13 Cancers in One Blood Test — but 75% False Alarms

A prospective cohort study led by Yang Shao, PhD, president and CEO of Geneseeq Technology Inc. and professor at Nanjing Medical University, Nanjing, published in Nature Medicine, on a blood test capable of simultaneously detecting 13 types of cancer. The test demonstrated high sensitivity and specificity and was able to identify early-stage cancers that often go unnoticed during routine screenings.

Traditional cancer screening methods are often invasive, expensive, and time-consuming, which can reduce patient adherence. In addition, several cancers — such as pancreatic cancer — are typically asymptomatic in their early stages and progress rapidly, with no established screening protocols currently available.

This prompted the development of less invasive approaches, such as multi-cancer early detection (MCED) blood tests that can detect a cancer signal from circulating cell-free DNA. These simple blood tests analyze plasma cell-free DNA using genetic and fragmentomic-based features from whole genome sequencing to simultaneously detect multiple cancer types. Although promising, current MCED tests still have relatively low sensitivity, typically less than 60%. Experts, including those from the American Cancer Society, cautioned that widespread use could create a false sense of reassurance and potentially deter patients from following up with standard screenings.

Researchers have developed an MCED blood test that detects 13 cancers: breast, cervical, colorectal, endometrial, esophageal, gastric, liver, lung, ovarian, pancreatic, prostate, biliary tract, and lymphoma. These cancers account for 66.6% of all new cases and 74% of cancer-related deaths worldwide. The test uses two main classifiers: the detection-of-cancer classifier, tasked with confirming the presence of cancer, and the tissue-of-origin classifier, responsible for pinpointing the primary site of malignancy by analyzing and integrating feature frameworks, including copy number variations, fragment size coverage, fragment size distribution, nucleosome footprint, and fragment-based methylation.

To develop the test, researchers analyzed 6553 blood samples, 3076 from patients with cancer and 3477 from healthy individuals, divided into a training dataset of 4807 samples and an internal validation dataset of 1746 samples.

Independent validation was performed using a prospectively enrolled cohort of 1465 participants in an age-matched fashion, comprising 732 patients with cancer and 733 non-cancer individuals between April and November 2021. In the third ongoing phase, 3724 asymptomatic adults aged 45-75 years in the Jinling cohort underwent both complete physical examinations and the MCED test in June 2023.

Positive Results

In independent validation, the MCED test showed an overall sensitivity of 87.4% and specificity of 97.8%. The sensitivity was particularly high for certain cancer types, such as 100% for liver and biliary tract cancer, 94.5% for lung cancer, and 82.3% for colorectal cancer. Even cancers that are difficult to diagnose early, such as pancreatic and ovarian cancers, showed a sensitivity of 76.9% for pancreatic cancer and 90.5% for ovarian cancer. Breast cancer had the lowest sensitivity at 63.8%.

The test was effective in detecting early-stage disease, with a sensitivity of 79.3% for stage I and 86.9% for stage II cancer. This increased to 92.4% for stage III and 97.1% for stage IV. When considering the top two tissue origin predictions, the accuracy increased to 90.7% for the internal set and 91.7% for the independent set. It performed best in identifying cancers of the colon-rectum, lungs, and liver but was less accurate for pancreatic and stomach cancers, correctly identifying the origin in 50% or fewer cases.

In a prospective screening cohort of asymptomatic individuals, the MCED test identified 23 of 43 cancer cases within 1 year, with an overall sensitivity of 53.5%. When limited to the 13 cancers that the test was designed to detect, the sensitivity increased to 62.1%. Most of these cases (93%) were early-stage cancers (stage 0, I, or II). The specificity remained high at 98.1%, with a positive predictive value (PPV) of 25% and a negative predictive value of 99.4%.

Notably, 8 of the 23 positive patients who received a positive MCED result had their cancers undetected through physical examination, and 4 had cancers for which there is currently no recommended screening, highlighting the potential of the MCED test to effectively detect cancers that would otherwise have gone undetected.

“Our study demonstrated high sensitivity, highlighting our classifier’s ability to detect cancer cases, even in populations with lower disease prevalence. This underscores the capacity of our classifier to effectively detect incident cancer cases under real-world screening conditions, facilitated by comprehensive physical examinations,” the authors concluded.

These findings suggest that the MCED test could be a valuable complement to existing screening methods, particularly for cancers without routine early detection tools. The ability to detect early-stage pancreatic and ovarian cancers is particularly promising. However, broader validation across diverse populations, cost-effectiveness analyses, and studies on the psychological impact of screening outcomes are critical before widespread clinical implementation.

One key limitation was that the PPV achieved was 25% for the MCED test, which was lower than the 38% reported in the PATHFINDER study published in 2023.

“The PATHFINDER trial enrolled participants with a higher cancer prevalence and utilized the MCED test results to trigger diagnostic workup, systematically investigating participants with positive test results, thus inherently enriching their cohort for cancer diagnoses within the workup pathway. Conversely, the Jinling study adopted a standardized comprehensive physical examination for all participants as the primary screening modality, independent of MCED test outcomes and within the context of lower cancer prevalence,” the researchers noted.

However, the absolute number of false positives in the Jinling study was low (20 of 3724 participants, or 0.54%). False-positive results can lead to unnecessary anxiety, further invasive diagnostic procedures, and potentially inappropriate treatment for patients. The researchers emphasized the need to improve the sensitivity of the MCED test for very early-stage cancers while reducing false positives.

This story was translated from Univadis Italy.