Name: Peter Papenhausen
Title: National Director of Cytogenetics, Laboratory Corporation of America
With more than 40 years of cytogenetics experience behind him, Peter Papenhausen witnessed the advent of most of the technologies used in laboratories today, from microscope-slide-based karyotyping to fluorescence in situ hybridization to chromosomal microarray analysis, and now to next-generation sequencing.
In his position at LabCorp, he has overseen the implementation of array-based testing, first in pediatric constitutional cases, later in prenatal constitutional cases, and, more recently, in cancer testing.
While CMA has increasingly been adopted for hematological malignancies, its adoption for assessing solid tumors has been slower, with some raising questions about the quality of information obtained from such heterogeneous samples, as well as the expense of using multiple technologies on the same sample.
Papenhausen, though, is an advocate of CMA, and has introduced through LabCorp array-based testing for a number of solid tumor categories, including brain tumors and Wilms' tumor. To better understand his view on using CMA for cancer testing, BioArray News interviewed Papenhausen recently. Below is an edited transcript of that interview.
How did LabCorp go about introducing chromosomal microarray analysis for cancer cases?
We felt we had to run constitutional studies for an extended period of time before we started doing cancer cases, because we needed to know what the normal variation was. We needed to know what we could expect as far as homozygosity was concerned. The main thing was being able to determine an acquired change from a constitutional change, and attribute that to cancer. You don't want to have to run normal tissue in a second array. The main objective in these cancer analyses is to be able to make a determination from a single run of the tissue. It helps to have some normal tissue in the test sample because you can see the difference between the two in the same analysis. You can see which regions are 100 percent normal, and what is a lesser percentage, presumably the acquired leukemic cells. This normal mix gives you an internal control standard.
So we ran about 50,000 constitutional samples before we started doing cancer, even though I was especially interested from the very beginning, going back to the 1990s, [when] I was trying to prove that mitotic recombination was taking place. That's why I wanted to go to SNPs, instead of a comparative genomic hybridization array. The dosage-related SNP calls allow for determination of both constitutional [uniparental disomy] and acquired [loss of heterozygosity]. What I didn't anticipate, however, was that they would be so common. Both constitutional UPD and acquired LOH are incredibly common. I was occasionally seeing evidence of a UPD in cytogenetics in leukemia, but only in cases where there was an observable translocation that had duplicated a derivative. I hadn't really considered how many driver mutations would lead to so-called copy-neutral LOH. But it became eminently clear in validation, and now we have been doing cancer cases for about two years.
Have you always been running it on CytoScan?
No, we first started running cancer cases on the Affymetrix 6.0. We have used other products in the past. We like CytoScan because of its robust coverage of cancer genes and its high percentage of total probes. The 6.0 was a GWAS array. It was originally designed for linking genes to SNPs, not for the cytogenetic analysis of imbalance, although the high density of the array still provided excellent results.
It seems like arrays were first adopted for hematological malignancies, perhaps because cytogenetic testing was already established for blood cancers.
It's not so much that any kind of establishment took place, but because they are dosage-related disorders. That is what the microarrays primarily reveal, although SNP arrays also detect copy-neutral changes. But what it doesn't get is balanced translocations. Therefore it is not optimal for any disorder that has a propensity for balanced translocations, such as small round cell tumors. More appropriate are brain tumors – most clonal evolution in brain tumors is dosage related. Wilms' tumor, on the other hand, is often based on a copy-neutral alteration. In terms of solid tumors, our main targets are Wilms' tumors and brain tumors, but we have been looking at all types.
And if the information is useful, what is that lending itself to? Better treatment decisions?
The hope is always that there will be some targeted therapy for the genes or the amplifications involved. In hematological disorders, a lot of what we do is based on prognosis. A lot of what we do is related to the type of therapy that is given. In other cases, such as chronic lymphocytic leukemia, we use it to stratify patients. If they are in a good prognosis subgroup, oncologists often leave them alone. If they are in a poor prognosis subgroup, they treat them. The best prognosis subgroup is a deletion 13, by itself. I've discussed one case ... in which the universally performed FISH analysis showed just a deletion of 13, and this person would have been placed in the best prognosis group. But as I showed, there were three other changes when the whole genome perspective was looked at, including loss of heterozygosity across p53, the poorest subgroup. So this person went from having the best prognosis to the poorest prognosis and really needs to be treated.
How do people order this testing? Do they specifically request a cancer array?
They order a chromosomal analysis. Sometimes they say, "Wilms' tumor, please run a chromosomal analysis." I might say, "Do you realize that the main abnormality can't be seen by chromosomal analysis? However, if you would like to order an array, we are usually able to obtain diagnostic results." It might go that way. A lot of times we talk to them, and explain the best uses of these tools, so they can order them appropriately.
But there are some people who are skeptical about the use of new technologies.
They are probably the same people who are always skeptical of new technologies. I have been in cytogenetics since 1969, but have always embraced changes. What I am suggesting could be a major change to conventional cytogenetics. But much of it, except the disorders with balanced translocations that require FISH analysis, would be better off done with an array. You have 200 times the resolution and the ability to detect copy neutral clonal evolution. Why would you continue to do something that provides you with lower resolution and fewer clone-specific changes?
Some would say it's too expensive.
Well, it is more expensive, but the price is going down. When we first started doing arrays for ALL, I thought, "We're doing FISH, chromosomes, and now arrays? I don't know if we are going to be able to afford all three and be cost effective." And over the course of time, we are finding so much on the ALLs that I am saying, "Don't do chromosomes anymore." You still need FISH because of the balanced translocations. Still, array and FISH is going to provide more clonal insight than chromosomes and FISH, and the quality of the analysis is just so much better due to the often poor chromosome morphology in ALL.
What do you think of the quality of the arrays on the market?
There were a lot of people contributing suggestions to the CytoScan design. I think it's a good product; I don't have experience with other arrays that have included SNPs. I strongly feel that arrays need to provide SNP genotype information, particularly for cancer.
It seems that more labs are adopting arrays to study miscarriage and stillbirth.
Well, there are many advantages. I gave a talk recently about using SNP arrays for [product of conception] analysis. There are so many problems with cytogenetics and doing POCs. You are putting in necrotic fetal tissue and [it is] competing against fresh maternal tissue. Guess what grows out? You can have 90 percent necrotic fetal tissue and the five percent of maternal tissue will completely replace it in a week. Allelic admixtures and homozygosity in the array results also offer diagnosis of molar pregnancies, maternal cell contamination and trisomy rescue, and uniparental disomy. We have been offering this at LabCorp for a few years now.
And there is also the question of where next-generation sequencing fits into this.
It's a massive job to sort out all of the nononcogenic alterations and the normal variation. It is difficult doing that with the microarrays, and it is just hugely magnified when you are doing a molecular sequencing analysis. The analysis has to be quantitative, too, but new techniques can help this and sorting out the complexity should just be a matter of time.