How NGS Testing Directs Lung Cancer Treatment

Next-generation sequencing has become the backbone of precision treatment for lung cancer, capable of scanning hundreds of cancer-related genes from a single tissue or blood sample to identify mutations that can be matched to targeted drugs. Clinical guidelines now recommend broad-panel NGS over older single-gene tests for anyone diagnosed with advanced non-small cell lung cancer, because the number of mutations that qualify a patient for a specific therapy has grown beyond what piecemeal testing can reasonably cover. Yet the technology carries its own set of complications, from sample-quality failures and turnaround delays to persistent gaps in who actually gets tested.

Why Broad-Panel Testing Replaced Single-Gene Tests

A decade ago, molecular testing in lung cancer often meant checking for one or two mutations at a time, usually EGFR and ALK. As researchers identified more and more “actionable” alterations, that sequential approach became impractical. Each additional single-gene test consumed more tissue, cost more money, and added days or weeks to the diagnostic timeline. Broad-panel NGS solved most of these problems by interrogating dozens to hundreds of genes simultaneously from one sample.

Guidelines from organizations like the National Comprehensive Cancer Network now recommend comprehensive NGS panels to search for all known actionable mutations rather than relying on one gene at a time.¹ The list of targets continues to grow and currently includes alterations in EGFR, ALK, ROS1, BRAF, KRAS, MET, RET, NTRK, ERBB2, and others. Missing even one of these can mean a patient receives chemotherapy or immunotherapy when a targeted drug with fewer side effects and better response rates was available.

That said, not every NGS panel is equal. A review of commercially available panels found that only about 6% captured all twelve biomarkers recommended by current lung cancer guidelines.² Gene fusions and rearrangements appeared in fewer than half of panels examined, and NTRK fusions, which are rare but highly targetable, were missing from the vast majority. MET amplification was included on only about one in eight panels. This means that even when a patient does get NGS, the panel chosen can leave blind spots.

What RNA-Based Sequencing Adds

Standard NGS panels typically analyze DNA, and they are excellent at finding point mutations and small insertions or deletions. They are less reliable at detecting gene fusions and certain structural rearrangements, events where two genes break apart and rejoin in a way that creates an abnormal protein driving the cancer. RNA-based NGS is better suited for these because it reads the messenger molecules the cell actually produces, making expressed fusions easier to spot.

A large study found that adding RNA-based sequencing identified roughly 15% more patients carrying actionable structural variants compared with DNA-based testing alone, including about 14% more fusions and nearly 19% more cases of MET exon 14 skipping.³ For ALK fusions specifically, real-world data from over 7,000 patients showed that about 15% of ALK-positive cases were picked up only by RNA sequencing and would have been missed by DNA alone.⁴ A separate analysis across more than 5,500 patients confirmed that RNA-based NGS detected a higher rate of fusions overall, with especially large gains for rarer targets like NTRK and NRG1 fusions.⁵

The practical takeaway is that DNA-only testing, while good, can miss a meaningful fraction of patients who would benefit from a targeted therapy. Laboratories increasingly run both DNA and RNA panels in parallel, though not all centers have adopted this practice yet.

Liquid Biopsy as an Alternative to Tissue

Tissue biopsies remain the gold standard for NGS, but they are not always feasible. Some tumors sit in locations that are difficult or dangerous to biopsy, and some patients are too sick to undergo the procedure. Liquid biopsy, which sequences fragments of tumor DNA circulating in the blood, offers an alternative. It is faster, less invasive, and can be repeated over time to track how the cancer changes.

A systematic review and meta-analysis of blood-based NGS panels in advanced non-small cell lung cancer found pooled specificity near 100% and sensitivity around 77%, with overall diagnostic accuracy reaching 0.99 on a standard scoring metric.⁶ In plain terms, when a liquid biopsy says a mutation is present, it is almost certainly correct. But it misses some mutations, particularly when the tumor is shedding little DNA into the blood.

That sensitivity gap matters most in earlier-stage disease. A real-world Chinese cohort study found that plasma-based NGS in stage III patients had a positive agreement rate of only about 29% with tissue results, while stage IV patients reached above 99%.⁷ A Dutch study comparing liquid biopsy to standard tissue-based testing found overall concordance of about 71% at the patient level, with two patients having actionable mutations missed entirely by blood testing, changes that would have directly affected their treatment.⁸ The consensus view is that a negative liquid biopsy in a patient who could have tissue tested should be followed up with a tissue biopsy whenever possible.

Detecting Cancer’s Return Before Scans Can

One of the most promising newer applications of blood-based NGS is molecular residual disease monitoring. After surgery or radiation aimed at curing early-stage lung cancer, tiny amounts of circulating tumor DNA can signal that microscopic cancer cells remain before any visible recurrence shows up on imaging. This lead time could, in theory, allow earlier intervention.

An early landmark study tracked patients after surgery and found that those with detectable circulating tumor DNA at a post-operative landmark timepoint had dramatically worse outcomes. At three years, none of the patients with detectable tumor DNA were free of recurrence, compared with 93% of those whose blood tested clean.⁹ A separate retrospective validation study confirmed this pattern across a larger combined cohort, finding that post-treatment tumor DNA detection was associated with roughly an 11-fold higher risk of recurrence and about an 8-fold higher risk of death.¹⁰

Another study using ultra-deep plasma sequencing found that tumor DNA positivity preceded visible recurrence on scans by a median of about three months, and in some cases by nearly a year.¹¹ The clinical question that remains is whether acting on that early signal, say by starting chemotherapy or targeted therapy months before a scan would have caught the recurrence, actually improves survival. Trials are underway, but for now this application is rapidly moving from research into clinical practice.

When Targeted Therapy Stops Working

Nearly every patient who responds to a targeted therapy eventually develops resistance. The tumor acquires new mutations or activates alternative pathways that allow it to grow despite the drug. NGS plays a central role in figuring out what changed and what to try next.

For EGFR-mutant cancers treated with first- or second-generation drugs, the most common resistance mechanism is a secondary mutation called T790M, found in roughly 63% of tumors that progressed on treatment in one NGS-based study.¹² That study also found that tumors without T790M had higher overall mutation burden and worse survival, suggesting those patients might benefit from immunotherapy instead. The same research identified a previously unknown resistance mutation, illustrating that NGS can uncover novel mechanisms that older, narrower tests would miss.

In ALK- and ROS1-positive cancers, resistance is similarly complex. Researchers have used combinations of NGS, direct sequencing, and other techniques to identify kinase domain mutations, gene copy number changes, and activation of entirely separate signaling pathways.¹³ The practical implication for patients is that re-biopsy and repeat NGS at the time of progression often reveal a new target, sometimes allowing a switch to a next-generation drug rather than falling back on chemotherapy.

Tumor Mutation Burden and Immunotherapy

Beyond matching patients to targeted drugs, NGS panels can estimate a tumor’s overall mutation burden, a count of mutations per stretch of DNA. The idea is that tumors with more mutations produce more abnormal proteins, giving the immune system more targets to recognize, and therefore respond better to checkpoint inhibitor immunotherapy.

This concept has gained traction, and the FDA has approved high tumor mutation burden as an indication for pembrolizumab across tumor types. But the reality is messier than the approval might suggest. No single threshold for “high” mutation burden has consistently predicted improved survival across randomized trials.¹⁴ Different NGS panels, because they cover different sets of genes and use different bioinformatic pipelines, can give different burden estimates for the same tumor. A harmonization study confirmed that available assays differ in gene content, panel size, and algorithms, leading to discrepancies that can directly affect which patients are classified as “high” versus “low.”¹⁵

For patients and oncologists, this means tumor mutation burden from NGS is a useful piece of the puzzle, but it should not be the sole factor driving immunotherapy decisions. PD-L1 expression, clinical features, and the patient’s overall health all play a role, and a “low” mutation burden on one panel might be scored differently on another.

Practical Hurdles in Getting Results

Even when NGS is ordered, the process from biopsy to actionable results can be frustratingly slow. Time matters, because patients with advanced lung cancer may deteriorate while waiting for molecular results. A quality-improvement initiative at one center showed that the median time from pathologic diagnosis to NGS results was 24 days before targeted interventions, and even after streamlining the process, it took 16 days. The biggest bottleneck was the gap between the biopsy result becoming available and someone actually ordering the NGS test, which was cut from seven days to one through workflow changes. Total time from diagnosis to starting the right treatment dropped from about 33 days to 22.¹⁶

Sample quality adds another layer of difficulty. Most lung cancer biopsies are small, and the tissue is preserved in formalin-fixed paraffin blocks that can degrade DNA. A study examining NGS failure rates found that samples with very low tumor content had a success rate of only about 44%, while physically small tissue fragments had a success rate around 75%.¹⁷ Interestingly, once tumor content exceeded about 10%, additional increases did not substantially improve sequencing outcomes. Failed sequencing means the patient either needs a repeat biopsy, a switch to liquid biopsy, or simply starts treatment without molecular guidance.

Is NGS Cost-Effective?

Broad-panel NGS is more expensive upfront than testing for one or two genes, which raises reasonable questions about value. Multiple health-economic analyses have weighed in, and the answers depend on the healthcare system.

A Spanish analysis found that NGS provided additional quality-adjusted life-years compared to sequential single-gene testing, at a cost that fell below standard willingness-to-pay thresholds, making it cost-effective in that setting.¹⁸ A U.S. study went further, concluding that large-panel NGS was actually a dominant strategy, meaning it was both more effective and less costly over five years than alternatives, because it more reliably identified patients for targeted therapies and avoided futile treatments.¹⁹ A Brazilian analysis, however, reached a different conclusion: while NGS detected about 24% more true cases, the additional cost per quality-adjusted life-year gained did not meet that country’s cost-effectiveness threshold.²⁰

The pattern suggests that NGS pays for itself when a healthcare system can act on the results, meaning affordable targeted drugs are available and reimbursed. In settings where access to those drugs is limited, the testing cost may not translate into better outcomes for patients.

Who Gets Tested and Who Gets Left Out

Access to NGS is not evenly distributed. A systematic review of 18 studies found that the majority reported racial and ethnic disparities in genomic testing among lung cancer patients. In seven of those studies, Black patients were less likely to receive testing compared with White or Asian patients.²¹

A study of more than 12,000 patients in the U.S. community oncology setting quantified the gap. Compared with White patients, Black patients had roughly a 7.5 percentage-point lower rate of NGS testing, and Latinx patients about an 8.3-point gap. The disparities were not simply a matter of certain practices serving certain populations. About half of the inequity for Black patients arose within the same practice, meaning that even at the same clinic, Black patients were less likely to be tested. The physician-level analysis showed a similar pattern, with the majority of the inequity occurring within the caseloads of the same doctors.²² These are gaps that cannot be explained away by geography or insurance alone, and they represent missed opportunities for patients who may have had targetable mutations.

Rare Histologies and Unexpected Finds

NGS is most discussed in the context of common lung adenocarcinoma, but it can be equally revealing, sometimes more so, in rarer subtypes. Invasive mucinous adenocarcinoma is one example. Comprehensive NGS and fusion testing of 200 of these tumors identified a driver alteration in 96% of cases. The majority were KRAS mutations, but among the KRAS-negative tumors, about half carried targetable fusions, including NRG1 and a previously unknown NRG2 fusion. Some of those patients went on to achieve durable responses to targeted therapy.²³ Without broad NGS, those rare fusions would likely have gone undetected.

Small cell lung cancer, long considered a genomically monotonous disease dominated by TP53 and RB1 loss, is also yielding more nuance under NGS. Molecular subtyping studies have identified recurrent alterations in genes like NOTCH1, PIK3CA, and CREBBP alongside the canonical TP53/RB1 pattern.²⁴ Transcription-factor-based subtyping has revealed that certain molecular subtypes carry enrichment for specific gene amplifications and deletions, including MYC amplification and PTEN loss, which could eventually guide treatment selection.²⁵ There are no approved targeted therapies for small cell lung cancer based on NGS results yet, but the field is moving in that direction.

Tumor Heterogeneity and Why a Single Biopsy May Not Tell the Full Story

Cancer is not a single uniform mass. Different regions of the same tumor can carry different mutations, a phenomenon that NGS has helped reveal in striking detail. The TRACERx study, which performed multi-region sequencing on hundreds of lung cancers, found that in adenocarcinoma, mutations in more than half of common cancer genes showed evidence of being present in only some parts of the tumor rather than throughout. In about one in five tumors, large-scale genome doubling events had occurred in some regions but not others.²⁶

An earlier phase of the same study showed that while certain well-known drivers like EGFR and TP53 mutations were almost always present throughout the entire tumor, other potentially important mutations in genes like PIK3CA and NF1 appeared in some regions but not others, present in more than 75% of the tumors analyzed.²⁷ Elevated copy-number variation between regions was associated with roughly a fivefold increase in the risk of recurrence or death. Multi-region whole genome sequencing of squamous cell carcinomas and neuroendocrine tumors found similar patterns of massive genomic instability and intra-tumor diversity.²⁸

This has real implications for clinical NGS. A single needle biopsy samples one small region, and the mutations found there may not represent what is happening elsewhere in the tumor or in its metastases. It is one reason why repeat biopsies at progression, or liquid biopsy to capture DNA shed from multiple tumor sites simultaneously, have become increasingly important in treatment planning.

AI and Variant Interpretation

An NGS panel can return hundreds of detected variants per patient, the vast majority of which are biologically irrelevant. Deciding which variants to report and which to filter out requires expert human review, a bottleneck as testing volumes grow. Artificial intelligence tools are being developed to assist with this triage step.

One study training tree-based machine-learning models on clinical variant-reporting data achieved precision-recall scores above 0.90, suggesting that automated systems can reliably predict whether a given variant should be flagged for clinical attention.²⁹ However, a separate evaluation of AI-based tools specifically designed to classify clinically actionable variants in lung cancer found that performance varied widely, with some tools showing high sensitivity but very low specificity, meaning they correctly caught important variants but also flagged many benign ones.³⁰ For now, AI serves as a helpful first pass rather than a replacement for expert molecular pathologists.

What Patients Understand About Their Testing

There is a persistent gap between what NGS delivers and what patients expect it to deliver. Surveys of cancer patients undergoing genomic sequencing have found that the majority, between two-thirds and three-quarters, assumed that incidental findings unrelated to their cancer, like genetic risk factors for diabetes, would automatically be disclosed to them, even after being told otherwise by study staff.³¹ A broader review confirmed that genomic literacy among cancer patients is generally low and that patients vary widely in how much information they want and can absorb.³²

This disconnect matters practically. A patient who believes their tumor’s NGS panel screened them for hereditary cancer syndromes may skip recommended germline genetic testing. Conversely, a patient who receives an NGS report full of unfamiliar gene names and variant classifications may feel overwhelmed and lose confidence in their treatment plan. Oncology teams increasingly rely on genetic counselors and nurse navigators to bridge this gap, but access to those professionals is uneven, particularly at smaller or rural practices.

Beyond Mutations and Into Methylation

Standard NGS panels focus on changes in the DNA sequence itself, but cancer also involves chemical modifications to DNA that do not alter the genetic code. DNA methylation, where small chemical groups are added to certain regions and silence gene activity, is one of the earliest events in cancer development. Abnormal methylation patterns can now be detected using techniques that combine bisulfite conversion with next-generation sequencing, even in non-invasive samples like blood and sputum.³³ These approaches are being explored for early lung cancer detection in high-risk populations, potentially supplementing low-dose CT screening. While methylation-based testing is not yet a routine part of lung cancer workups, it represents a natural extension of what NGS technology can reveal beyond the mutation-centric view that currently dominates clinical practice.