Dr. Jonathan Frampton, Product Manager - Diagnostics, Horizon Discovery
Dr. Hadas Amit, Product Development - Diagnostics, Horizon Discovery
We will discuss the effects of formalin on clinical diagnostics and how these can be minimized.
How do we manufacture our reference standards?
At Horizon, we have developed a library of over 550 genetically defined cells harboring mutation found in cancer patients as well as their control cell lines from the same genetic background.
These cells are extensively validated before undergoing cell line engineering to produce the mutant cell line of interest. As part of the cell line establishment process it is critical that we single-cell dilute the original “wild type” cell line to ensure clonality. For pre- and post-validation we run SNP 6.0 analysis to confirm cell line identity, sanger sequencing to confirm engineering has been accurately targeted to the endogenous gene, digital PCR to confirm copy number and allelic frequency and RT-PCR to analyze gene expression.
Using our engineered cells lines we manufacture a range of genetically defined reference standards – DNA, RNA, FFPE blocks and sections, and cells slides for IHC and FISH.
In the next slides I will expand more on our new products, formalin compromised DNA reference standards.
Looking at DNA samples derived from different formalin fixed paraffin embedded FFPE sections used routinely in diagnostic labs, a high degree of variability can be observed. This variability at least partially originate from different formalin intensity levels leading to differences in, fragmentation levels, amplifiability and DNA chemical modification – such as cytosine deamination and cross linking. These characteristics can have a profound impact on clinical diagnostics leading to assay failure.
Our formalin compromised DNA RS mimic DNA derived from FFPE sections. Using matched pairs of formalin-compromised DNA and non-compromised DNA samples from a common mixture of genetically defined cell lines can be then used as powerful controls when assessing assay performance while taking into consideration the impact of formalin.
When assessing assay performance, one very important consideration is DNA quantification.
There are different DNA quantification methods available, including spectrophotometric-based, such as the nanodrop, Fluorescent-based which rely on a fluorescent intercalating agent that binds dsDNA – for example the QF assay available from Promega and the Qubit Assay by Invitrogen, as well as the PCR-based methods such as qPCR and ddPCR.
For formalin compromised DNA RS we do not recommend using PCR-based methods to quantify DNA because the amplifiability of such samples are markedly reduced due to the fragmentation and cross linking. qPCR assays can rather provide information on the amplifiability of the sample and thus serve as a quality control method.
We have looked into the DNA quantification methods for formalin compromised material and measured matched non-compromised and formalin-compromised DNA RS by QF, Tapestation and nanodrop. The graph on the top represent the non-compromised DNA and the one on the bottom represents the matched formalin compromised DNA. As you can see there is variation in the concentration of DNA measured by different methods, with an overestimation of DNA conc using the nanodrop and even higher overestimation in formalin compromised DNA. Similar results were obtained when we compared another fluorescent assay the Qubit assay and the nanodrop and again we have found an over estimation in the nanodrop reads vs the qubit.
Altogether this results are consistent with a recent study published a few weeks ago providing insights on failures in mutation detection on FFPE-derived DNA. The study showed that inadequate DNA quantification is a major concern leading to assay failure and B-Raf and EGFR mutation detection. The study also showed in some cases assay failure originated from DNA quantification by the nanodrop which overestimates DNA concentration when compared to the qubit assay or due to improper quantification using the qubit. If you are interested to learn a bit more about the study, you can download it for free.
Now let’s have a look at the effect of formalin on mutant variant detection. Here we were in particular interested in the EGFR and B-Raf locus as these regions are known to be vulnerable to formalin artefacts due to their structural complexity.
We have generated matched pairs of non-compromised and formalin compromised B-Raf RS with defined B-Raf AF, ranging from 5% down to 0.2% and measured allelic frequency using ddPCR. As you can see formalin-compromised DNA shows slightly higher quantified AF. These artefacts probably result from variability in amplifiability of the mutant versus the wt alleles resulting and artefacts in ratios.
Here is another example showing the impact of formalin on assay sensitivity and sensitivity. We produced a range of formalin-compromised DNA samples with increasing levels of formalin intensity levels, all samples are 100% wild type for EGFR. As you can see there is an artificial increase in mutant calling by ddPCR with increased formalin intensity. These data is consistent with the known deamination effects arising from formalin on the EGFR T790 locus leading to non-specific amplification of the mutant allele.
To take it to the next level, we produced a range of products with a gradient of EGFR T790M mutant AF from 3% down to 0.1% or wild type control as well as matched pairs of non-compromised and formalin-compromised DNA RS. Samples were sent blinded to one of our customers who are developing a novel PCR-based technology for the detection of EGFR T790M. Here you can see the mutant calling measured in 8 replicates and two independent, blinded samples for each. Using our standards they could provide compelling evidence for the specificity and sensitivity of their assay. It is also evident that this platform handles well increasingly lower mutant copies / AF with increased formalin intensity which can be pretty challenging to achieve the correct mutant calling.