Whole-genome pre-amplification as a viable approach for genomic screening of FFPE-derived DNA samples

This study demonstrates that DNA ligation-mediated multiple displacement amplification (DLMDA) effectively boosts DNA yield and preserves global copy-number alteration patterns in aged FFPE prostate tumor samples for whole-genome sequencing, although it significantly reduces the sensitivity for detecting specific amplifications and deletions.

The Gist

Imagine you are a detective trying to solve a crime by looking at a shredded, water-damaged, and faded piece of evidence. This is exactly the challenge scientists face when studying cancer DNA taken from old medical samples stored in wax blocks (called FFPE).

Over time, the DNA in these blocks gets broken into tiny, useless fragments. Often, there isn't enough of it left to run a full genetic scan (Whole-Genome Sequencing) to find the "criminal" mutations driving the cancer.

The Problem: Too Little Evidence

To get a full picture of the cancer, you need a lot of DNA. But with old samples, you might only have a few crumbs. If you try to scan them directly, the computer can't see the whole picture.

The Solution: The "Photocopier" (DLMDA)

Scientists tried using a special molecular "photocopier" called DLMDA (DNA Ligation-Mediated Multiple Displacement Amplification).

• How it works: First, it uses a "glue" to stick the tiny broken DNA fragments back together into longer chains. Then, it uses a super-efficient enzyme (a biological machine) to make millions of copies of that DNA.
• The Goal: To turn a few crumbs of DNA into a full feast, so the scientists can scan the whole genome.

The Experiment: Testing the Photocopier

The researchers took 22 old prostate cancer samples (some 5 years old, some 15, some 20) and compared two groups:

1. The Control Group: DNA scanned without copying.
2. The Test Group: DNA that was glued and copied (amplified) for either 2 hours or 8 hours.

The Results: A Mixed Bag

1. The Good News: We Got More DNA!
The "photocopier" worked incredibly well. It increased the amount of DNA by 42 to 86 times. This means samples that were previously too weak to test could now be analyzed. It's like turning a single grain of sand into a whole beach.

2. The Bad News: We Missed Some Clues
While the photocopier made more DNA, it wasn't perfect at copying everything.

• The "Dropout" Effect: The machine missed some specific parts of the genetic map. It failed to detect certain "deletions" (missing pieces) and "amplifications" (extra copies) that were actually there in the original sample.
• The Analogy: Imagine you are trying to copy a map of a city. The photocopier makes a huge, clear copy of the whole city, but it accidentally leaves out a few small alleyways and parks. You can still see the main roads, but you miss some specific details.

3. The Surprising Good News: No Fake Clues
Usually, when you copy something imperfectly, you create fake errors (false positives). You might think a street exists when it doesn't.

• The Finding: The researchers were relieved to find that this method did not create fake errors. It didn't invent new mutations or distort the map in a specific, biased way. The "missing" parts were just randomly scattered; the machine didn't preferentially ignore specific neighborhoods.

The Verdict: Is It Useful?

Yes, but with a caveat.

Think of this method as a wide-net fishing strategy.

• Without the net (no amplification): You catch nothing because the fish (DNA) are too small and scattered.
• With the net (DLMDA): You catch a huge amount of fish, so you can study the ocean. However, the net has some holes, so you miss a few specific fish.

The Conclusion:
This technique is a viable tool for screening. It allows scientists to study old, precious cancer samples that were previously impossible to analyze. It gives a reliable "big picture" of the cancer's genome without inventing fake problems.

However, because it misses some small details (the "holes in the net"), scientists need to be careful not to miss important targets. They need to keep improving the "glue" step to make the net finer, ensuring they catch every single clue without losing any.

In short: It's a powerful rescue tool that saves old data from being wasted, but it's not yet perfect enough to see every tiny detail. It's a "good enough" solution for now, with room for improvement.

Technical Summary

1. The Problem

Formalin-fixed paraffin-embedded (FFPE) tissue samples are the primary source of archival clinical material for retrospective cancer studies. However, formalin fixation causes DNA fragmentation, cross-linking, and chemical damage, leading to low DNA yields and poor quality.

• Limitation: Whole-genome sequencing (WGS) is the gold standard for comprehensive genomic analysis (detecting Copy Number Alterations [CNAs] and Single-Nucleotide Variants [SNVs]), but it often requires more DNA input than can be extracted from small or old FFPE biopsies (e.g., prostate cancer).
• Current Solution & Flaw: Whole-genome amplification (WGA), specifically Multiple Displacement Amplification (MDA), is used to increase DNA yield. However, standard MDA is known to introduce representation bias and allelic imbalance, often distorting CNA profiles and generating false-positive CNA calls, making it unreliable for CNA screening in degraded FFPE samples.

2. Methodology

The study evaluated DNA ligation-mediated MDA (DLMDA), a protocol designed to repair fragmented DNA templates before amplification, to determine its suitability for WGS-based CNA analysis.

• Sample Cohort: 22 archival FFPE prostate tumor blocks were analyzed, stratified by age: 5-year-old (n=8), 15-year-old (n=9), and 20-year-old (n=5). All samples had ≥30% tumor content confirmed by H&E staining.
• Experimental Design:
  • DNA was extracted from each block.
  • DLMDA Protocol: DNA underwent random ligation (24°C, 30 min) followed by MDA amplification using the REPLI-g FFPE Kit.
  • Time Points: Samples were split into three groups: Non-amplified (0h), 2-hour amplification, and 8-hour amplification.
• Sequencing & Analysis:
  • Libraries were prepared and sequenced on an Illumina NovaSeq 6000 (2x101bp).
  • Bioinformatics: Reads were aligned to GRCh38, deduplicated, and processed using HMMcopy to detect CNAs.
  • Metrics: The study calculated the Genomic Instability Index (GII), Jaccard Index (JI) for overlap of CNA regions, and compared the frequency of deletions and amplifications between amplified and non-amplified samples.

3. Key Contributions

• Validation of DLMDA for FFPE: The study provides empirical evidence that DLMDA can successfully recover DNA from highly degraded, archival FFPE samples (up to 20 years old) for WGS.
• Bias Characterization: Unlike previous studies suggesting WGA creates false positives, this study demonstrates that DLMDA primarily causes false negatives (loss of signal) rather than creating artificial CNA events.
• Independence from Sample Age: The performance of the technique was shown to be robust regardless of the archival age of the FFPE block (5, 15, or 20 years).
• Random Distribution of Errors: The study proved that the loss of CNA detection is randomly distributed across the genome, meaning DLMDA does not introduce regional bias that would skew specific genomic loci.

4. Key Results

• DNA Yield: DLMDA achieved a massive increase in DNA yield, ranging from 42-fold (2h) to 86-fold (8h) compared to non-amplified inputs.
• Global CNA Preservation: Global CNA patterns were largely preserved. The overall genomic instability profiles remained consistent between amplified and non-amplified samples.
• Reduction in CNA Sensitivity (False Negatives):
  • There was a significant reduction in the number of detected CNA deletions and amplifications in pre-amplified samples compared to non-amplified controls.
  • This reduction was statistically significant for both 2h and 8h reactions and was independent of FFPE block age.
  • Consequently, the Genomic Instability Index (GII) decreased in amplified samples.
• No Regional Bias:
  • Analysis of the Jaccard Index showed that the CNA "dropouts" (events detected in non-amplified but not amplified samples) were randomly distributed.
  • There was no evidence that specific genomic regions were systematically missed, indicating the technique does not distort the genomic map.
• Reaction Time: Extending the reaction from 2h to 8h increased DNA yield significantly but did not introduce additional bias or further reduce CNA sensitivity beyond the initial drop seen at 2h.

5. Significance and Conclusion

• Clinical Utility: DLMDA is a viable strategy for CNA screening pipelines in WGS when working with low-yield FFPE samples. It allows researchers to utilize otherwise unusable archival blocks, expanding cohort sizes for retrospective studies.
• Trade-off: The technique offers a trade-off: it eliminates the risk of false-positive CNA artifacts (a major issue with standard MDA) but introduces a risk of false negatives (reduced sensitivity).
• Recommendation: The authors conclude that DLMDA is suitable for screening purposes where avoiding false positives is critical. However, to improve sensitivity and minimize false negatives, future work must focus on optimizing the DNA ligation step to ensure more complete reconstitution of fragmented templates.
• Future Directions: The study suggests that comparative analyses using equally pre-amplified samples as a baseline can mitigate false-negative risks. Further research is needed to compare DLMDA against other WGA methods (like MALBAC) specifically in FFPE contexts and to evaluate its impact on SNV calling (though MDA is generally known to preserve SNVs well).

In summary, the paper establishes DLMDA as a robust tool for recovering DNA from difficult FFPE samples for WGS, provided that the user accepts a reduction in CNA detection sensitivity in exchange for high-fidelity, bias-free amplification.