Moving companion diagnostics toward spatial patterns rather than single molecules is where biomarker development is headed, but the path from a promising spatial signature to an assay a physician can actually order is long, regulatory as much as scientific, and considerably less traveled than the underlying biology might suggest. Understanding that path, not just the biology behind it, is what separates a spatial biomarker with genuine commercial potential from an interesting research finding.
Key takeaways
|
Why spatial biomarkers are different
A conventional biomarker, such as PD-L1 expression measured by immunohistochemistry, reports a single value: how much of a protein is present. A spatial biomarker reports something structurally different: a pattern, defined by the arrangement, density, or co-occurrence of multiple features across tissue, which cannot be reduced to a single number without losing the information that makes it predictive in the first place.
That difference has direct consequences for validation. A single-molecule assay’s performance can be summarized by sensitivity and specificity against a defined cutoff. A spatial signature, involving multiple markers, their relative positions, and often a computational model that weighs all of them together, requires validating not just each individual measurement but the entire analytical pipeline that turns raw tissue images into a single predictive output, including any algorithm involved. Reader variability, meaning different pathologists or different image analysis systems scoring the same tissue differently, becomes a bigger and more consequential problem as the number of features being jointly assessed increases.
From signature to assay
Turning a discovered spatial signature into a standardized, reproducible assay is a distinct undertaking from discovering the signature itself, and it is where many promising spatial biomarkers stall.
A multidisciplinary approach called AstroPath, developed at Johns Hopkins University and combining principles from immunology, pathology, computer science, and astronomy, illustrates one route through this transition. It was built specifically to lay the foundation for rapid, efficient biomarker discovery through image analysis, turning discovered predictive signatures into analytically and clinically validated assays rather than leaving them as one-off research findings. The astronomy connection is not incidental: techniques originally developed for processing astronomical imaging data at scale transfer surprisingly well to standardizing multiplexed tissue imaging, since both problems involve extracting consistent, quantitative signal from large, complex image datasets.
Image analysis validation specifically remains an area where regulatory clearance lags the underlying science. Beyond a small number of biomarkers, image analysis applications have rarely received regulatory clearance for use as an aid to pathologists, and until image analysis is routinely validated and accepted as a standard component of immunohistochemistry-based companion diagnostic tests, reader variability will remain a critical consideration in verifying any slide-based assay, spatial or otherwise. A spatial biomarker program should treat this not as a solved problem to work around, but as a validation requirement to plan for explicitly from the earliest development stage.
The companion diagnostic path
Two genuinely different regulatory categories exist for a biomarker test, and the distinction matters enormously for a spatial biomarker program deciding which path to pursue.
Category | What it requires | What it enables |
Laboratory-developed test | Performance specifications sufficient to ensure assay consistency, under Clinical Laboratory Improvement Amendments and College of American Pathologists guidelines; historically self-certified by the performing laboratory | Clinical use at the certified laboratory, without full FDA premarket review |
FDA-approved companion diagnostic | Formal FDA review of analytical and clinical validation; for most products, a premarket approval application as a Class III device | Distribution as a labeled in vitro diagnostic device, and inclusion in the paired therapeutic’s FDA label as a required test |
Table 1. The two regulatory categories a spatial biomarker test can occupy. A drug label that mandates an FDA-approved companion diagnostic specifically cannot be satisfied by a laboratory-developed test alone, regardless of the underlying assay’s scientific validity.
That distinction has become considerably more consequential recently. In 2024, the FDA finalized a rule phasing out its general enforcement discretion policy for laboratory-developed tests, establishing a timeline for LDTs to come under full device regulation. This has direct implications for spatial biomarkers specifically: complex multiplex and spatial assays have frequently been run as LDTs precisely because their complexity made a full premarket approval submission burdensome relative to a simpler single-marker test, and many biomarker tests that function as de facto companion diagnostics in clinical practice have never gone through formal FDA clearance or approval at all. A spatial biomarker program built on an assumption that the LDT route remains a stable, lower-effort long-term path should revisit that assumption in light of this change.
For a manufacturer not intending to distribute a device broadly, a single-site premarket approval pathway exists as a narrower route to Class III approval, subject to many of the same quality system requirements as a full in vitro diagnostic manufacturer but applied to one location rather than distributed use. This can be a practical option for a spatial assay initially offered only through a single central laboratory rather than distributed broadly to hospital pathology labs.
Analytical and clinical validation
Regardless of which regulatory path a spatial biomarker follows, the underlying validation discipline is the same in structure, though considerably more demanding in practice than for a single-marker test.
Three requirements recur across spatial biomarker validation specifically.
- A training set genuinely representative of real-world clinical samples. Selecting a training set that reflects the actual samples collected in clinical practice, rather than an idealized or unusually clean research cohort, is critical to ensuring the assay achieves appropriate analytical sensitivity and specificity once deployed beyond the development lab.
- Validation by someone other than the assay’s own developer. The pathologist engaged in validation and verification is typically an expert who has become particularly proficient in scoring that specific assay, not a pathologist who would be scoring it in ordinary clinical practice, which is a documented source of performance overestimation worth explicitly correcting for.
- Explicit characterization of image analysis performance, not just the biomarker’s biological validity. Given how few image analysis applications have received regulatory clearance as pathologist aids, a spatial biomarker program should expect to validate its computational scoring pipeline as rigorously as its underlying biology.
These requirements are not unique to spatial biology in principle, but the number of individually validated components, multiple markers, their spatial relationships, and the algorithm combining them, multiplies the validation burden relative to a single-marker assay in a way that a development timeline should account for explicitly rather than discover midway through.
Commercial and regulatory realities
The historical base rate for the entire companion diagnostic category is worth stating plainly, since it tempers any assumption that spatial biomarkers will translate into approved diagnostics quickly or in large numbers.
Since the first FDA companion diagnostic approval in the late 1990s, approximately 50 to 60 CDx devices have been approved or cleared in the United States altogether, across every modality and every therapeutic area, not just oncology. The approval rate has increased over time, from fewer than one per year through 2010 to roughly three per year from 2011 through 2024, but that remains a modest absolute number for a field now being asked to accommodate an entirely new class of spatial and multiplex assays layered on top of established single-marker tests.
The regulatory pathway split reinforces the same caution. Historical trends indicate most current companion diagnostic products required a premarket approval application, a significantly more rigorous process than 510(k) clearance; among products currently available, only two have received 510(k) clearance or been granted De Novo classification. A spatial biomarker program should plan its regulatory strategy assuming the more demanding premarket approval pathway is the likely route, rather than hoping for the exception.
Three practical implications follow for anyone building a spatial biomarker program with commercial ambitions.
- Budget development timelines against the harder pathway, not the easier one. With only two products having achieved the lighter-touch route among all current CDx products, assuming a premarket approval timeline from the outset avoids a painful later correction.
- Treat the 2024 LDT rule change as a planning input, not a future contingency. A spatial assay intended for real clinical use should be designed with eventual FDA device regulation in mind from early development, rather than assuming indefinite LDT flexibility.
- Expect image analysis validation to be a first-class development workstream. Given how rarely computational scoring tools have themselves received regulatory clearance, a spatial biomarker’s algorithm deserves the same validation rigor and timeline allocation as its underlying biology.
Detailed treatment of each stage introduced here is developed across this section’s companion guides: a foundational primer is available in What Are Spatial Biomarkers? A Primer for Drug Developers, discovery methodology in Spatial Biology in Biomarker Discovery, the development pathway in full in From Spatial Biomarker to Companion Diagnostic: The Development Path, patient stratification applications in Predictive Spatial Signatures for Patient Stratification, and the specific oncology application space in Precision Oncology and Spatial Profiling.
The biomarker-to-combination-therapy logic this hub builds toward is developed in Spatial Biology in Immuno-Oncology: Targets, Signatures, and Combinations, and for where biomarkers and companion diagnostics sit within the full spatial biology pipeline, see Spatial Biology in Drug Discovery: From Target Discovery to Translational Medicine.
This article was produced under Drug Discovery News’s AI editorial policies.
