Transforming Early Cancer Detection with Liquid Biopsy, Automation, and AI

Early detection of cancer improves survival rates and quality of life in patients. Traditionally, cancer screening and detection have been carried out by performing biopsies of solid tumors. An invasive technique with a high human error rate, solid tumor biopsies have been associated with biased diagnoses and poor prognoses. The invasive nature of traditional solid tumor biopsies also compromises patient safety and comfort. And in cases where the tumor is inaccessible, biopsies cannot be performed, making tumor profiling and monitoring almost impossible. 

However, in 2016, the U.S. FDA approved the first liquid biopsy test, cobas EGFR Mutation Test v2, for use with plasma specimens. The test is a companion diagnostic used to detect mutations in circulating-free tumor DNA to determine whether a patient is a candidate for treatment with Tarceva (erlotinib). 

This first approval of a liquid biopsy test marked the beginning of a new era of cancer testing. According to Nitish Kumar, principal analyst at BIS Research, the liquid biopsy market, which accounted for about 55% of the North American market value in 2021, is expected to grow by about 20% from 2022 to 2032. 

Here’s everything you need to know about liquid biopsies.

Liquid biopsy: The advent of noninvasive cancer diagnostics

Biological fluids, including blood, urine, cerebrospinal fluid, and saliva, contain several biomolecules of clinical significance. Screening blood for circulating DNA, RNA, proteins, metabolites, etc., has been shown to unlock deeper, molecular-level insights into cellular happenings and disease states.

Clinical oncology studies analyze DNA released by tumor cells into the bloodstream, called circulating tumor DNA or ctDNA. Applying the precision and accuracy of next-generation sequencing (NGS), researchers can identify genetic changes in tumors and their microenvironment—using samples from a simple blood draw. Hence, the name liquid biopsy. 

In 2016, one of the largest-of-its-kind studies demonstrated the equivalence and efficacy of liquid biopsy with traditional solid tumor biopsies: Liquid biopsy was able to identify the genomic changes and mutations identified by tumor profiling studies. The method even detected mutations linked to treatment resistance that its traditional counterpart failed to detect.

In the study, California-based Guardant Health developed a liquid biopsy test (Guardant360), that used NGS to analyze ctDNA from more than 15,000 patients with more than 50 types of tumors. Guardant360 detected alterations in 70 cancer-related genes at the same distribution and frequency observed in published genomic profiling studies that used solid tumor biopsy. 

Based on accurate and robust results from liquid biopsies, the University of California Davis Comprehensive Cancer Center researchers identified a set of tumor biomarkers that could inform clinicians of cancer progression in their patients and drive personalized treatment development.

Assessing biomarkers using liquid biopsy

One major advantage of liquid biopsy is the ease of sample collection and preparation. The most commonly used body fluid is blood. Peripheral blood samples are screened for a variety of biomarkers that reflect the cellular activity in the body. DNA, RNA, proteins, circulating tumor cells, extracellular vesicles, and a variety of biomolecules have been used as markers of tumor development and progression and treatment responses.

Circulating tumor cells

Liquid biopsies initially investigated circulating tumor cells (CTCs), which are released into the blood by a tumor and travel through blood or lymph “to other areas of the body—having the potential to cause distant metastases.” CTCs carry several biomarkers at varying levels depending on the type and stage of cancer. 

However, CTCs occur at very low concentrations (< 10 CTCs per mL of blood) in circulation, even in late-stage, aggressive, and/or metastatic cancers, requiring highly sensitive technologies to isolate and analyze these cells efficiently. Also, the diversity and variation of CTC markers—owing to the heterogeneity in the tumor microenvironment before and across treatment stages—make it challenging to define the CTC population.

Cell-free nucleic acids

Cell-free DNA (cfDNA) is composed of DNA fragments released into body fluids either by cells undergoing apoptosis or necrosis or by replicating tumor cells (known as ctDNA). cfDNA can be found in blood (plasma and serum), urine, saliva, and cerebrospinal fluid, and is present in both healthy and diseased patients.  

ctDNA amounts to around 1–2 percent of overall cfDNA in patients with cancer. It is characterized by epigenetic changes, like methylation and somatic mutations, and can be of great clinical significance. ctDNA markers have been used to identify cervical, breast, pancreatic, and several other cancers.

The process of transcription naturally amplifies RNA in the cells, making detection easier. Choosing RNA fragments as biomarkers can give direct insights into gene activity. Purified and quantified circulating cell-free mRNA and microRNA (miRNA) have been used to accurately detect and prognosticate ovarian and metastatic colorectal cancers, as well as germ cell tumors.

Mutations in certain hotspots in the noncoding regions of the genes DGRG6, PLEKHS1, WDR74, TBC1D12, and LEPROTL1 have been shown to occur frequently in patients with bladder cancer. A 2023 study targeted these noncoding regions in urine samples and found that the mutations can be used to detect bladder cancer “with 66 percent sensitivity at 92 percent specificity.”

Proteins

Single-cell technologies have accelerated cancer research, but they continue to evolve and currently need standardization. Protein biomarkers have been historically used to detect and monitor cancers. Prostate-specific antigen (PSA), carbohydrate antigen 125 (CA125), and carbohydrate antigen 19-9 (CA19-9) have been used for the diagnosis and prognosis of prostate, ovarian, and pancreatic cancers, respectively. Proteomic analyses in combination with traditional flow cytometry have been used to explore the dynamic CTC populations in head and neck squamous cell carcinoma cases.

However, the complexity of proteomes in body fluids, their heterogeneity among cancers and within tumors, and inconsistent yield from samples outweigh the pros of using protein biomarkers for early cancer detection. 

Extracellular vesicles

Critical to intracellular communication and cargo, extracellular vesicles (EVs) offer clinically significant cellular information. Released into biofluids, EVs—such as exosomes—can be easily isolated and treated to extract DNA, RNA, proteins, lipids, and other potential markers indicative of the pathophysiological status of healthy and tumor cells. 

EV-bound mRNA and exosomal protein and miRNA markers have been shown to aid in the early detection and progression of oral cancer. Exosome-associated miRNAs were used to detect colorectal nasopharyngeal, ovarian, and breast cancers. 

Several other biomolecules, such as tumor-educated platelets and autoantibodies, have been used to detect, monitor, and prognosticate various cancer types and subtypes. Advances in NGS and single-cell sequencing continue to fuel the search for more precise, accessible, and robust predictive biomarkers that inform cancer diagnostics. 

But, challenges remain.

Barriers to translating liquid biopsies

Most robust biomarkers detect cancers after onset and aid tumor progression studies—very few are predictive. And since molecular markers vary with cancer types, subtypes, and stages, biomarkers can be expensive, time-consuming, and laborious to characterize, quantify, and standardize. 

Liquid biopsies are expensive assays and warrant skilled personnel due to the NGS component. Cost, intrinsic variability, lack of resources, and insufficient clinical trial data are a few of the many barriers that have prevented noninvasive cancer screening from being fully adapted to the clinical setting. To intervene early and control oncogenesis, researchers and laboratories need intuitive technology to screen more samples, build datasets, predict and analyze trends, and develop precise and personalized therapies. Automating liquid biopsies may enable clinical labs to handle high throughput without compromising accuracy and patient safety.

Can automation enhance liquid biopsy’s applicability?

Automation can improve the sensitivity and scalability of liquid biopsies and boost the productivity of research and clinical labs. By supporting validation and standardization studies and larger clinical trials, automation can make cancer screening affordable and accessible to patients and profitable to healthcare providers too.

“Standardization of liquid biopsy helps different providers accurately compare results. When the industry achieves that standardization, the discussion will be less about how you gathered the data for liquid biopsy, and more about what the data actually says. Artificial intelligence will help determine how much biomarkers progress over time,” said Manuel Bauer, PhD, associate director of mass spectrometry for Tecan, in a press briefing at the SLAS2024 conference in Boston, MA.

Automation can also be used to develop rapid, point-of-care tests and devices. A recent study, published in AACR’s Cancer Research Communications, described an automated, liquid biopsy assay for breast cancer. The LBx-BCM, or Liquid Biopsy for Breast Cancer Methylation, prototype accurately detects nine methylated markers within 4.5 hours and is comparable to the highly sensitive cMethDNA test.

Applying artificial intelligence to advance noninvasive cancer screening

Analyzing high-throughput clinical data is crucial to precision medicine. Applying machine learning (ML) methods to identify data patterns can help build predictive models to assist in early cancer detection.

Digital pathology employs ML models to analyze digitalized biopsy slides and identify cancer types. ML models trained on large datasets of solid tumor biopsy images can accurately detect cancer-specific genomic biomarkers.

A 2023 study in Scientific Reports used an automated ML model to predict mortality preoperatively in gastric cancer patients due for gastrectomy. The model was trained on existing data to identify stage 1–3 gastric cancer patients undergoing surgery and could predict 90-day mortality well in larger cohorts. Such predictive models can inform patient prognosis and improve patient selection for surgeries and treatments.

Liquid biopsy is a valuable cancer screening and detection tool, and standardization will help it meet current clinical and regulatory requirements.

Leveraging automation and AI can, to an extent, help overcome existing barriers and bring liquid biopsies into clinical practice but will require researchers, manufacturers, healthcare providers, and policymakers to prioritize and invest in patient-centric technology and innovation.