC response?

For ADC therapeutics, the effector molecule is the surface protein itself – which means protein-level assessment is the most direct and biologically relevant way to evaluate patient suitability for a given ADC. This gives IHC paired with digital computational analysis a fundamental advantage over genomic profiling, which uses DNA copy number or gene amplification as surrogates for protein expression, and over RNA profiling, which measures transcript levels that may not faithfully predict final protein abundance or membrane localization. When the goal is to determine whether an ADC target is present at a therapeutically meaningful level on the tumor cell surface, protein-based measurement is the preferred approach.

That said, digital pathology is most powerful when integrated with complementary modalities. For example, combining IHC with AI-powered H&E analysis can reveal genomic and transcriptomic profiles from standard histology slides – a potentially democratizing development, given that H&E images are universally available while genomic sequencing is not always accessible. AI applied to H&E can, for instance, infer EGFR mutation status, helping prioritize which patients require formal molecular testing.

Flow cytometry offers another complementary approach for assessing protein expression in liquid tumors or dissociated cell preparations, particularly relevant for hematologic malignancies. And as discussed, spatial omics approaches add the important dimension of microenvironmental context – immune cell composition, spatial architecture, and biomarker co-localization – that can refine response prediction beyond protein expression levels alone. Together, these modalities form an increasingly multimodal diagnostic ecosystem, with IHC and digital pathology as the central pillar for solid tumor ADC assessment.

If you look five to ten years ahead, what might ADC biomarker assessment look like, and what developments do you expect will have the most influence?

Looking five to ten years ahead, I expect the most transformative development will be the clinical standardization of multiplex assays that simultaneously evaluate multiple ADC targets in the same tumor specimen. As more ADCs are approved for individual tumor types – a near-certainty given the current pipeline – single-target IHC will become insufficient and inefficient. Multiplex panels assessing three to five ADC targets at once will become standard of care in lung, breast, colorectal, and other high-prevalence cancers, enabled by automated staining platforms paired with AI-powered computational analysis.

Computational pathology will be indispensable in this future. The broader adoption of digital pathology across clinical laboratories – currently still below 20 percent globally – will accelerate, driven by the regulatory approval of AI-enabled companion diagnostics and by demonstrated clinical utility that overcomes current adoption barriers. AI will not only enable pathologists to interpret complex multiplex data but will also introduce fundamentally new biomarker paradigms: quantitative continuous scoring, normalized membrane expression ratios, spatial interaction metrics, and multimodal biomarkers derived from combined IHC, H&E, and potentially radiology data.

The TROP2 QCS and normalized membrane ratio represent early prototypes of this next generation of computational companion diagnostics. Their regulatory success will encourage other pharmaceutical companies to develop analogous approaches for their own ADC programs. Beyond ADCs, this same infrastructure – multiplex protein assays combined with AI-driven analysis – will support the growing field of multispecific antibodies and will deepen the integration of ADC therapy with immune checkpoint inhibition. The vision is a future in which a single, AI-interpreted multiplex assay provides oncologists with a comprehensive molecular portrait of a tumor, enabling them to select the optimal therapy – or combination of therapies – for each individual patient at the right point in their cancer journey. 

Newsletters

Receive the latest analytical science news, personalities, education, and career development – weekly to your inbox.

I have read and understand the Privacy Notice*

Subscribe