Innovative blood test boosts accuracy of lung cancer screening


In a recent study published in Cancer Discovery, researchers developed and validated a blood-based, cell-free deoxyribonucleic acid (cfDNA) fragmentome assay for lung cancer detection, which, if the results were positive, would be followed by low-dose computed tomography (LDCT).

Study: Clinical validation of a cell-free DNA fragmentome assay for augmentation of lung cancer early detection. Image Credit: MMD Creative/Shutterstock.comStudy: Clinical validation of a cell-free DNA fragmentome assay for augmentation of lung cancer early detection. Image Credit: MMD Creative/Shutterstock.com

Introduction

Lung cancer is a major death cause, and yearly screening is crucial. However, chest LDCT has low adoption due to patient barriers like inadequate awareness, radiation concerns, and limited availability.

Other challenges include poor smoking history recording, a lack of defined practices, and specialist follow-up.

A blood-based lung malignancy screening test, like the fragmentome technique, could increase screening rates by analyzing specific chromatin configurations in peripheral blood.

About the study

In the present DELFI-L101 study, researchers developed a hematological test using machine learning to analyze DNA fragmentomes and identify individuals at risk of lung cancer. Individuals testing positive would undergo LDCT.

Beginning March 2021, the researchers enrolled 958 individuals aged 50–80 with ≥20 pack-years of smoking across 47 United States (US) facilities. Eligibility features resembled the LDCT screening criteria of the 2015 National Health Interview Survey (NHIS).

They excluded individuals with cancer therapy within one year, a history of hematologic malignancy or myelodysplasia, organ tissue transplantation, blood product transfusion within 120 days of enrollment, pregnancy, and participation in other trials.

The team divided the study participants into three groups: A (lung cancer), B (non-cancer controls), and C (cancer other than lung cancer).

The American Joint Committee on Cancer’s Cancer Staging Manual (AJCC) criteria ascertained the disease stage. Changes in cfDNA fragmentation patterns (fragmentomes) in blood revealed genomic and chromatin features of lung cancer.

The researchers trained the classifier on 576 cases and controls before validating it on another 382 cases and controls.

They used whole genome sequences from the training dataset to assess fragmentations in 504 non-overlapping-type 5.0 MB sections with strong mappability. Each region included 80,000 pieces and covered a genome size of 2.50 GB.

The team examined genome-wide alterations to Hi-C open-type (A compartment) and closed-type (B compartment) chromatin.

They created the classifier using principal component analysis (PCA) and logistic regressions, incorporating chromosomal arm-level changes, cfDNA fractions derived from the mitochondrial genome, and cfDNA fragment length distributions.

The researchers performed Monte Carlo simulations on 15 million individuals under three scenarios:

  • Base Scenario: Current practices without hematological screening.
  • Low Scenario: 10% uptake of hematological screening for individuals eligible for pulmonary cancer screening but not subjected to low-dose CT in the first year, increasing to 25% in five years.
  • High Scenario: 20% uptake of hematological screening for the same group in the first year, increasing to 50% in five years.

Results

The researchers observed 58% test specificity, 84% sensitivity, and 99.8% negative predictive value (NPV). Applying the rest to the screening-eligible group with 0.7% lung cancer prevalence, the number needed to screen (NNS) was 143.

Study validations showed negative and positive results related to NNS with LDCT imaging to detect 414 and 76 cases, respectively, yielding a 5.5 relative risk value. The positive predictive value (PPV) was almost double that of the LDCT qualifying requirements alone.

The cfDNA fragmentomes of lung squamous cell carcinoma (LUSC) patients comprised a component resembling cfDNA profiles from non-cancer individuals and another resembling A/B-type compartments noted in LUSC tissues.

Non-cancer individuals showed cfDNA patterns approximating lymphoblastoid Hi-C findings. Within common locations, fragmentations among samples provided by individuals with cancer presence and absence were similar.

Lung cancer patients had increased cell-free DNA representations fpr 1q, 3q, 5p, 8q, and 12p, as well as lower 1p, 3p, 4q, 5q, 10q, and 17p levels. Their cfDNA fragmentations differed from controls, revealing more closely packed chromatin in cfDNA of closed LUSC spaces, while lymphoblastoid reference regions showed the reverse impact.

At the cut-off of 0.2, ten-fold cross-validation with ten repeats within the training population yielded 50% overall specificity and sensitivities of 75%, 90%, 96%, and 97% for stages I, II, III, and IV, respectively. Sensitivity was constant across ages, with younger people having higher specificity. Using the 2015 NHIS data yielded 80% sensitivity and 58% specificity.

From the ‘base’ scenario (24,489 cases), lung cancer cases identified by screening increased to 63,523 (the ‘low’ scenario) and 100,346 (the ‘high’ scenario). In contrast, stage I cases increased by 4.80% and 9.70%, while stage IV diagnoses decreased by 4.20% and 8.70%, respectively.

In total, 4,720 deaths from lung malignancies could be averted in the ‘base’ scenario, 7,652 in the ‘low’ scenario, and 14,264 deaths in the ‘high’ scenario. LDCT use in screening could reduce the number of tests required to identify lung cancers from 202 (‘base’ scenario) to 150 (‘low’ scenario) and 139 (‘high’ scenario).

Conclusion

Based on the study findings, the DNA fragmentome assay provides a novel, accurate, affordable, blood-based tool for initial lung cancer evaluation with LDCT follow-ups.

The assay could contribute to preventing lung cancer-related deaths, with moderate adoption rates possibly lowering late-stage diagnoses and fatalities.



Source link

Spread the love

Leave a Reply

Your email address will not be published. Required fields are marked *