Dr. Nelson N. Stone spoke at the 24th International Prostate Cancer Update on Thursday, February 20, 2014 on “Can Radiation Toxicity and Response be Predicted by Genetics?” In his presentation, Dr. Stone discusses ways to identify patients who may be more susceptible to radiation because of inherited mutations and what the future holds.

Presentation: 

Keywords:  mutation, radiation, SNPS, toxicity, symptoms, toxicity, genetic

Cite: How to cite: Stone, Nelson N. “Can Radiation Toxicity and Response Be Predicted by Genetics?” Grand Rounds in Urology. January 12, 2015. Accessed Apr 2020. https://grandroundsinurology.com/prostate-cancer-nelson-n-stone-radiation-toxicity-and-response/.

Transcript

Can Radiation Toxicity and Response be Predicted by Genetics?

I think I have some questions first before we do the next talk, which is on genetics and morbidity of radiation related toxicity. And you had this question this morning from Dr. Partin, so I want to test your recall. Genetic testing is well established in identifying populations with inherited germ line, inherited equals germ line, risk of several cancers. If one adds family history, age, and five selected at-risk SNPs to the relative risk for prostate cancer risk, so we know from family history, it’s around two or 2.5, but if you add everything together including the SNPs, what is the increase in relative risk? Is it, this should be 100% because Dr. Partin presented it this morning. Dr. Partin, are you here? I want you to go beat everybody up. The answer is ninefold. Okay.

The patient with prostate, this is a harder question, a patient with a prostate cancer desires to be treated with brachytherapy but is found to have an ATM alteration. His risk of radiation brought titers above the normal population is how much higher? Zero percent higher, no change; 20% higher; 30% higher; 40%; or 50%? That’s not bad; it’s actually between 20-30%. Very good, okay. Let’s go on to the slides.

Okay, so this was actually the first question. You have to add up all the things that make up a risk for prostate cancer that you know about like family history and age. But when you add the five of five selected at-risk SNPs, the relative risk increase for prostate cancer goes up ninefold. The problem is it only is described in 1.4% of the population. So it’s not like getting wild now, order this test on your patients and get this, because it’s only a small segment of the population that actually has, of the risk population that has these characteristics. So it’s not really very useful from a practical clinical perspective.

So let’s go a little bit over genetics. When we deliver radiation therapy to prostate cancer, we’re not just irradiating the prostate. We’re obviously irradiating the contiguous structures. And for the argument’s sake, you’re irradiating the rectum, you’re irradiating the bladder, you’re really irradiating the urethra because it’s right smack in the middle of your target, and no matter how well you give external beam radiation, your urethra is going to get probably at least 100% of the target dose. And you’re irradiating the neurovascular bundle.

And the damage done to DNA is dependent upon the amount of radiation you deliver. So this, what this data looks at is the number of genetic changes related to the specific Gray deliver, so with the most severe damage, which is a double-strand break, or a crosslink, you don’t get a lot of those events until you really kick up the dose of radiation.

So as we’re raising the dose of radiation and improving our targeting, we’ve gone, like I said, from 60 Gy way back 20 years ago to now the standard in the U.S. is 81 Gy, we are increasing these events. And we’re increasing these events not only in the cancer tissue, but we’re increasing these events in the normal tissue. So when we get a failed repair of a cell, because the cells are always doing to try and fix themselves, and if you get a double-strand break, you’re not likely to get a repair; you’re going to get cell death. And this cell death will occur in not only the prostate tissue but in the contiguous and urethral tissues that are around or in the prostate.

This phenomenon is not a straight line. It’s going to vary based on our genetic makeup. That’s the whole basis of this type of investigation is to try and find out ways to identify patients who may be more susceptible to this because of what they inherit, so-called germ line mutations, another name for inherited mutations.

The work really began at, in Philadelphia and in New York at Columbia by Dr. Hall, radiobiologist at Columbia when they first identified the ataxia gene and found that in a small population that had this gene mutation, there was a much greater likelihood of having late effects after radiation. This was a prostate study. It goes back to 1998. So that was really the beginning.

And we had known for many years that this gene was a real problem. It was known that individuals who have this mutation and go out in the sun and get exposed to sunlight get a real severe burn on their skin, much greater than what you would anticipate. And since the sun delivers photons and that’s what’s causing the burn, it made a lot of sense that radiation therapy, X-rays, which also deliver photons, would deliver the same type of damage to the tissue around the prostate, and exactly what they found.

One of, the radiobiologist who was working at Mount Sinai, Dr. Barry Rosenstein decided to get together with us and just, and see if we could investigate this same phenomenon in our brachytherapy patients. Now, the brachytherapy patients were sort of a unique population because they were so well studied and categorized. This is an example of intraoperative plan where the seeds are around the prostate, in the prostate, and we had the rectum and the urethra. And the yellow is the radiation dose cloud, and because we did intraoperative organ reconstruction followed by post-implant CT scans, we could extremely well characterize the dose of radiation to all of the structures that were of interest.

So brachytherapy is, for our purposes doing this genetic study, we see became a very good model because we were delivering very high doses. The dosimetry was variable because when you put the seeds in, you didn’t all, unlike external beam where you prescribe anyone Gray, sometimes you would get 140, sometimes you got 180, so you could measure the effects of variable doses on the structures and use that6 as one of the variables to look at the genetic influences.

We know that toxicity does happen. We can measure multiple toxicities. Patients live, so you can look at the late toxicity, very important because some of these things happen ten years later.

The toxicities had clinical meaning: urinary symptoms, rectal bleeding, erectile dysfunction. And this was also an important issue: we are at a point because of our biopsy data where the local failure rate was so low, that didn’t become a variable where tumor was growing and you had to put the patients on other therapies to interfere with their ability to measure the effects of the radiation vis-à-vis the toxicities.

So all these factors led to the realization that, yes, gosh, we’ve got a great thing here going, and this makes for a great opportunity.

The other thing is in 1990 when we started our implant program, I had come from a BPH, medical BPH clinical study environment, so I was very used to using detailed surveys for the patients. So when we started the brachytherapy program, we had patient questionnaires that we gave to all the patients before they had their brachytherapy, and then at regular intervals, every three months, and then six months thereafter, so we had a massive amount of data that we could analyze in terms of measuring the toxicity.

This was one of the first papers that we published looking at rectal toxicity based on radiation dose. And what we found, this was Wai, one of our medical students at the time, who became our radiation resident, that if we measured the volume of the rectum, and then measured how much of the dose was delivered to that volume, we could very accurately describe the likelihood of developing grade two radiation proctitis. So there’s the rectal dose, going higher, and here is the incidence of proctitis. So this became the standard of care. Everybody now knows they want to deliver the rectal dose below 1 cc of rectal volume to limit radiation proctitis.

But this is how the whole population looks. When we started to investigate the ATM alteration, we didn’t expect this but this is, was kind of an amazing result, that those patients who didn’t have the alteration, their likelihood of getting radiation proctitis was extremely low until you got up to a fairly substantial dose – – one way up. Here is the graph I showed you before, and here is the patient population with the ATM abnormality. So it is about a 20-30% increase over the normal that was the question.

The real issue for the clinicians, even though this is not present in a lot of patients, you don’t know who has it. So the bottom line is you always want to keep the rectal dose as low as possible because you don’t want to get into the situation of, oh my goodness, I gave him a normal dose of radiation to the rectum and he got a fistula. And you’re scratching your head saying, why the heck did that happen? It’s probably because of this reason. So always try and keep that rectal dose as low as possible.

Now, when we first started this hunt, it was using something called candidate genes, and those were genes that had already been identified, which were associated with some type of abnormality. And the candidate genes, like the ATM gene and TGF beta-1 and SOD2 and SRCC1, all have been implicated in damage following radiation therapy.

So we studied this. Now, this was a very laborious exercise because we had to do high pressure liquid chromatography to isolate these things, so you couldn’t go hunt for everything. But we did find associations, so if a patient had TGF beta-1 mutation, then the incidence of ED increased from 56% to 33% over the normal population. This is typically what we publish for our long-term data that ED is about one-third of the patients, but it almost was 60% if they had that mutation.

Our group also was very interested in breast cancer because that’s where a lot of the initial data came from. And so, we also looked at these phenomena, at ATM and the XRCCC, SOD2, and TGB, TG beta-1, and found associations with fibrosis after radiation therapy for breast cancer. So that’s no surprise that that’s showing up.

This is something like, remember what I said at the beginning, when you have the five SNPs and all of the information, and only have a 1.4% of the population? So this is not a type of investigation you can do in one institution. You have to have multiple institutions, literally thousands of patients you have to study to look at these associations. So we formed a cooperative group, and the cooperative group included several centers in the U.S., Denmark, Israel, Switzerland, France, and Japan, funded by the Department of Defense, New York State, the Danish Cancer Society, and the cohort group in France. And guys have gotten together and worked very well.

One of the major problems in doing this, unfortunately, is everybody likes to collect their toxicity data based on whatever they believe is the most appropriate instrument, and trying to marry those toxicities, the genetic output’s going to be the same, but the toxicity reporting is going to be different. And that’s been a major challenge.

Fast-forward: new technology. So no longer are we doing high pressure liquid chromatography; we’re doing genome-wide arrays to look for specific SNPs. So basically, what that is is you have to decide based on the fact that there are ten million SNPs in the human genome, you have to pick a group of SNPs; you can’t study all ten million. So you do, you get a chip that’s got maybe a million SNPs on it, and you do first a quick pass, and you see maybe 100,000 good hits. Then you refine it and make a SNP, make a chip, you can actually have the company make the chip with 100,000 SNPs, and then you get more refined.

And this is just number crunching. You go through 100,000 SNPs that you decided you want on your chip, and then you look at the outputs and you look at what correlates to specific outputs in terms of P values. And I’ll go over that in a minute.

But using that technology, we reported first this paper in African-Americans and ED after external beam, and when you use the clinical model, older age, hormones, whatever, this is what predicted ED; this is the area under the curve. And when you use the genetic model, the area under the curve went up substantially. So we were able to show that the genetic model was actually better than the clinical model in finding the patients that had erectile dysfunction after external beam.

Then we went into our own database, that wasn’t our database, that came out of a database in Queens, and started looking at our patients. And we divided the patients for this study into patients we called cases, they had ED after brachytherapy, or controls, they did not have ED. And they’re fairly well-matched in terms of age. Of course, you would think that there were more patients who had ED who got hormones and that was true. It was 68% of the cases versus 35% of the controls. And they were well-matched for all the other characteristics.

And so, the bottom line is here’s are all the SNPs that were identified that had significance, statistically significance. But notice none of them are really high: this one is 10.4 and this is 10-3. Nothing really is, strikes you as really way significant. Look at age: age had a P value of 10-15. Well, we know that’s always been true.
Turns out if you combine the SNPs, you get an increase in score that takes you down to 10-19. So here on this graph, I’m showing again the area under the curve to predict ED. There is the flip of the coin. The blue line is the clinical, the green line is the cumulative SNP score, and this line which is 89% is a combination of SNPs plus the clinical score.

So going forward, in order to probably refine what we’re going to do, it’s not probably going to be just the SNPs; it’s probably going to be a combination, in this case, the age, the use of ADT, and the use of EBRT in addition to the implant, all combined with the SNPs to be your predictor of erectile dysfunction.

What about urinary symptoms? Remember the urethra runs right through, and sometimes you’d see patients who all got the same dose of radiation, and he’s screaming and yelling at you two years out, I can’t stand this pain, and you’re going to like, what the heck is going on?

Well, it turns out there is also an association between the SNPs and the incidence of urinary dysfunction. So here’s the study. Now, this is baseline, these are AUA symptoms scores, and these are changes from baseline at each time point going from a baseline out to five years. And I have divided this data up into presenting symptoms.

So these are the patients in this line who present with severe urinary symptoms at the outset, AUA symptom score of 20 or higher. These are ones with intermediate symptoms, and these are the ones with the fewest symptoms, score of zero to seven.

So you would think, oh my goodness, if you have a patient with a AUA symptom score of 25, and you put seeds in them, he’s gone. Turns out over time, they do the best; they have the greatest decrease in symptoms by the change from baseline versus the ones who have fewer symptoms.

But if you model this now, average it all together and model it with the SNP, we found that was significant. Turns out those patients who have this mutation, it’s not really a mutation, it’s a SNP is different than a mutation, but we’re not going to go into that, they have worse urinary symptoms. The dotted line shows you a change that’s greater than five because we found if the AUA symptom score changes by five points from your baseline, that correlates with a change in quality of life. So they go higher, they stay higher longer, and they don’t return to baseline if they have that anomaly.

I’m going to switch gears now because this is like sounding so exciting. Let’s go out and file a patent and we’ll get all these SNPs, and then we’ll have a test that we can sell it. And we did that. We filed a patent. After a year, we realized this is not going to work as easily, it’s not as easy as it sounds, and so, I’ll show you some information as to why it doesn’t work so well.

There is a company that has been very aggressive called 23andMe, and they started selling their SNP test to identify patients with a whole host of diseases. And it hadn’t been on the market very long before the FDA told them to pull it. And one of the commentators said it was borderline absurd to think someone was going to get a mastectomy based on a $99 test, because that’s basically what the company was pushing. They weren’t advising for a mastectomy, but they were giving you information that would make you believe, gees, I’ve got to have my breast out because I’ve got this SNP abnormality.

A company spokeswoman said she had hoped to reach a million customers by early next year. So a million customers and $100 bucks, that’s $100 million by selling a blood test in one year. That’s pretty aggressive. So they pulled this test, and they interviewed this one woman in this article who talked about her experience in sending off her blood to these different companies that offer these tests.

And this is what she found: she sent the blood to 23andMe and they told her she had an elevated risk for psoriasis and rheumatoid arthritis. And then she sent it to another company called Genetic Testing Laboratories, and they said she had no risk for psoriasis and rheumatoid arthritis. So because of these discrepancies, one has to be extremely careful before they start going out and marketing these things because there’s too many mistakes to be made.

Scientists have identified about ten million SNPs within our three billion nucleotides. But an entire genome sequencing, which we’re not doing, we’re just taking little looks at whatever we put on the chip, you decide you want it on the chip, then that’s what you’re going to find. When you look at three billion of these nucleotides, that would cost at least $3000 per patient and it would take a long time to do. And clearly , we’re not there yet.
So the message is here is we have to be very careful about doing these type of analyses and then reporting the data. It looks really attractive in terms of the radiation, but we still have a lot of work to do. And we’ve got to make sure that we reach that threshold of at least 10-8 to achieve what we call genome-wide significance.

And so, we’ve gone further and in a collaborate effort, we formed this new group called STROGAR, and we’re actively investigating the correct patient populations; what type of radiation exposure; the phenotype is basically the output, that’s the outcome you’re looking at; what type of genomic typing strategy we should use, and quality control is very important; how should we do these associations; and what type of secondary analysis we should do. So this is ongoing work, not ready for clinical use yet.

So in conclusion, multiple SNPs can affect normal tissue response to radiation. Genome-wide studies and high-powered statistical modeling allow us to identify those SNPs that are associated with specific morbidities. Further studies with a greater number of patients will hopefully allow the creation of predictive models to screen patients for specific radiation side effects, so the idea there is, of course, one day, the patient has the blood test, he has a high risk of proctitis, he’s not getting radiation; he’s going to have some other form of therapy.

Once adequately validated, patient treatment selection may be further refined. That’s going to be our goal. Thank you.

References

Andreassen CN, Alsner J, Overgaard J, et al. TGFB1 polymorphisms are associated with risk of late normal tissue complications in the breast after radiotherapy for early breast cancer. Radiother Oncol. 2005 Apr;75(1):18-21. http://www.ncbi.nlm.nih.gov/pubmed/15878096

Andreassen CN, Overgaard J, Alsner J, et al. ATM sequence variants and risk of radiation-induced subcutaneous fibrosis after postmastectomy radiotherapy. Int J Radiat Oncol Biol Phys. 2006 Mar 1;64(3):776-83. http://www.ncbi.nlm.nih.gov/pubmed/16338099

Anscher MS, Murase T, Prescott DM, et al. Changes in plasma TGF beta levels during pulmonary radiotherapy as a predictor of the risk of developing radiation pneumonitis. Int J Radiat Oncol Biol Phys. 1994 Oct 15;30(3):671-6. http://www.ncbi.nlm.nih.gov/pubmed/7928499

Cesaretti JA, Stock RG, Atencio DP, et al. A genetically determined dose-volume histogram predicts for rectal bleeding among patients treated withprostate brachytherapy. Int J Radiat Oncol Biol Phys. 2007 Aug 1;68(5):1410-6. http://www.ncbi.nlm.nih.gov/pubmed/17490827

De Ruyck K, Van Eijkeren M, Claes K, et al. TGFbeta1 polymorphisms and late clinical radiosensitivity in patients treated for gynecologic tumors. Int J Radiat Oncol Biol Phys. 2006 Jul 15;65(4):1240-8. http://www.ncbi.nlm.nih.gov/pubmed/16798416

Hall EJ, Schiff PB, Hanks GE, et al. A preliminary report: frequency of A-T heterozygotes among prostate cancer patients with severe late responses to radiation therapy. Cancer J Sci Am. 1998 Nov-Dec;4(6):385-9. http://www.ncbi.nlm.nih.gov/pubmed/9853138

Kerns SL, Stock R, Stone N, et al. A 2-stage genome-wide association study to identify single nucleotide polymorphisms associated with development of erectile dysfunction following radiation therapy for prostate cancer. Int J Radiat Oncol Biol Phys. 2013 Jan 1;85(1):e21-8. http://www.ncbi.nlm.nih.gov/pubmed/23021708

Kerns SL, de Ruysscher D, Andreassen CN, et al. STROGAR – STrengthening the Reporting Of Genetic Association studies in Radiogenomics. Radiother Oncol. 2014 Jan;110(1):182-8. http://www.ncbi.nlm.nih.gov/pubmed/23993398

Panagiotou OA, Ioannidis JP; Genome-Wide Significance Project.
What should the genome-wide significance threshold be? Empirical replication of borderline genetic associations. Int J Epidemiol. 2012 Feb;41(1):273-86. http://www.ncbi.nlm.nih.gov/pubmed/12569017

Peters CA, Stock RG, Cesaretti JA,  et al. TGFB1 single nucleotide polymorphisms are associated with adverse quality of life in prostate cancer patientstreated with radiotherapy. Int J Radiat Oncol Biol Phys. 2008 Mar 1;70(3):752-9. http://www.ncbi.nlm.nih.gov/pubmed/12569017

Quarmby S, Fakhoury H, Levine E, et al. Association of transforming growth factor beta-1 single nucleotide polymorphisms with radiation-induced damage to normal tissues in breast cancer patients. Int J Radiat Biol. 2003 Feb;79(2):137-43. http://www.ncbi.nlm.nih.gov/pubmed/22253303

Snyder KM, Stock RG, Hong SM, et al. Defining the risk of developing grade 2 proctitis following 125I prostate brachytherapy using a rectal dose-volume histogram analysis. Int J Radiat Oncol Biol Phys. 2001 Jun 1;50(2):335-41. http://www.ncbi.nlm.nih.gov/pubmed/11380219