Basic Molecular Pathology and Pre-Analytical Variables

Objectives

The learning objectives for today will be to review basic molecular biology and then to identify sources of pre-analytical variables that contribute to failed molecular diagnostic testing and sources of contamination that can happen in the gross room and histology laboratory. This is an area of disconnect since they are typically different labs that are not managed by the same pathologist. Then we'll conceptualize a model for standardizing cutting molecular curls and what might happen between the histology lab and the molecular lab.

This talk will consist of a brief review of molecular biology, preanalytical variables in surgical pathology, how tissues are processed in the path lab, and possible special considerations for molecular testing.

Basic Review of Molecular Pathology

Genetic information is encoded by DNA. These are packed into chromosomes that are pieces of DNA complexed with protein. There are 22 autosomes, including a pair of sex chromosomes. The important thing of the DNA structure is the nucleosome, which has 8 histone molecules. Why is it important in pathology? Nucleosomes are an octamer of histone protein, so 8 histones form a nucleosome bead, and a nucleosome has about 147 base pairs of DNA wound around it, which is important in surgical pathology because formalin stabilizes the histone and DNA bonds. The DNA molecules that wrap around these histones are somewhat protected, whereas during processing when we extract DNA away from FFPE tissue, the rest of the genome is somewhat fragmented in processing. DNA is generally protected.

Thinking about some of the circulating tumor versus cell-free DNA and ChIP seq; ChIP seq is a type of sequencing where you digest through sonication and MNase digestion, looking at the DNA that is wound around the histone-nucleosome complexes. You're digesting between the nucleosome molecules and looking at the DNA fragments that are here, which is why it is important to know why these DNA structures are important in downstream molecular testing. The other reason this is important is because we now know the circulating tumor DNA versus cell-free DNA has about a 20 base pair differential, and we think that is where this difference is. The yellow arrow is pointing at the melanoma circulating tumor DNA. They are typically about 147 base pairs, whereas cell-free DNA not from tumor are typically about 165 base pairs.

Moving on to the nucleic acid composition, the difference between DNA and RNA, is extremely important in our molecular lab because have to think about the bases of every assay. Are we testing DNA? Are we testing RNA? How we treat these depends on these two groups. The deoxyribose sugar at the two-prime position for deoxyribose—the H group—versus the OH group is extremely important. The other important thing is the base pairs, DNA having a thymine and RNA having a uracil.

The other key point in the review of basic molecular biology is the hydrogen bonds. Between adenine and thymine, you have 2 hydrogen bonds, whereas between cytosine and guanine-based pairs, you have 3 hydrogen bonds. This is the premise of melting temperatures, building primers, building any complementary sequence, as well as temperature melting curves, etc. These are very important DNA sequences where we look at the sugar phosphate backbone, 5' to 3' direction. Looking at our complementary sequence, A only binds to T or U (having 2 base pairs) and G binds only to C (having 3 base pairs), which dictates what our melting temperature in PCR, our next sequencing library preparation, and what all those reactions will be.

Next is RNA. RNA is transcribed from the DNA template, is generally single-stranded and therefore are not as stable, and can pair with single-stranded DNA or RNA. Some of the clinically-relevant RNA include messenger RNA, ribosomal RNA, and micro RNA. Here is a picture of transfer RNA.

The central dogma of molecular biology includes starting with DNA in the nucleus. Through transcription, you get mRNA, which is transported out of the nucleus into the cytoplasm, where it is translated into protein. This is a very condensed form; a lot more steps happen between there, but I will not belabor the details today in this basic review.

The next step in our basic review is gene structure and transcription. It is important to know that with the genes and molecular tests that we look at today, most of our next generation sequencing panels actually have targeted hot spot or exon sequencing. It means that we're focusing on these darker blue areas. However, it is important to know that the gene structure comprises a huge area. We're talking upstream. These could be very distal, far away from the gene. You can have distal control elements such as enhancers, more proximal control elements as upstream from your promoter, terminal regions, and intronic sequences, all of which play a role in production of mRNA and the final protein product.

I'm going to talk about splicing of introns because this is very important. This is a multi-unit protein that builds up what you call the spliceosome. The spliceosome binds to two specific sequences at the intron/exon junction, the GU and the AG, between where your intron and your exon site. These sequences are very important because they dictate whether you keep that intron or not. This leaves as a lariat structure with this protein, the spliceosome protein.

I will go into the types of mutations that are small mutations and big mutations. In this slide I'm showing a small example of types of point mutations that you might see. You may see the word synonymous or silent mutations. This is where you have a single base pair changewhich does not create any type of amino acid change. While we think of silent mutations as typically not causing any detrimental disease or alteration, this is not always the case. Sometimes silent mutations actually cause some phenotype downstream, but for most cases, we do assume that the silent mutations don't cause many problems.

Types of point mutations can also be substitutions or missense mutations, which means a single base pair change can lead to an amino acid change and cause a small change in the final protein product. If this happens to be in a key interaction site for your protein, this could be detrimental, but typically not too bad unless it is in a specific reaction site.

Then there is termination or nonsense mutation, where your point mutation/single base change causes a stop codon. The mRNA and protein would then be truncated or the mRNA would be the same, but the protein would be truncated and you would have a shortened protein. There are also frameshifts, where you typically have an indel or small insertion or deletion causing every amino acid thereafter to change from where your mutation is. Again, if you have a splice site mutation right where the intron is and a base pair change that causes your entire intron to be retained or lost where it should be or should not be, you have a major change in your protein downstream.

Other types of genetic mutations are the larger ones, which you can see from a karyogram. You can have a duplication, where a large chunk of the chromosome is duplicated upon itself, an inversion where the same large chunk of chromosome is flipped, the chunk of chromosome is gone or deleted, or it could jump into another chromosome called an insertion, a translocation where two chromosomes exchange chunks of themselves.

The last part of the review is nomenclature. While many of you may not be doing molecular testing yourselves, I'm sure many of you have seen a molecular report or will have to interpret some of this. You can refer to this website where a lot of us go to for guidance on how to name the mutations we see. There is a standard way to look at these. In nomenclature rules, we always refer to a reference sequence.

There are standard prefixes that tell what we're looking at, which are a genomic sequence, a mitochondrial sequence, a coding sequence, but most commonly are a genomic sequence, protein sequence or RNA transcript. For example, a genomic sequence would have your reference sequence, your location, and what the exact change was. Same thing for RNA transcripts, it would be what the reference sequence would be, where exactly it was, and what the change was. Same thing with protein, it would be your reference sequence, where the location is, and what the amino acid change is.

Molecular Testing

In general, we have some sort of nucleic acid manipulation. We may extract the nucleic acid out, denature it and keep it on the slide, or preserve it in some way so that we can do signal amplification. In the next step, that is where we amplify what we're trying to see. We may take the DNA or RNA molecule and amplify that through PCR, through library preparation, or through signal amplification where we attach a probe and then the probe has a signal amplification. Then we read this out in many ways.

The pathway from a patient to the molecular lab is convoluted. It can go to the CP lab and then come to the molecular pathology lab or can go from the patient to the anatomic pathology lab. Every hospital system is different and there are a lot of pre-analytical variables that happen between the patient and molecular lab. That is one of the things that we as pathologists deal with every day.

The gold standard for molecular testing is snap freezing, which is immediate storage in -80 degrees or -200 degrees until molecular testing can be done. However, this isn't the solution for every hospital system and is not a good solution for everything because there is always a loss of temperature control. This is extremely expensive and there is also a lot of risk with keeping a large freezer farm. There are risks associated with liquid nitrogen storage, including liquid nitrogen burns, supply tank explosions, and suffocation from liquid nitrogen leaks in a closed space. There have to be better ways of transporting specimens to the molecular lab.

Let's take a moment here to have a case vignette. It's 6 p.m. on a Friday and a lovely specimen walks in fresh. What do you do? Take 2 seconds to think about this in your office. This is a huge spleen. You might think of dried smears or taking frozen tissue. Do you want to make a molecular block? What is a molecular block? Do I need to use special fixatives for molecular pathology? Do you have molecular fixatives in your surgical pathology gross room? Do you need to worry about contamination? Could this be infection and if so, how sterile do you need to be in grossing this? What contributes to degradation? Has the spleen be embolized before removal? Those are some of the questions that as you're grossing this in, you could be thinking about. That may be an art we're losing now.

Preanalytical Variables in Surgical Pathology

What are the preanalytical variables that we think about? Variables that can sometimes occur when the test is ordered by the physician until the sample is ready for analysis. Pre-analytical variables can account for up to 75% of lab error, and most of that can be out of our hands. In surgical pathology, for today's talk, we will talk about from patient to diagnosis.

Specifically, factors that impact the molecular diagnostics are the specimen adequacy, the time and temperature between the patient and molecular lab, type of fixative that that specimen has seen, and how close to the protocol adherence to that fixative it was seen.

Contamination is a major problem, especially in assay sensitivity. How sensitive do we have to be? Downstream molecular testing and sensitivity of PCR is 1:104-6. Nested PCR is 1:107-8. Are you looking for an infectious organism or BCR-ABL? That is important to think about because contamination is a problem for false positives. Nucleic acids are highly susceptible to degradation, especially when wet.

Specimen adequacy. In a paper done by Melissa Austin and Colin Pritchard at UW, they looked at all the oncoplex cases, looked at what failed and what didn't, and said, "What is the minimum amount of tissue that we need to get a successful next generation sequencing run?" That is that increasingly in pathology we get tiny biopsy specimens, bone marrow cores, tiny needle cores, and FNA and smears, and people are asking us to get full next generation sequencing panels. This is a Hodgkin's lymphoma. It is a beautiful specimen. If only we could get this all the time. We don't get that type of nice excisions anymore. What they found out is with the needle core biopsy, the minimum amount of tissue needed would be two 1 cm cores by an 18-gauge needle. That's the paper.

A requirement of CAP is recording the histologic assessment of neoplastic cell content for molecular testing. One thing to realize is that no tumor specimen is ever 100%. In this Ewing sarcoma you still have stromal cell components in the background. In this invasive breast carcinoma you have a lot of stromal cells in the background that still have DNA content that isn't 100%.

What is the minimum tumor percentage to get a sensitivity in an NGS assay if the heterozygous mutation sensitivity is about 10%? Molecular pathologists think about what they need in a surgical pathology specimen and at the minimum what they need to see to call a mutation. We assume there are two copies of the gene only, but know that karyotypes of tumors are not always two copies. The percent of DNA with mutation is the tumor percentage divided by 2. For a heterozygous, only one of those two will have it, so 20% of the cells must be tumor, but you want a minimum of 40% tumor for your specimen. When thinking about submitting a specimen for molecular testing, think about how much tumor percentage would be in the block required to detect a mutation.

Time and temperature to fixation. Hypoxia, ischemia, metabolic stress, and macromolecule degradation start as soon as vessels are ligated in surgery, not when it removed from the patient. An embolization procedure 2 or 3 days before is when you start seeing degradation, ischemic change, and hypoxia. Higher temperatures without fixative also speed up autolysis. What happened to that tissue since it was removed from the patient? Was it sitting on the counter for 4 hours before coming to the grossing lab?

This is a study performed by Rob Bradley, where he took blood and solid tumor specimens and tested the RNA expression at 0, 4, 8, 24, and 48 hours. Key genes were also published in Cell, Science, etc., to have significant differences. He noted for instance that NOTCH2 had a significant expression difference in as little as 8 hours, and in LEF1 and PHF20, as little as 4 hours. In coding RNA, little difference could be seen in 4 hours, but by 8 hours, major differences. The solid line is room temperature and the dotted line is the same experiment performed with the tissue on ice. Incubating on ice did preserve a lot of the RNA. If processing time is delayed, put anything needing RNA work on ice.

Fixatives. Next I'll talk about fixatives and what could be done in the gross room to help ameliorate damages. There are different types of fixatives. Crosslinking fixatives are typically aldehyde-based, connecting two parts of the macromolecule, leading to enzyme inactivation. Coagulating fixatives allow better penetration of fixation into the tissue. Precipitating fixatives are typically alcohol-based, reducing the solubility of proteins and disrupt the hydrophobic interactions. Noncoagulating fixatives form a gel barrier, making penetration into tissue harder for other types of solutions.

Fixatives typically used are coagulating, the alcohol-based, and noncoagulating, the crosslinking being the formalin we typically are used to. Since this is a molecular talk, we will avoid mercuric-based fixatives and Bouin's-based fixatives.

Formalin, the 4% aqueous solution, actually preserves fairly good morphology. The crosslinking reaction causes the conformational changes and is the reason we need antigen retrieval in immunohistochemistry. It also fragments RNA, DNA, miRNA and other proteins. The fragmentation is caused by the crosslinking reaction, but this also preserves these nucleic acids. Now that the next generation sequencing is one of the most common modalities of molecular testing, it also requires fragmented nucleic acid and ends up being okay for FFPE and downstream NGS testing. Formalin requires 25 hours for every 3.9 mm of tissue to properly penetrate into the tissue, as well as for the crosslinking reactions to happen. You want 1mm to no more than 3 mm thick tissue, which is extremely difficult to cut. I, myself, have a hard time cutting that thin, but I definitely try. Remember, you need 15 to 20 times the volume of formalin to tissue to properly fix that in this time period.

What is wrong with freezing is that transporting the specimen is cumbersome. Storing the specimen is extremely expensive and requires temperature monitoring. The architecture would be disrupted and would not work with many special stains. Freezing would be good for RNA and DNA, but may be problematic in certain proteins because the thawing process may break some of that up.

Effects of decalcification on tissue. HCL solution extracts from RNA in the purine/pyrimidine bases, which is faster, but it leads to poor morphology. We like to use EDTA decalcification. It preserves nucleic acid and has a good enzyme for IHC, but is slower. If we think about immunohistochemistry and molecular-friendly fixatives, there are several options.

One paper by Dr. Singh is on the analysis of multiple different decalcifications. He looked at the mRNA yield and DNA yield from all these different fixatives. He found EDTA, immunocal, formical are okay and produce a fairly decent amount of RNA yield, as well as having an RNA CT, which means they performed PCR with these. Having a lower CT value is better and means there is good-quality RNA to do PCR. This study was performed by a second paper by Dr. Schrijver in 2016. This is some of the morphology and additional data supporting that work. You can look at some of the other fixatives if looking for instance for lipids or mucopolysaccharides.

Molecular-friendly fixatives. There are many proprietary fixatives that are great, but quite expensive. From a laboratory perspective, you would have to change your protocols and see how that impacts your own SOPs and validation of workflow, etc. In a clinical lab, each of these has different types of workflows. If you're a research lab, some of these you can get a recipe online to make yourself. The propriety ones here you can buy commercially and look up the recipes online.

This paper compared the fixatives we just looked at, including RNAlater, Allprotect, Paxgene, fresh frozen, and a few others. They discovered the red numbers looked similar for each of the different samples, Paxgene is a fairly decent comparator to fresh frozen. Again, having a lower CT value is better. The conclusion is that in terms of morphology and molecular tests, that was the best thing. Paxgene is about 9 times more than formalin, but check that yourselves I might be wrong on the current pricing.

For the morphology of Paxgene, the difference between formalin and Paxgene is similar, but not exactly the same in marrow and in spleen. But when it came to this lymphoma specimen, there was some shrinkage artifact in the Paxgene not identified on the formalin. We couldn't look at the nuclear detail here.

This is a comparison of RNA RIN values, which is a way for the molecular lab to look at the quality of the RNA. Higher RIN values are better, so our gold standard RIN values are up around 8-9 and formalin is around 7-8. FFPE is the lowest after processing at 4-7, but Paxgene stayed high around 7.

UMFIX is another type of fixative. You can see fairly decent nuclear detail here. This is an RT-PCR for a control gene that we typically use. When you compare the fresh frozen to UMFIX versus formalin, it is improved over formalin, but not as good as fresh tissue.

RNAlater versus standard is another type of fixative. What they did in this type of PCR is look at 28S and 18S RNA PCR, as well as P53 and GAPDH PCR. They did not do any CT values, but just did a gel. In this study where they looked at autopsy tissue fixed in RNAlater, the morphology is fairly well-preserved. In this study, they looked at lung cancer fixed in four different methods: FFPE, FineFIX, RCL2, and HOPE. The FFPE has the clearest, most distinct morphology. You sacrifice nuclear detail and cytoplasmic morphology with the rest of the fixatives. When looking at molecular advantages, FineFIX and HOPE are better, but I would not expect any pathologists to sacrifice morphology for that. Unless specifically collecting tissue for molecular, I would say morphology is still the most important, so we can deal with the FFPE and formalin.

Eight golden rules of proper fixation.

1. We want to make sure that fixative is penetrating from all sides and tissue is adequately fixed with formalin.
2. Opening cavities.
3. Perfuse the specimen (especially lung or complex vascular system).
4. Thickness is important (4 mm max).
5. Agitation is useful.
6. Adequate volume is useful, 15-20 at least in comparison to tissue versus fixative.
7. Allow sufficient time, not just penetration time, but time for crosslinking to allow for best preservation of nucleic acid if thinking about doing downstream molecular testing.
8. Warming of formalin can increase penetration time but does not necessarily increase crosslinking time, so room temperature is best.

How Tissues are Processed in the Path Lab

Contaminants in the gross lab can happen almost anywhere, including your processing board, filters, gloves, instruments, and gadgets. Every gross lab is different. Take what I'm showing in my lab to apply to your own lab. If you have questions, I'd be more than happy to answer to see how we can apply it to every situation. Our gross bench looks pretty clean, much cleaner than where I was a resident and a fellow. Where on this grossing bench is actually sterile and not contaminated? Pretty much nothing. In our gross lab is sterile or uncontaminated today, except for maybe the inside of the sterile gloves and the autoclave tools. We have contaminants possibly in paraffin. There are floaters in our water bath, grossing areas, and reusable cassettes. What are your cleaning schedules in your lab? Part of our SOP is how often you do that and how often you clean out your processors. Then time and heating components, how much does your tissue actually see in your heating schedule before moving on?

Think about your embedding station. How often do techs change gloves during embedding? In terms of tissues, are they cleaning forceps between specimens or using the same forceps the whole time? Are they wiping down hot and cold surfaces and are brushes are available to clean off teeth? If you have a papillary-type of carcinoma with friable tissue that falls all over the place, does anything prevent pieces of that tumor from getting into the next specimen?

Contamination can happen at any step. For instance, a specimen that was abscessed. Is there a cleaning process for all the equipment between that specimen and the next one? During grossing station, does any cleaning happen between one specimen and the next? It is probably unreasonable to do a complete sterilization procedure between every specimen. I recognize that because I have dual hats as an anatomical pathologist, as well as in the molecular lab. We also load 20-100 patients into the processor in every single hospital system. Each cassette has holes and cells could travel in between the specimens. Microtomes are used to cut sections from multiple different blocks. Patient material from one patient could be transferred into the next patient. How often are blades used during cutting changed? Are they changed between patients or not? Is the tech cutting the block wearing a hairnet, a face mask or what other protective materials are they wearing? If sick, are they working and is there a sick policy? Setting up the assay on a molecular bench means as we leave the surgical grossing bench and move into the molecular lab, many things change. A more sterile environment is treated with those.

Degradation of RNA and DNA depends on RNAses and DNAses. RNAses are very stable and hard to get rid of, but DNAses are not so stable and can destroy your sample. Molecular curls are important. We have a protocol, but not many places do, so be sure that you have one. Necrosis and molecular testing should be avoided if possible, but can be possible if you need a distinction between karylolysis and apoptosis and your necrosis is primarily apoptotic. When making a molecular block, think about an area of primarily viable tumor. Avoid necrotic tumor. You may consider a second area of normal tissue.

Take Home Points

Preanalytical factors important in surgical pathology to think about are: do you have enough specimen to have a successful molecular assay? What was the time and temperature to that specimen before sending it to the molecular lab, even before it hits your lab, that would allow for a successful molecular test? What type of fixative will you put on there? Can that fixative preserve your nucleic acid enough to allow for a successful test? I hope that as pathologist we can all ensure that we deliver the best care for our patients by following these steps and working together. We then allow for the best data to get out there and help our patients. The other important things to our molecular lab is contamination and degradation: not allowing the nucleic acid to degrade also not contaminating the specimen, especially when looking for things low in signal. We don't want those confounders. For AP, in the grossing room you can impact what the molecular test results look like. For CP, think about the preanalytical variables. If you get a funny result, question what is going on. Consider the molecular-friendly fixatives.

This is SCCA Pathology Lab. Kristen Shimp helped me with a lot of the stuff, so thank you very much. Here are some references that will be available. At this point I will turn it back to Rick for questions.

Q&A

Q: We still see many labs overprocessing their tissue. With this in mind, does the removal of bound water have any detrimental effects on FFPE molecular testing? 
A: Water would degrade nucleic acid, so having a desiccated specimen is actually better.

Q: How long can a tissue stay in formalin before proceeding to subsequent histological procedure in a situation where facilities are not immediately available? You said the minimum duration of fixation in formalin is at least 25 hours; what is the maximum time of fixating in formalin that will not disrupt the histological morphology of the specimen?
A: Historically we have thought that formalin would be a good fixative for forever. I have looked at specimens that have been fixed over 3 months. Morphologically they look good, but there are actually new CAP guidelines specifically for immunohistochemistry where overfixation impacts against the quality of IHC. Also, in a lot of the antigen markers they get degraded, so there is a sweet spot in how long you want your tissue to be in formalin, specifically when looking at antigens. If there isn't a good alternative, it is better to fix and keep it in fix than to not have anything. Except you want to mark your time, so you know what the limitations are because degraded tissue can't go back. Whereas, if antigens for specific IHC markers are lost, you may be able to do others.

Q: Can the thawing and freezing affect DNA or RNA samples?
A: Absolutely. The first thing we do in the molecular lab when we get a new reagent or new control is aliquot our RNA and DNA controls into tiny, separate containers, so we don't have to freeze/thaw our reagents and controls. Thawing and freezing has been shown to impact quality. For instance, freezing and thawing a block can definitely impact the quality of nucleic acid in there.

Q: Is there a good fixative for cfDNA and is it superior to routinely-available blood tubes?
A: There are a couple of cell-free DNA fixatives out there similar to the proprietary products I've shown you. The one I am most familiar with is the Streck tube. They're very expensive. If you are in a facility and can actually get to the specimen of EDTA-drawn blood within 4 hours and process it within 2 hours, you may not need that extra expense. What you would do is double-spin the blood within 2 hours and freeze it right away. If you have to process it over 24 hours later, then absolutely you will need one of those special tubes.

Q: Removal of bound water has an effect on protein conformation, but does the removal have an effect on FFPE molecular testing? 
A: Yes. Water is our enemy. We want to get rid of any hydration molecules because wet molecules/wet tissue are how you initiate allow for degradation reactions to happen at a molecular level. It doesn't just degrade protein; it degrades RNA or DNA and allows for the DNAses and RNAses to work, so having all of that removed is important.

Q: Can I store a specimen in 70% ethanol after fixation with formalin? 
A: I have heard of labs that store their specimen in ethanol, but I don't know the percentage.

Q: Lots of very nice comments. Which is better to run: ALK/ROS1 FISH or ALK ROS1/NGS testing?
A: It depends on your specimen, the tumor type, and also your lab. If you are thinking about cost or the lab, I am biased. I am a molecular pathologist and hematopathologist, so I would always want to do the molecular test because I think that by doing the next generation sequencing testing you get additional information. However, recognize that by doing the next generation sequencing test the cost could be higher because doing FISH on two probes essentially is generally lower. Secondly, the processing for FISH assay typically would be 2 days at our institution, whereas a next generation sequencing test is at the fastest 5 days, but typically 10 days. How fast do you need that result? Also, it is the overall cost of running two FISH probes versus running a next generation sequencing panel. Are you looking for other mutations as well? You may be able to look for the fusion, but may be able to find additional mutations that are actionable.

Q: How effective is it to retrieve DNA or RNA material from paraffin-embedded tissue?
A: We used to think we aren't able to do that, but increasingly it is becoming more standard. Retrieving DNA from FFPE tissue is quite standard now, especially in next generation sequencing labs because the DNA that is in FFPE is already fragmented to about the right size we need for library preparation. That is a pretty good specimen to do this. I would say RNA is more challenging, but we do it in our lab and typically get fairly decent amounts to do RNA work.

This follows another question that came in on the private line that says: how do you store long-term specimen? I think FFPE blocks are wonderful. We have done extraction of specimens that have been 20, 40 years old and have successful next generation sequencing runs and molecular studies.

Q: Is there any method to keep DNA or RNA samples with high integrity?
A: Assuming I have all the money in the world and this is perfect, I want to store extracted RNA/DNA in a cryotank on every patient—on a temperature-controlled freezer farm forever. That would be optimal and wonderful if we could do that.