ROS1 Rearrangement in Non-Small Cell Lung Cancer: A Predictive Biomarker Worth Getting Right

Introduction

Of all the actionable alterations we look for in lung adenocarcinoma, ROS1 rearrangement is one I never want to miss. It's uncommon — present in roughly one to two percent of non-small cell lung cancers — but the therapeutic payoff for the patients who carry it is substantial. This is a predictive biomarker in the strict sense: it doesn't tell us how the tumor will behave on its own, but it does tell us whether a specific class of targeted drugs is likely to work. That distinction matters, and I'll return to it. What follows is a practical walk through what the test measures, how we run it, and where the interpretation still gives us trouble.

What the Test Measures

ROS1 is a receptor tyrosine kinase. In the relevant tumors, a chromosomal rearrangement fuses the ROS1 kinase domain to one of several partner genes — CD74 is the classic one, but there are many. The fusion produces a constitutively active kinase that drives proliferation independent of normal ligand signaling. That's the vulnerability we exploit: a tumor addicted to aberrant ROS1 signaling is one that can be shut down by blocking the kinase [1].

So the test is really answering a single binary question. Is a ROS1 fusion present, or isn't it?

How It's Tested

Testing typically starts on formalin-fixed, paraffin-embedded (FFPE) tissue, though circulating tumor DNA (ctDNA) from plasma is increasingly used when tissue is scarce or a rebiopsy isn't feasible.

The workflow usually runs in stages. Immunohistochemistry (IHC) for ROS1 protein is a reasonable, inexpensive screen — high sensitivity makes it a good first filter. But here's the catch, and it's a real trap in practice: ROS1 IHC can produce false-positive staining, in part through non-specific antibody cross-reactivity, so a positive IHC result is never the end of the story. It must be confirmed molecularly. The confirmatory step is non-negotiable.

Confirmation comes from either fluorescence in situ hybridization (FISH) or next-generation sequencing (NGS). FISH uses break-apart probes to detect the physical rearrangement regardless of partner. NGS — particularly RNA-based assays — can identify the fusion and often name the partner gene, which is useful because RNA-based sequencing captures fusion transcripts that DNA-based panels sometimes miss when the breakpoints fall in large intronic regions.

Scoring, at the end of all this, is deliberately simple. Fusion present versus absent. The complexity lives in the workflow, not the readout.

What Each Result State Means

ROS1 fusion present. A molecularly confirmed fusion identifies a tumor driven by aberrant ROS1 signaling. This is the result that opens a therapeutic door.

ROS1 fusion absent. No confirmed rearrangement means the tumor is not ROS1-driven, and ROS1-directed therapy would not be expected to help. In these cases attention turns to other biomarkers.

And the awkward middle case — IHC-positive but molecularly unconfirmed — is not a result at all. It's an incomplete test. Treat it as such until the confirmatory assay reports.

What It Determines for Treatment Eligibility

A confirmed ROS1 fusion informs eligibility for the class of drugs known as ROS1 tyrosine kinase inhibitors (TKIs). This is where the "predictive" label earns its keep: the biomarker gates access to a therapy class whose benefit is tied directly to the presence of the target [1,2].

The clinical rationale rests on trial data in ROS1-positive NSCLC, where these agents have produced meaningful and often durable responses — the kind of activity that led to regulatory approvals in this setting, including the approval of repotrectinib for locally advanced or metastatic ROS1-positive disease [2]. I'll be explicit about the boundary here: identifying the fusion establishes eligibility for the drug class. It does not constitute treatment advice, and decisions about any specific agent belong to the treating oncologist and patient.

Caveats and What's Evolving

A few things keep me cautious.

First, the IHC-confirmation issue bears repeating because it's the most common error I see. A positive screen is a lead, not a diagnosis. Skip the molecular confirmation and you risk directing a patient toward targeted therapy on the strength of a stain that cross-reacted.

Second, assay platforms don't detect everything equally. FISH confirms rearrangement but won't name the partner. DNA-based NGS can miss fusions with breakpoints in difficult intronic territory. RNA-based NGS improves fusion detection but depends on adequate RNA quality — and FFPE is not kind to RNA. When results conflict across platforms, the case deserves a second look rather than a reflexive call.

Third, ctDNA is a genuine advance for tissue-limited cases, but a negative plasma result never rules out a fusion. Tumors shed DNA at variable rates. Absence of a detected fusion in plasma is not absence of the fusion in the tumor.

Finally, resistance is the frontier. ROS1 TKI responses, while often durable, are not permanent, and the resistance mechanisms that emerge are an active area of study [1,2]. This part of the field is moving quickly, and today's testing algorithm may not be next year's.

A Closing Pearl

If you remember one thing, make it this: in ROS1 testing, the screen and the confirmation are two different tests doing two different jobs, and only the confirmed fusion counts. The direction of travel is toward broad, upfront RNA-inclusive NGS panels that catch ROS1 alongside every other actionable driver in a single run — which will eventually make the IHC-then-confirm dance a legacy workflow. We're not fully there yet. Until we are, respect the confirmatory step, and read a lone positive stain with the skepticism it deserves.

References

1. Lung Cancer: Targeted Therapy in 2025. Curr Oncol. 2025. PMC11941068.
2. FDA Approval Summary authors. FDA Approval Summary: Repotrectinib for Locally Advanced or Metastatic ROS1-Positive Non-Small Cell Lung Cancer. PubMed. 2024. PMID:38875108.