AACR Results Seek Solutions to Precision Medicine Challenges

Use of testing to detect and diagnose cancer, and then guide treatment decisions, is an essential part of precision medicine in oncology. Unfortunately, testing in oncology is below optimal levels, in part because of inconsistent payer coverage. Biomarker testing to find actionable mutations may be limited to fewer genes than an oncologist desires, or results may not be properly interpreted.

Meanwhile, the rise of circulating tumor DNA (ctDNA) testing, or “liquid biopsy,” has raised questions whether this technology is as accurate as a traditional tissue biopsy.

Studies about testing in oncology, including some methods that incorporate AI, were abundant at the 2025 annual meeting of the American Association of Cancer Research (AACR). Besides ctDNA testing, several abstracts looked at multicancer early detection (MCED), which aims to look for multiple types of cancer from a single test.

Proponents of MCED say population-level screening could find certain high-mortality cancers, such as pancreatic or hepatobiliary; these are often found only when cancer has reached an advanced stage. However, the American Cancer Society reports that some tests on the market produce high rates of false-positives, meaning patients are told the MCED test detected cancer, but follow-up testing says otherwise. Right now, MCED tests are not approved by FDA, although some are available as lab-developed tests.

Using Tissue and Liquid Biopsy Together Produced Optimal Results for Patients With Solid Tumors

When it comes to tissue vs liquid biopsies to detect genomic alterations to guide cancer treatment, a study highlighted at AACR’s Tuesday press conference found that using both tests is better than either one alone.

Patients with advanced solid tumors saw significantly improved survival outcomes when they had access to both tests. Results were then evaluated by a molecular tumor board, according to Paolo Marchetti, MD, scientific director at the Istituto Dermopatico dell’Immacolata in Rome, Italy, who presented an exploratory analysis of the ROME Trial.¹

Paolo Marchetti, MD | Image provided by AACR

Testing to identify actionable targets has revolutionized cancer care. “However, the interpretation of the genomic data from both liquid and solid biopsies remain complex, posing significant challenges for clinicians, where often struggle to translate these findings into actionable treatment decision,” Marchetti said at the press conference.

Each test type has its strengths and weaknesses. Tissue biopsies evaluate samples from the tumor itself, but they require surgery and sometimes there is not enough tissue, or the sample may not catch all mutations. Biopsies from blood samples are easier on patients, but they may not find mutations if tumors do not “shed” enough cells, which is more likely in certain types of disease, including brain cancer.

ROME is a phase 2 multicenter study that enrolled 1794 patients with advanced or metastatic solid tumors who were on their second or third line of treatment. Patients provided samples for 2 Foundation Medicine tests: the FoundationOne Liquid CDx and the FoundationOne CDx test to evaluate a tissue biopsy.

Samples underwent next-generation sequencing before a tumor board evaluated data for concordance and discordance, with concordance defined as detection of the same actionable alterations in both biopsy types. Discordance was defined as an actionable alteration in only 1 test. This process identified 400 patients with actionable genomic alterations. Among these patients:

• 197 patients, or 49.2%, had actionable alterations in both the tissue and liquid biopsy.
• Actionable alterations were detected only in the tissue biopsy for 139 patients (34.7%) or only in the liquid biopsy for 64 patients (16%).
• In each arm, patients were randomly assigned to have either tailored therapy or standard of care.

Results. Patients who had alterations in both tissue and liquid biopsy, median overall survival (OS) was 11.05 months for those who received tailored therapy, vs 7.7 months for those in receiving standard of care, for a 26% reduction in the risk of death. Median progression-free survival (PFS) was 4.93 months vs 2.8 months, respectively, for a 45% reduction in risk of progression among those with concordant results.

Among those with discordant results, both OS and PFS were shorter. OS in the group where alterations were found only in tissue was 9.93 months, followed by 4.05 months among those where only the liquid sample showed an alteration. PFS was similar, with the tissue-only group seeing 3.06 months, with 2.07 months for the liquid-only group. In addition:

• The 12-month OS rate was 47.8% in the concordant group receiving tailored therapy, vs 38.8% in the standard-of-care group. The 12-month PFS rates were 27.2% and 9.1%, respectively.
• Among the concordant patients, objective response rate was 20% for patients receiving tailored therapy 11.8% with standard of care.

At the press conference, Marchetti emphasized the role of the tumor board. “The molecular tumor board was not just an advisory group, but was the engine to drive in clinical decision making for every patient,” he said.

Thus, the best results were seen among patients who had both concordant findings and input from a tumor board that could optimize those findings into individual treatment strategies. Future work, he said, should take a deeper dive into other factors that drove the results, such as differences by tumor type, metastatic sites, and the location where tissue was gathered.

Discordant results were attributed to discrepancies found in the detection of molecular alterations (43.3%), high tumor mutational burden (35%), test failures (21%) and microsatellite instability (1%). Pathways with the highest discordance rates were PI3K/PTEN/AKT/mTOR and ERBB2.

“By addressing the challenges of discordance and leveraging the strengths of both biopsy modalities, future strategies can refine precision oncology algorithms and enhance clinical outcomes for patients with advanced cancers,” Marchetti said.

A chief limitation of the study was its exploratory nature of the analysis, which Marchetti said meant it was not powered for subgroup comparisons. Also, the samples for tissue and liquid biopsies were taken at different times. Funders for the study included Roche, Bristol Myers Squibb, Incyte, Novartis, Pfizer, Takeda, Merck, and Eli Lilly and Company.

Framework Offers Performance Assessment in Blood-Based Early Detection

Harbinger Health, which is developing an MCED method that aims to find cancer before it becomes clinically apparent, presented these findings at a poster session Tuesday. Results were derived from CORE-HH, a project undertaken with Sarah Cannon Research Institute. The Cancer ORigin Epigenetics- Harbinger Health (CORE-HH) study (NCT05435066) is enrolling up to 10,000 participants with and without cancer at up to 150 sites in the United States, for the purpose of validating diagnostic accuracy of its early-stage cancer platform.

One Harbinger abstract involved framework to evaluate how well its platform predicts a given cancer type. Not all tests predict all types of cancer, so giving physicians a metric that shows how well the test predicts a certain cancer would offer more insight on whether results should be followed with a tissue biopsy or other test. Results reported the positive predictive value (PPV) for various types of cancer detected with Harbinger’s sequential reflect test. PPV measures the likelihood that the person tested actually has cancer.²

Peripheral blood samples from CORE-HH, including 1039 people with cancer and 1000 without, were used to evaluate the sequential reflex test, in which the primary test examines ctDNA methylation for cancer signals, with an automatic follow-up that uses specific biomarkers to confirm presence of a given cancer type.

Results. Study authors reported a specificity of 98.7% in the sequential reflex test. Patients in the cohort had a mean age of 57.6 years; 61.5% were female, 71.6% White, and 17.4% were Black. The researchers focused on high-mortality cancers, including several that lack screening.

According to the abstract, PPVs were 94% for upper gastrointestinal (UGI), 85% for colorectal, 80% for hepatobiliary (HB), 39% for lung, 38% for pancreaticobiliary (PB, pancreas, gallbladder), and 34% for head and neck. Corresponding odds of correct case-type were 16:1 for UGI, 6:1 for colorectal, 4:1 for HB, 2:3 for lung, 3:5 for PB, and 1:2 for head and neck.

Authors reported that early-stage intrinsic accuracy (sensitivity) was 70% for head and neck, 51% for HB, and 40% for PB, with lower sensitivities for lung and GI cancers. “In a modeled cohort of 100,000 people taking the test, 107 pancreaticobiliary cases are expected, including 38 early stage; the test identifies 67 of the 107, including 15 early stage, with odds in favor of correct classification of 3:5,” the authors wrote.

Hutan Ashrafian, MD, PhD, MBA | Image credit: Harbinger Health

The study authors concluded that this approach could promote optimal use of MCED testing and allow for comparisons of different platforms. “The reported sequential reflex test demonstrates high performance for early cancer detection and localization, warranting prospective validation in diverse populations,” they wrote.

“While innovations in cancer therapy have revolutionized the treatment landscape, the tools to detect and diagnose it early when survival rates are highest have lagged behind,” Hutan Ashrafian, MD, PhD, MBA, Harbinger’s chief medical officer, said in a statement.³

“At Harbinger, our mission is to disrupt conventional cancer detection approaches that either lack an appropriately high diagnostic accuracy for clinically impactful outcomes or only provide detection for a single cancer at a time. By pioneering blood-based tests based on DNA methylation signatures with novel applications including sequential reflex testing and individualized analyses, we aim to provide solutions that give physicians highly accurate and reliable information to get their patients on the right treatment pathways, right away. The data presented [Monday] underscore our commitment to setting a new standard in quantifying test performance, with the ultimate goal of adequately informing earlier clinical intervention and improving outcomes for patients across a wide spectrum of cancers.”³

References

1. Marchetti P, Scagnoli S, Crimini E, et al. Combined tissue and liquid biopsy improve outcomes in advanced solid tumors: an exploratory analysis of the ROME Trial. Presented at: American Association for Cancer Research 2025 annual meeting, April 25-30, 2025; Chicago, IL. Abstract 6372.

2. Massaad E, DiRienzo AG, Withers JB, et al. Novel performance quantification of MCED testing to aid clinical decisions: Analysis of a sequential reflex blood-based methylated ctDNA test. Presented at: American Association for Cancer Research 2025 annual meeting, April 25-30, 2025; Chicago, IL. Abstract LB252.

3. Harbinger Health unveils promising new data on early cancer detection platform at AACR 2025. News release. Harbinger Health. April 25, 2025. Accessed April 29, 2025. https://harbinger-health.com/2025/04/25/harbinger-health-unveils-promising-new-data-on-early-cancer-detection-platform-at-aacr-2025/