Please can you give a brief introduction to papillary thyroid carcinoma (PTC) and who it affects?
There are two types of thyroid cells and therefore there are two broad types of thyroid cancer. Medullary carcinoma is derived from parafollicular or C cells, whereas follicular cells give rise to several types of thyroid cancers.
Papillary carcinoma is derived from follicular cells and is the most common type, representing about 85% of cases.
PTC affects a wide range of patients, from children to late adults. The incidence of thyroid cancer peaks in the 45 to 55 range and the median age is 50.
Women are affected about 3 times more often than men. People with radiation exposure are at risk of developing thyroid cancer.
PTC sometimes runs in families. It can be part of a syndrome or it can run in families by itself, which is called familial nonmedullary thyroid cancer. The genetics of familial nonmedullary thyroid cancer is unclear, but a lot of work is being done in this area.
How much evidence is there that the incidence of PTC has increased over the last few decades and what do you think are the reasons for this?
It is very clear from NCI databases, such as the SEER database, that the incidence of thyroid cancer has been steadily increasing since 1990, during which time the incidence has almost tripled.
It has been reported in the New England Journal and the New York Times that in Korea the incidence has increased almost tenfold.
There are probably many factors involved. In Korea there has been a lot of discussion about overdiagnosis and overscreening, but in the United States it is probably a little more multifactorial.
Overall, doctors are trying harder to find thyroid nodules and cancers. Ultrasound of the neck has improved and become more widely use, and so it directly follows that we are finding more nodules, some of which turn out to be cancers.
Also, pathologists are trying harder by examining more of the thyroid tissue that is resected during thyroid surgery.
Finally, pathologists have expanded the types of nodules that are diagnosed as carcinomas, and this is an area of controversy in thyroid pathology. Many people are excited by the prospects for molecular pathology to help resolve some of this diagnostic controversy.
So there has been a lot of excitement in the thyroid field for ancillary molecular testing to help determine who has thyroid cancer and how we should treat them. But having said all that, there is also evidence that the true incidence of thyroid cancer has also increased and the reasons for that are not entirely clear.
Could you please outline your work in mapping genetic mutations found in thyroid cancer?
We did a large study of nearly 500 tumors through The Cancer Genome Atlas program (TCGA), which is a joint effort of the National Cancer Institute (NCI) and the National Human Genome Research Institute – both of which are within the National Institutes of Health (NIH).
TCGA started in 2006 as a pilot and then was expanded in 2009 with funds from the Stimulus package, which allowed more types of cancer to be studied in this program. Thyroid cancer was included in the expanded phase and since PTC is the most common type, it was chosen for study.
TCGA has a standardised scientific pipeline, much of which is based on high throughput DNA sequencing. The technical aspects are a little complicated but essentially we’ve done DNA sequencing of what’s called the whole exome, which is the coding regions of the genome; RNA sequencing, microRNA sequencing, DNA methylation, copy number assessment (either gains or losses in the DNA) and protein expression.
The amount of data produced is quite amazing. The challenge is using all the data to tell pathologically and clinically relevant stories. To do that, we divided the analysis into three integrated steps.
First we determine the acquired or somatic mutations present in these 496 tumors. We found entirely novel mutations and also novel forms of existing mutations.
One of the most common mutations is a mutation in a gene called BRAF. The most common hotspot mutation in BRAF is called V600E and so about 60% of our tumors had BRAF-V600E and about 13% had a mutation in one of the RAS genes.
We then exploited that fact that tumors with BRAF-V600E and RAS mutations were mutually exclusive. Using some sophisticated bioinformatics, we showed that these tumors have distinct signalling properties. This is relevant to targeted therapy as thyroid cancer progresses to a more targeted therapy approach.
We also examined the role of thyroid differentiation which is a very important factor in thyroid cancer overall. Using a measure called the Thyroid Differentiation Score, we showed there is more biological diversity in these tumors than was previously appreciated.
Finally, we used all the data to derive molecular classifications of PTC. We showed that tumors with a RAS-like phenotype were strikingly different from tumors with a BRAF-V600E-like phenotype.
Moreover, we showed that the BRAF-V600E-like group has distinct molecular subtypes, which is important because this group is most often treated as a homogenous group in research studies. Our studies suggest that it’s really no longer appropriate to consider that as a homogeneous clinical group.
Our paper could essentially be split into three separate parts, but because we used the data from part one to inform part two and data from part two to inform part three, if we broke it up, it would be diminished.
We tried very hard to publish in a journal that would let us tell the story in its whole glory and Cell allowed us to do that. We were very pleased when we were published in Cell and they let us entend most of their guidelines for how long a paper can be.
Why are many papillary cancers still not genetically understood?
It is largely related to the advances in technology that we’ve seen since the completion of the Human Genome Project in 2000. People have studied thyroid cancer for a fair amount of time, but the technology that they have had at their disposal was so rudimentary compared to what we can do now.
The power of the current sequencing technologies is phenomenal and so we can examine the mutations in cancers much more thoroughly and easily now.
These advances will continue. People have talked about the $1,000 genome for a while now, but I think this will happen. Genomics is transforming nearly every field of biology and medicine and the only tension for actual medical translation and application is cost.
But the optimistic view is that doing genomics in clinical medicine will actually save money and pay for itself. In cancer, the example, if you can genetically determine that someone won’t respond to a particular course of therapy, then not only is that good for the patient but you spare the cost of doing something that’s destined not to work.
I think it’s a double-edged sword. It’s a very costly technology but, again, it might save money down the road. But the other point is we still have a lot of work to do to really fully translate this into the clinic.
How did your recent research reduce the fraction of PTC cases with unknown oncogenic driver?
Simply by finding novel mutations or novel forms of mutations in genes known to play a role in papillary carcinoma. For instance, as I mentioned, BRAF-V600E is the most common mutation in papillary carcinoma. Occasionally, BRAF can also become an oncogene by fusing with another genes. In our study, we found nine different BRAF partner genes, as well as some unusual BRAF deletions, which clearly shows that BRAF can be activated in a variety of different ways.
Using that as an example, we found ALK fusions, and new fusions of RET, and some PTEN deletions. By looking very thoroughly, we were able to reduce the fraction of PTC cases with unknown oncogenic drivers.
At the beginning of our study, if you genotyped 100 thyroid cancers with the existing knowledge, you only found drivers in about 70%, but we drove that up quite substantially.
We also found in a subset of tumors that they have what are called arm-level copy-number changes, meaning that they’ve either gained or lost a whole arm of a chromosome. While we didn’t actually pinpoint the genes that might be the drivers in that, because they’re mutually exclusive with the known drivers, we were able to argue that these copy-number changes represent novel drivers. It’s important for the field to recognize that this is a possibility. Again, we have to do more work in this area, but we think that those events can be oncogenic drivers.
If you’ll allow us to include those, along with some other rare mutations that we think may be drivers but we didn’t prove, we found drivers in almost all 402 cases except for 5. We thought that was a really big success of our study.
The really interesting thing is there are several molecular tests that are being developed and some of these mutations have already been incorporated into these molecular diagnostic tests and they’re actually finding some of the same things that we found. So, we’ve already had an impact in the care of patients by making the molecular diagnostic tests perform better.
How did they manage to incorporate the mutations into the tests so quickly?
One of the tests is led by my colleague and friend, Dr. Yuri Nikiforov at the University of Pittsburgh, and Yuri is part of this study. One of the advantages of being part of this study is you get a early view of the data and can act on it. So, that’s why I think it’s been implemented so quickly.
The truth is this study took about four years to complete. We started officially back in May 2010 and between sample acquisition, data generation and then analysis and paper writing, it took us four years to get it all done, so quite a lot of work.
Why do your results call for a reclassification of thyroid cancers?
Currently, the RAS-like and BRAF-V600E-like tumors are classified together as papillary carcinoma. However, out study shows that they have fundamentally distinct biologies across all genomic platforms. Also, these tumor have different pathologic growth patterns. So putting the pathology and genomic data together, our results suggest that these should not be considered part of the same tumor group.
Back in the ‘60s and ‘70s tumors that had the follicular growth patterns were actually classified as follicular carcinomas and then pathologists decided that they had some shared features with papillary carcinoma, so they were moved into the papillary carcinoma group.
In our paper, we suggest that it may be time to revisit and go back to the way it was done a long time ago. I think we’re just now starting to catalyze this conversation.
Some have argued ‘why bother?’ since they get treated the same, but as we more to more targeted therapy, the underlying biology does matter. Some early research that was published in the New England Journal by the group at Memorial Sloan Kettering suggests that these groups do respond differently to targeted therapy. So I fully expect that this study will catalysed a re-examination of the classification of thyroid cancer.
In fact, we just published a provocative editorial in the journal Thyroid with the goal of trying to get the conversation started. The first author is Sylvia Asa. We took a very provocative approach there, even asking the question “Does the follicular variant really even exist? Should it be part of follicular carcinoma?”
Thyroid pathologists are a passionate group and we will probably argue this for quite some time. But I think, again, we’re just catalyzing the discussion, which is exciting because a lot of the genomic studies the TCGA has done have not really gone down to this fundamental basic pathology classification level. So, I think the impact of our study will be great.
What impact do you think this reclassification would have on the management of the disease?
As the field moves to more targeted therapy, knowing the underlying biology and genetics of a given tumor will become essential. Having thyroid cancer reclassified along these lines will facilitate the transition to targeted therapy.
Beyond that, just having the most biologically relevant classification pays dividends in all kinds of ways, from basic science to pathology to clinical management.
More specifically, there’s a group of tumors that we call the encapsulated follicular variant of pathway thyroid cancer. If you go back to 1960, pathologists called those benign; they called them follicular adenomas. Then we discovered that they had nuclear features that were similar to papillary thyroid cancer, and so we’ve brought them into papillary carcinoma, but the truth is they very rarely act like true cancers.
Some have argued that pathologists have actually not done the right thing in overdiagnosing all these encapsulated, non-invasive follicular variants of papillary carcinoma, and that’s part of what our editorial is getting at--”What is the proper way to call something malignant? Is it the nuclei? Is it invasion?
In follicular carcinoma, which is the partner to papillary, you need to see invasion before you can call it cancer. If you don’t see invasion, we call it a benign nodule, we call it an adenoma.
I think this reclassification might actually impact the incidence issue that we discussed in question 1. We might go back to calling encapsulated follicular variant of papillary carcinomas either adenomas or, people have suggested a term borrowed from the breast world where we have ‘ductal carcinoma in situ’; follicular carcinomas in situ.
The advantage of calling one of those encapsulated follicular variants of papillary a ‘follicular carcinoma in situ’ is it gets the message across that the likelihood of this nodule actually becoming a clinically significant cancer is really low.
The benefit of that is that people then don’t get overtreated because one of the potentials for overtreatment is what’s called a completion thyroidectomy, where you have one lobe with a nodule taken out, it’s called cancer, and they go back and take the other lobe out, just thinking that that’s beneficial; or preparing to give the patient radioactive iodine.
It is fairly clear that in some places, patients with some of these non-invasive encapsulated tumors get overtreated. If we were to revise the classification to more precisely reflect the genetics and the biology, maybe that we could more precisely tailor the treatment to those features.
Would this lead to a watch and wait approach?
Yes, exactly. So, if you had one of these follicular variants - if I had one of those or a family member, I would say “You’re fine with a lobectomy. There’s no need to get a total thyroidectomy, there’s no need for radioactive iodine.”
But doing all of that extra treatment is probably too much. So, with the thyroid, the real challenge is more precisely matching up the disease with the treatment; having just this umbrella of papillary thyroid cancer is not granular enough.
What further research is needed to increase our understanding of the genomics of PTC?
One limitation of our study is that the cohort was rather recent, meaning we don’t have long term follow-up to see which patients recurred and did poorly. In thyroid cancer, because the prognosis is excellent, to do the long term follow-up really well, you need 20 years of data and it was just not practical for us to find 500 papillary thyroid cancers that old. So, that is a limitation of our study.
Some centers are studying older cohorts to overcome this limitation. There’s a group at MD Anderson that does have an older cohort, and they’re doing a sequencing study. It’s not as thorough as TCGA in terms of genomic platforms, but they’re looking for those mutations that are associated with recurrent disease and outcome.
However, I think, rather than doing more genomics on PTC, the better investment would be focusing on the other types of thyroid cancer beyond PTC, which represent the more aggressive forms. This is happening already at some cancer canters across the country. For instance, I know there’s a study at Memorial Sloan Kettering that is using a very comprehensive targeted genotyping assay called IMPACT that they developed there to look at some of these more aggressive cancers. Then we will be able to combine the results from TCGA with these future studies and together they will paint a fairly complete picture of the genomic landscape of thyroid cancer.
We argued to try to get some of those done through TCGA. The problem is finding the right tumor cohorts because the TCGA is under a strict timeframe and it was determined that we could not actually find enough of those tumors to really get it off the ground.
Where can readers find more information?
Our paper is published in Cell and is freely available.
More basic information on thyroid cancer can be found on the NCI website. The SEER program is excellent and has links to many other sites: http://seer.cancer.gov/
The Thyroid Cancer Survivors’ Association is also a rich source of information: www.thyca.org
See recent editorial published in Thyroid: Thyroid. 2014 Nov 19. Implications of the TCGA Genomic Characterization of Papillary Thyroid Carcinoma for Thyroid Pathology: Does Follicular Variant Papillary Thyroid Carcinoma Exist? Asa SL1, Giordano TJ, LiVolsi VA.
About Prof. Thomas J. Giordano
After AP training, he joined the faculty of the Department of Pathology at the University of Michigan Medical School as Assistant Professor, was promoted to Associate Professor in 2001, and Professor in 2008. He also holds a joint appointment in the Metabolism and Endocrinology Division of the Department of Internal Medicine.