Progress in early cancer detection in the clinical laboratory

Currently, cancer accounts for almost 10 million deaths each year, making it the second leading cause of death globally.¹ This burden is expected to rise further, and by 2050, it is estimated that cancer will account for 18.5 million deaths each year, making it a serious and worsening global health issue.²

Early detection of cancer is key to improving cancer care. When identified at the earliest possible stage, cancer is more likely to respond to treatment, resulting in a greater likelihood of survival.³ Although some detection methods are available, there is still an urgent need for better methods to find cancers earlier.⁴

At Roche Experience Days 2024, Professor Xu Songxiao of Zhejiang Cancer Hospital, China, spoke about the potential for cell-free DNA (cfDNA) based approaches and combined biomarker models to meet this need.

Currently, available methods for cancer detection include physical examinations, imaging tests, biopsies, and laboratory tests.⁵ In the laboratory, biomarker testing is a well-established approach to cancer detection, and an emerging area of interest is that of cell-free DNA (cfDNA). cfDNA are fragments of DNA released by cells into the bloodstream to degrade. cfDNA is present in low concentrations in healthy individuals, but in individuals with cancer, these levels are elevated with a proportion of cfDNA coming from tumor cells called circulating tumor DNA (ctDNA).⁶

As concentrations of ctDNA are very low in cfDNA they require highly sensitive technology to be detected in blood samples. However, there are several advantages to cfDNA testing over traditional testing methods. Firstly, the technique is non-invasive when compared to other sampling techniques, as testing only requires taking a blood sample from the patient.⁷ The method also allows for cancers to be detected at a very early stage, increasing the time available for early treatment intervention. In addition, cfDNA testing makes it possible to detect multiple cancers at once, as information on the tissue of origin is available.⁷

Professor Xu Songxiao highlights four cfDNA characteristics that can be analyzed in the lab - mutation, methylation, fragmentomics, and microbiome – and outlines the benefits and limitations of each.

The mutation-based sequencing approach involves identifying and analyzing genetic mutations within the cfDNA. Cancer-associated mutations can be captured using technologies such as quantitative polymerase chain reaction (PCR), digital PCR, or next-generation sequencing before further analysis to confirm the presence of cancer.  

The limitations of this approach are that it can be difficult to detect in samples with low cfDNA yields, and there is a potential for false negative results. As Professor Xu explains, “For example, lung adenocarcinoma has a very hot mutation of EGFR, but only about 50% of lung adenocarcinoma carry this mutation.^8,9 So, if we use this mutation from the cfDNA to do the diagnosis, we're going to miss half of the lung adenocarcinoma.” On the other hand, because DNA mutations also occur in normal tissue samples, this can interfere with testing and create false positive results.

The mechanism of abnormal methylation in cfDNA is another characteristic that occurs in all kinds of cancers and can be analyzed to detect cancer. “One interesting thing is that some methylation happens before gene mutation, so this can help us detect very early cancer,” says Professor Xu. However, detection methods used, such as bisulfite-dependent conversion or bisulfite-independent conversion can lead to DNA damage, making it very difficult to analyze results. Professor Xu notes it is also “high cost and not popular in clinical laboratories.”

cfDNA fragmentomics is a “very new and interesting character” in the field of pan-cancer screening, says Professor Xu. In combination with computational technology such as artificial intelligence (AI), it is possible to detect a wide range of tumors through the evaluation of the fragmentation patterns of cfDNA in the bloodstream. Given that this is a new approach, its limitation comes from the fact that studies are still at the proof-of-concept stage and require further validation with larger cohorts. Professor Xu’s lab has conducted a fragmentation study in gastric cancer and has seen good results so far, including “a very good performance even in stage one gastric cancer”. They hope to publish the results early next year.

Another characteristic of interest is the microbiome, specifically the complicated ‘crosstalk’ between the microbiome and cancer. “The underlying mechanism has been studied recently,” explains Professor Xu, “and the study indicated that the alteration of circulating microorganisms could serve as promising tumor biomarkers.” Again, since circulating microbiome DNA-based methods have only been tested in a limited number of studies, they require further validation and evaluation with large-scale studies.

Beyond cfDNA characteristics, Professor Xu highlights an additional direction in the field of early cancer detection as the combination of biomarkers for analysis through digital algorithms. 

Although biomarkers are used in the detection of cancer, single biomarker testing is often limited in terms of sensitivity or specificity. Professor Xu gives the example of using Alpha-fetoprotein for liver cancer: “30% of hepatocellular carcinoma (HCC) has no AFP change, so we’re going to miss at least 30% of them if we use only AFP as a diagnostic biomarker.”¹⁰ However, by using multifactorial models powered by digital technologies, it is possible to increase diagnostic accuracy by including parameters beyond a single biomarker. Professor Xu cites algorithms such as GAAD or GALAD in HCC, which have demonstrated better performance on HCC early detection in comparison to testing individual biomarkers alone.¹¹

Professor Xu believes digital technologies will increasingly help in cancer care, noting the ability of AI to analyze images, including historical images, to provide additional information to researchers. However, the professor sees the greatest benefit for laboratories from data processing: “We can use AI to analyze our data and build models to find more accurate information for the patient's diagnosis or prognosis evaluation.” Xu believes the combination of new approaches to biomarkers such as cfDNA, aided by AI, is going to change the way we fight against cancer in the future.

For further details on current studies in cfDNA, listen to Professor Xu’s full presentation here.