Potential utility of longitudinal somatic mutation and methylation profiling for predicting molecular residual disease in postoperative non‐small cell lung cancer patients

Hang Li, Zelin Ma, Bin Li, Yunjian Pan, Jiaqing Xiang, Yawei Zhang, Yihua Sun, Ting Hou, Analyn Lizaso, Yan Chen, Xi Li, Hong HuCancer 2021 · DOI

TL;DR

This study explores the combined power of tracking both DNA mutations and methylation patterns in blood over time to detect minimal residual disease (MRD) in non-small cell lung cancer (NSCLC) patients after surgery. The findings suggest this dual approach can predict recurrence earlier than current methods, offering a more sensitive tool for personalized post-treatment monitoring.

What Problem Does This Paper Address?

For patients with early-stage non-small cell lung cancer (NSCLC), curative surgery is the standard treatment. However, a significant portion of these patients experience disease recurrence within years, even after successful surgery. Current staging systems, while useful, don't always accurately predict individual patient outcomes. Detecting "minimal residual disease" (MRD) – tiny amounts of cancer cells left behind after treatment – is crucial for identifying high-risk patients who might benefit from further therapy and for sparing low-risk patients unnecessary treatments. Traditional methods for detecting MRD, like imaging, are often too late, only picking up disease when it has grown into a detectable tumor. Circulating tumor DNA (ctDNA) in blood offers a less invasive way to monitor for MRD, but detecting low-frequency mutations in early-stage cancers can be challenging due to the limited amount of ctDNA. This paper addresses the need for more sensitive and earlier detection of MRD in postoperative NSCLC patients by investigating the combined utility of two types of ctDNA analysis: somatic mutations and DNA methylation.

Methods

The researchers enrolled 65 patients with resectable NSCLC and collected tumor tissue and longitudinal blood samples at various time points before and after surgery. They employed two key molecular profiling techniques:

  • Ultra-deep targeted sequencing: This was used to analyze somatic mutations in both tumor tissue and plasma-derived cell-free DNA (cfDNA). The "ultra-deep" aspect refers to high sequencing depth, allowing for the detection of very low-frequency mutations, potentially down to 0.05% with the use of unique molecular identifiers (UMIs).
  • Targeted bisulfite sequencing: This method was used to profile DNA methylation patterns. Specifically, they developed a "tumor-informed MRD prediction model." This model uses methylation patterns from the patient's own tumor tissue as a reference to assess the methylation signal intensity in their plasma cfDNA. This "tumor-informed" approach aims to minimize inter-patient variability and focus on cancer-specific epigenetic changes.

The core of their approach involved creating an "MRD score" based on these methylation profiles, which reflects the presence of cancer-specific methylation signatures in the blood. They then longitudinally tracked both mutation levels (specifically, maximum allelic fraction or maxAF) and this methylation-based MRD score in the plasma of patients over time.

Datasets

The study utilized data from 65 patients with resectable NSCLC. This included:

  • Tumor tissues: Collected from all 65 patients.
  • Longitudinal blood samples: Collected from patients at baseline (before surgery) and at multiple postoperative follow-up time points.

The study did not introduce a new benchmark dataset but rather analyzed data from a prospectively enrolled cohort.

Key Insights and Ideas

  • Methylation as a Sensitive Early Indicator: The study highlights that DNA methylation profiling, particularly when using a tumor-informed approach, appears to be a more sensitive indicator of MRD than mutation detection alone, especially in early-stage NSCLC. This is likely due to methylation changes occurring earlier in tumorigenesis and potentially being more stable or abundant in cfDNA.
  • Longitudinal Monitoring is Key: Tracking both mutations and methylation patterns over time, rather than just at a single time point, is crucial for understanding disease dynamics and predicting recurrence. Dynamic changes in these markers provide a more comprehensive picture of residual disease.
  • Early Recurrence Prediction: The most striking finding is that elevated MRD scores (based on methylation) were observed in all five relapsed patients in their sub-cohort, and this elevation occurred significantly before radiological confirmation of recurrence (0.5-7 months prior). This suggests a powerful predictive capability for this approach.
  • Complementary Nature of Mutation and Methylation: While mutation detection alone has limitations, especially in early-stage disease, combining it with methylation profiling enhances both sensitivity and specificity. Four out of five relapsed patients showed an increase in mutation allelic fraction alongside elevated MRD scores, indicating a synergistic effect.

Key Results

  • Mutation Detection: At baseline, somatic mutations were detected in 100% of tumor tissues and 63.1% of paired plasma samples. Post-surgery, the detection rate of ctDNA mutations significantly dropped to 23-25% at follow-up.
  • MRD Score (Methylation-based): The MRD score was positive in 95.7% of patients at baseline, dropping to 74% at the first follow-up and 70% at the second.
  • Recurrence Prediction: In the sub-cohort of 5 relapsed patients with parallel mutation and methylation profiling, all 5 showed elevated MRD scores between 0.5 and 7 months before radiological recurrence.
  • Four of these patients also had a concomitant increase in mutation allelic fraction.
  • One patient had no detectable mutations throughout follow-up but still showed an elevated MRD score prior to recurrence.
  • Correlation: A strong positive correlation (0.89) was found between the maximum allelic fraction of mutations (maxAF) and the methylation MRD score, suggesting their combined utility.

Limitations and Future Work

The authors acknowledge several limitations:

  • Small Cohort Size: The study included 65 patients, and the sub-cohort for longitudinal analysis was even smaller (35 patients). This limits the generalizability of the findings.
  • Short Follow-up Time: The median follow-up was 132 days, with a maximum of 676 days. Some patients with elevated MRD scores might experience relapse later, beyond the observed follow-up period.
  • Limited Biomarker Integration: The study focused solely on mutation and methylation profiling. Integrating other biomarkers, clinical symptoms, and imaging data could further improve risk stratification.
  • Need for Larger, Longer Studies: The authors emphasize the need for studies with larger cohorts, consistent sampling times, and longer follow-up periods to validate these findings and refine the prediction models.

Future work should focus on these areas to solidify the role of longitudinal ctDNA mutation and methylation profiling as a robust tool for predicting recurrence and guiding treatment decisions in NSCLC patients.

Summary updated

Abstract

GROWING EFFORTS ARE BEING INVESTED IN INVESTIGATING VARIOUS MOLECULAR APPROACHES TO DETECT MINIMAL RESIDUAL DISEASE (MRD) AND PREDICT DISEASE RECURRENCE. IN OUR STUDY, WE INVESTIGATED THE UTILITY OF PARALLEL LONGITUDINAL ANALYSIS OF MUTATION AND DNA METHYLATION PROFILES FOR PREDICTING MRD IN POSTOPERATIVE NON-SMALL-CELL LUNG CANCER (NSCLC) PATIENTS. TUMOR TISSUES AND LONGITUDINAL BLOOD SAMPLES WERE OBTAINED FROM 65 PATIENTS WITH RESECTED STAGE IA-IIIB NSCLC. SOMATIC MUTATION AND DNA METHYLATION PROFILING WERE PERFORMED USING ULTRA-DEEP TARGETED SEQUENCING AND TARGETED BISULFITE SEQUENCING, RESPECTIVELY. DYNAMIC CHANGES IN PLASMA-BASED MUTATION AND TUMOR-INFORMED METHYLATION PROFILES, REFLECTED AS MRD SCORE, WERE OBSERVED FROM BEFORE SURGERY (BASELINE) TO POSTOPERATIVE FOLLOW-UP, REFLECTING THE DECREASE IN TUMOR BURDEN OF THE PATIENTS WITH RESECTED NSCLC. MUTATIONS WERE DETECTED FROM PLASMA SAMPLES IN 63% OF THE PATIENTS AT BASELINE, WHICH SIGNIFICANTLY REDUCED TO 23-25% DURING POST-OPERATIVE FOLLOW-UPS. MRD SCORE POSITIVE RATE WAS 95.7% AT BASELINE, WHICH REDUCED TO 74% AT THE FIRST AND 70% AT THE SECOND FOLLOW-UP. AMONG THE 5 RELAPSED PATIENTS WITH PARALLEL LONGITUDINAL ANALYSIS OF MUTATION AND METHYLATION PROFILE, ELEVATED MRD SCORE WAS OBSERVED AT FOLLOW-UP BETWEEN 0.5-7 MONTHS PRIOR TO RADIOLOGIC RECURRENCE FOR ALL 5 PATIENTS. OF THEM, 4 PATIENTS ALSO HAD CONCOMITANT INCREASE IN ALLELIC FRACTION OF MUTATIONS IN AT LEAST 1 FOLLOW-UP TIME POINT, BUT ONE PATIENT HAD NO MUTATION DETECTED THROUGHOUT ALL FOLLOW-UPS. OUR RESULTS DEMONSTRATE THAT LONGITUDINAL PROFILING OF MUTATION AND DNA METHYLATION MAY HAVE POTENTIAL FOR DETECTING MRD AND PREDICTING RECURRENCE IN POSTOPERATIVE NSCLC PATIENTS.
BibTeX
@inproceedings{li2021potential,
  author = {Hang Li and Zelin Ma and Bin Li and Yunjian Pan and Jiaqing Xiang and Yawei Zhang and Yihua Sun and Ting Hou and Analyn Lizaso and Yan Chen and Xi Li and Hong Hu},
  title = {{Potential utility of longitudinal somatic mutation and methylation profiling for predicting molecular residual disease in postoperative non‐small cell lung cancer patients}},
  booktitle = {Cancer},
  year = {2021},
  doi = {10.1002/cam4.4339},
}
Figure