Cofactor Genomics, a privately held molecular diagnostics company based in St. Louis, announced on October 16, 2026, the publication of clinical validation findings for OncoPrism-NSCLC, an RNA-based test designed to identify patients with non-small cell lung cancer who are more likely to benefit from immune checkpoint inhibitor therapy. The peer-reviewed study, published in npj Precision Oncology, evaluated 1,487 patient samples across four datasets and found that patients classified as OncoPrism high had significantly better treatment response and survival outcomes in the immunotherapy-treated cohorts.
The most clinically relevant results came from the PREDAPT cohort, which included 89 evaluable patients treated with anti-PD-1-based therapy. Among those classified as OncoPrism high, the objective response rate was 71%, compared with 36% in both the low- and medium-scoring groups. Median overall survival reached 22.9 months in the high group, versus 13.0 months and 9.2 months in the respective lower-scoring categories.
The study also found that conventional PD-L1 tumor proportion score did not significantly predict response, progression-free survival or overall survival within the evaluable PREDAPT comparison population. OncoPrism retained a significant association with outcomes after statistical adjustment for several clinical factors, suggesting that its multidimensional RNA signature may offer information beyond a single protein-expression measurement. However, this finding does not establish that PD-L1 testing is clinically obsolete or that OncoPrism should automatically determine treatment selection.
The publication represents an important validation milestone for a platform already used in recurrent or metastatic head and neck squamous cell carcinoma. Yet the current lung cancer findings remain primarily retrospective analyses of clinical outcomes. The decisive next question is whether using the RNA test prospectively to guide treatment decisions actually improves outcomes compared with established approaches.
What did the 1,487-patient validation study establish about OncoPrism?
The analysis combined four non-small cell lung cancer cohorts with different treatment histories and research purposes. These included 195 patients treated with an anti-PD-L1 checkpoint inhibitor, 89 patients treated with anti-PD-1-based therapy, 193 patients receiving non-immunotherapy treatment and 1,010 patients drawn from a separate cancer research dataset. Their combined size was 1,487, but not all participants received immunotherapy.
This distinction matters because the headline population is much larger than the number of patients directly contributing to the principal PREDAPT response-rate comparison. The 1,010-patient dataset helped establish reference score distributions and investigate the relationship between the biomarker and outcomes outside immunotherapy. It did not represent more than a thousand prospectively assigned patients receiving treatment according to OncoPrism results.
The study assessed whether an RNA-derived score could distinguish groups with different outcomes following checkpoint inhibitor therapy. Investigators examined objective response, progression-free survival and overall survival, while also evaluating whether the biomarker appeared to predict prognosis independently of treatment. A useful predictive marker should identify differences associated with benefit from a particular therapy rather than simply identifying patients with generally more favorable disease.
The findings supported an association between higher OncoPrism scores and better outcomes in both immunotherapy-treated populations. The researchers also found no corresponding survival association in the non-immunotherapy datasets. That combination strengthens the hypothesis that the signature captures something relevant to immunotherapy sensitivity.
However, the evidence does not demonstrate that patients assigned treatment using OncoPrism perform better than those managed without the test. That would require a prospective clinical strategy study or other appropriately designed investigation.
How were the OncoPrism high, medium and low groups established?
OncoPrism uses RNA expression measurements from tumor tissue to characterize the immune environment surrounding malignant cells. The underlying platform analyzes multiple biological signals and generates a score rather than relying on a single tumor protein or genetic mutation. Cofactor developed its lung cancer adaptation using an earlier biomarker model established for head and neck cancer.
The researchers examined RNA data from 1,010 non-small cell lung cancer samples to establish an NSCLC-specific score distribution. They divided the reference population into three approximate tertiles, creating low, medium and high groups. The resulting OncoPrism-NSCLC thresholds were scores of 0 to 40 for low, 41 to 57 for medium and 58 to 100 for high.
This calibration is important because the immune characteristics and typical treatment responses of different cancers are not identical. A threshold developed for one tumor type may not provide optimal classification in another. Adjusting the score distribution for lung cancer therefore represents a defined analytical development step rather than merely relabeling the original head and neck assay.
The study also examined related RNA signatures across additional cancer types, including clear cell renal cell carcinoma, colon adenocarcinoma, triple-negative breast cancer and bladder carcinoma. However, identifying the biomarker’s presence across these tumors is not equivalent to establishing validated immunotherapy prediction for every disease.
The potential breadth of the underlying technology remains an opportunity for future research. Each indication would require appropriate calibration, analytical validation and clinical outcome evidence.
What do the 71% response rate and survival findings mean for PREDAPT patients?
The PREDAPT analysis initially involved 176 enrolled patients from 17 healthcare systems. After eligibility and quality-control assessment, 89 samples from 12 sites were included in the validation dataset. This attrition is relevant because the final analysis represents a selected subset rather than every patient initially enrolled.
The average patient age was 67 years, and the cohort included different non-small cell lung cancer histologies and treatment regimens. Approximately 71% received checkpoint inhibition combined with chemotherapy, while the remaining patients received immunotherapy alone. These differences create potential influences on response outcomes that researchers must address when evaluating the biomarker.
In the high-scoring group, 71% of patients achieved objective responses, compared with 36% in each of the low and medium groups. The overall response rate across the study population was 47%. The association between OncoPrism score and objective response was statistically significant, with a reported p-value of approximately 0.008.
Median progression-free survival was 18.6 months in the high group, compared with 7.0 months in the low group and 5.5 months in the medium group. Median overall survival was 22.9 months in the high group, compared with 13.0 and 9.2 months, respectively. These comparisons demonstrate substantial separation between the observed outcomes of different score categories.
However, patients were not randomized to receive different treatments according to their scores. The data therefore show that OncoPrism classification was associated with outcomes among treated patients, not that the test itself caused longer survival. The distinction is central to determining the clinical role of any predictive diagnostic.
How did the independent OAK cohort support the lung cancer findings?
The researchers also analyzed 195 patients receiving atezolizumab, an anti-PD-L1 antibody, in the OAK clinical trial. These participants had advanced non-small cell lung cancer and had previously received one or two lines of chemotherapy. This cohort provided an independent opportunity to evaluate the relationship between OncoPrism score and outcomes after checkpoint inhibition.
Patients classified as OncoPrism high had median progression-free survival of 4.7 months. The corresponding figures were 2.8 months in the low group and 1.5 months in the medium group. The difference was statistically significant, supporting an association between a high RNA biomarker score and more favorable disease control.
Overall survival showed a similar pattern. Median survival was 20.4 months in high-scoring patients, compared with 12.7 months and 10.5 months in the low and medium groups. The study reported a significant relationship between OncoPrism classification and overall survival.
These results are relevant because the OAK treatment setting differed from PREDAPT. The patients had received previous chemotherapy and were treated with an anti-PD-L1 antibody, whereas PREDAPT included anti-PD-1-based regimens, often combined with chemotherapy. A consistent directional association across the two cohorts provides additional support for the biomarker.
Nevertheless, the differences between study populations also limit direct comparison of median survival estimates. Outcomes depend on treatment line, patient fitness, cancer characteristics and subsequent care. The OAK and PREDAPT medians should not be treated as interchangeable measures of test performance.
Why did PD-L1 testing fail to predict outcomes in the PREDAPT comparison?
PD-L1 tumor proportion score measures the percentage of viable tumor cells expressing the PD-L1 protein through immunohistochemistry. It is widely used to help guide immunotherapy decisions in non-small cell lung cancer, although its predictive performance is imperfect. Some patients with high PD-L1 expression do not respond, while others with lower expression may benefit from checkpoint inhibitors.
In the PREDAPT cohort, PD-L1 measurements were available for 83 of the 89 evaluable patients. Within that population, the researchers did not find a statistically significant association between conventional PD-L1 score groups and objective response, progression-free survival or overall survival. OncoPrism classifications remained significantly associated with the evaluated outcomes.
The comparison also identified some patients with high OncoPrism scores despite having lower PD-L1 tumor proportion scores. This suggests that RNA analysis may capture aspects of the immune environment not reflected by PD-L1 expression alone. The investigators consequently proposed that the tests could provide complementary information.
However, a lack of statistical significance in 83 patients does not establish that PD-L1 lacks clinical value. Larger randomized studies have supported the use of PD-L1 in defined treatment decisions, and the PREDAPT population included a mixture of monotherapy and combination treatment. The influence of chemotherapy can complicate comparisons between biomarkers.
The researchers themselves identified treatment selection and disease histology as potential confounding factors. OncoPrism may offer useful additional information, but the current study does not support removing established testing from clinical practice.
What distinguishes a predictive biomarker from a general survival marker?
A prognostic biomarker provides information about the likely course of a disease regardless of a particular treatment. A predictive biomarker is intended to identify differences in the benefit expected from a specific therapy. The distinction is especially important in immunotherapy because an assay used to guide treatment should do more than identify patients whose cancer is generally less aggressive.
To investigate this issue, the researchers evaluated OncoPrism in patients who did not receive checkpoint inhibitor therapy. The non-immunotherapy cohorts included 193 patients receiving docetaxel in OAK and 1,010 patients from The Cancer Genome Atlas data collection. These analyses provided a way to assess whether the score was associated with survival in different treatment settings.
OncoPrism was not significantly associated with overall survival in either non-immunotherapy dataset. It also did not significantly predict progression-free survival in the docetaxel-treated cohort. The absence of a similar association outside immunotherapy supports the interpretation that the test may be connected specifically to checkpoint inhibitor benefit.
However, these findings do not constitute definitive proof of a treatment-biomarker interaction across randomized groups selected by test status. Differences between the datasets, treatment approaches and patient characteristics remain important. Further prospective confirmation would strengthen the argument that the score can guide therapy selection.
The distinction has substantial commercial implications. A test that identifies likely immunotherapy responders may help physicians decide how aggressively to treat, while a test that merely estimates overall prognosis may have a more limited role in selecting medicines.
How does OncoPrism analyze RNA from routine tumor biopsy samples?
The test is based on RNA expression analysis rather than exclusively identifying DNA mutations or measuring one protein. RNA levels can provide information about biological activity within tumor tissue, including signaling associated with immune cells. Cofactor combines these measurements through a multidimensional classifier intended to characterize the tumor immune environment.
The method uses formalin-fixed, paraffin-embedded tissue, a common format for preserving samples collected during routine cancer diagnosis. This could support implementation without requiring a completely separate biopsy procedure for every patient. Nevertheless, sufficient tumor content and specimen quality remain necessary for reliable analysis.
The validation study required an adequate tissue section and excluded samples with insufficient tumor cellularity or other quality problems. Of the 176 patients assessed for inclusion in the PREDAPT analysis, 46 were excluded for eligibility reasons and another 41 failed quality control. This illustrates the practical importance of specimen availability and processing.
Analytical reproducibility was also examined using repeated measurements of different biological samples. The researchers reported approximately 95% categorical agreement across valid technical comparisons. This provides supportive information about consistency, although testing across broader laboratory workflows and real-world sample conditions remains important.
OncoPrism is offered as a laboratory-developed test, with its performance characteristics established within Cofactor’s laboratory framework. It has not received FDA clearance or approval as a companion diagnostic. Clinical validation and regulatory authorization remain distinct concepts.
How does OncoPrism-NSCLC build on the existing head and neck cancer test?
Cofactor previously developed OncoPrism-HNSCC for recurrent or metastatic head and neck squamous cell carcinoma. That test uses a related RNA biomarker approach to help identify patients more likely to benefit from checkpoint inhibitor treatment. Its validation programme helped establish the scientific and analytical foundation for extending the technology into other tumor types.
The head and neck application has obtained Medicare coverage under the relevant molecular diagnostics framework. That commercial status is significant because reimbursement can influence whether a test becomes practical in routine oncology care. However, the head and neck coverage should not automatically be described as Medicare coverage for every new OncoPrism indication.
The October announcement concerns a lung cancer-specific adaptation supported by separate clinical validation. The underlying biomarker is similar, but the score thresholds were adjusted for the NSCLC population. This distinction matters because test performance cannot simply be transferred between cancers without appropriate research.
Cofactor is seeking to position OncoPrism as a broader precision oncology platform rather than a single-indication test. The company has also received investment from Labcorp and Ascension Ventures, which may support engagement with healthcare networks and commercial expansion. However, the October study does not disclose a new financing round, an FDA approval or a quantified revenue contribution from lung cancer testing.
The potential market for immunotherapy selection in NSCLC is substantial, but commercial adoption will depend on reimbursement, laboratory access and confidence that test results improve actual treatment decisions.
What limitations must be addressed before OncoPrism can guide lung cancer treatment routinely?
The most immediate limitation is that the PREDAPT findings were obtained through retrospective analysis of patient samples and outcomes. The final clinical validation cohort contained 89 patients, including individuals treated with checkpoint inhibitor monotherapy and others receiving immunotherapy plus chemotherapy. This creates uncertainty about the extent to which the biomarker specifically identifies benefit from the immunotherapy component.
The study also had a median follow-up of 9.3 months within the PREDAPT cohort, although survival estimates were generated using available longitudinal data. Longer follow-up and a larger cohort would provide more precise estimates, particularly for patients with different PD-L1 levels and treatment regimens. The investigators acknowledged these issues in their discussion.
Another important question concerns clinical actionability. A low score does not mean that a patient cannot benefit from immunotherapy, and some responders appeared in the lower-scoring groups. Withholding effective treatment solely on the basis of an unproven selection strategy could create harm if the assay incorrectly identifies a patient as unlikely to respond.
The most useful future study would prospectively evaluate whether incorporating OncoPrism into treatment decisions improves outcomes or avoids unnecessary treatment compared with established care. Such research could also examine cost-effectiveness, patient preferences and the effect of test turnaround times on clinical practice.
Cofactor’s validation results represent an important advance in developing RNA-based immunotherapy diagnostics. The findings show that the classifier distinguished patient groups with markedly different response and survival outcomes and retained predictive associations that were not observed with PD-L1 alone in the evaluated cohort.
The next challenge is moving from retrospective prediction to demonstrated clinical utility. If additional studies confirm that OncoPrism helps physicians select more effective treatment strategies, the platform could become a useful complement to established genomic and immunohistochemical testing. Until then, it remains a clinically validated laboratory-developed assay whose ultimate effect on lung cancer treatment decisions has yet to be established.
