Path to Durable Complete Remission in Prostate Cancer: What Do We Know and What Do We Need to Learn?

Path to Durable Complete Remission in Prostate Cancer: What Do We Know and What Do We Need to Learn?

The topic I’m talking about complete remission is something that I talked to David over the years and it’s helped shaped my views. The other person who has helped shaped my views is Phil Kantoff who has talked to me over the years because we shared a patient with widespread metastatic disease who went into a durable complete remission in 1990 and is still in complete remission.

I have no financial conflicts of interest, but I do have a bias because as David mentioned, I had metastatic prostate cancer in 1999, and have been in a durable complete remission ever since so I have a bias about its importance.

I’m going to step back and leave prostate cancer oncology and talk about medical oncology and that is the lessons from the last 50 years of cancer treatment is that complete remissions matter. Of most cancers, partial remissions are good for months to a year impact on survival with a few exceptions. Complete remissions are required for long term disease free survival. And with only a few exceptions like choriocarcinoma, gestational choriocarcinoma, complete remissions require well designed combinations. Also interestingly, complete remissions are best if they are obtained rapidly usually within three to six months.

How can we induce complete remissions? Well the traditional approach with combination therapy that was responsible for curing diseases like Hodgkin’s disease involved combining agents with different mechanisms of action. And actually this idea first developed post World War 2 in tuberculosis and then moved into oncology. The fundamental principle is that probabilities multiple so if you have a drug that has a 1 in 100 change of failing, if you combine two of them, it’s 1/100 x 1/100 and that’s 1/10,000. Three drugs, it becomes 1/1,000,000 so you see the major advantage. Now most cancer drugs, mutation resistance occurs in 1/100,000 cells so three drugs, it would be 1/10(18). So that’s the fundamental rationale for combining drugs.

Another key thing is toxicities should not overlap. Independent mechanisms of action and overlapping side effects. These principles are applied where ever resistance is a problem. The cancers we cure today like Hodgkin’s, childhood leukemia, cancer of the testes are all dependent on combinations. Modern treatment of AIDS involves four drugs for the same reason. Tuberculosis, same thing, four drugs. Actually this theory of combination therapy has seen greater exploitation and elaboration in infectious disease than in medical oncology. Unfortunately, despite the early successes of this approach it has failed in prostate cancer and many other cancers of adult life.

So the question is, why has this failed and what can we do about it? Well, first complete remissions are not common. We now have some examples of this in the PREVAIL trial, just under 20% of the people on the Xtandi arm went into complete remission. Well, this excited me because if you look at medical oncology in general, this is essentially a third line agent and getting a 20% complete remission in any cancers of adult life with any oral agent would be uncommon. So it’s a nice observation, but it’s pretty much isolated.

Another example that I’d like to talk about is the CHAARTED trial which we all know about. It showed dramatic benefit. In this slide was sent to me by Phil Kantoff because of the discussions we had, and he pointed out the big jump in PSA undetectable at 6 and 12 months in patients on the combined treatment. This is just two agents, and this is what I would call a classic example of synergy and best – – in the clinic. Unfortunately the trial looked only at PSA biochemical complete remissions and not radiologic complete remissions and it would really be nice to know if there were radiologic remissions.

So we have some agents that cause complete remissions but not a lot. And the remissions are infrequent in number, under 30%.

So we have a difficult problem, and one of the best ways of thinking about a problem if you can’t figure it out is to take a different perspective.

And in talking to one of my patients who is a farmer, he said well why don’t you think about a cornfield as an analogy for this. So this is a picture of a cornfield, best of the corn rows and it looks like the cornfield is completely without any order, around the corner, and suddenly what appears to be a disordered growing of plants becomes highly ordered. I think this was important learning heuristic when you’re approaching problems like we do with prostate cancer.

So the rest of this talk I’m going to talk about how to change the perspective, how to really look at this. And fortunately, the molecular revolution has given us a new insight into what’s going on with prostate cancer.

The first one I’d like to talk about are mutational events. Lung cancer, adenocarcinoma of the lung was one of the first cancers revolutionized by this approach. In 2003 adenocarcinoma of the lung was one disease. In 2012, a series of mutational events were described, drugs were developed for those mutational events, and now adenocarcinoma of the lung is broken up into many different therapeutic subtypes, each with their own treatment philosophy.

So the process is to identify key mutational events, develop drugs that target these mutational events, use well designed phase 3 trials to prove effectiveness. So the result is that the adenocarcinoma of the lung is a much different disease than it was before.

What about prostate cancer? Well, there are many molecular subtypes of prostate cancer that have been identified. I think the most spectacular example was published in the New England Journal of Medicine in October 28 looking at DNA repair mutations in prostate cancer. Now there are many different DNA repair mutations that play a role in cancer risk and development. BRCA1 and BRCA2 are the best known because of their role in breast and ovarian cancer. BRCA2 is also associated with prostate cancer and the prostate cancer that develops in these patients is much more aggressive than usual. In Tour de France, they talk about a climb that’s beyond category and these cancers are beyond category in terms of their aggression.

When BRCA1 and BRCA2 are lost, the cells have a backup called PARP that provides DNA repair, less effective, but sufficient to allow the cell survive. And when PARP is inhibited, BRCA1 and BRCA2 cancers die. Olaparib is a PARP inhibitor that’s approved for ovarian cancer. It turns out that PARP is a backup not only for BRCA1 and BRCA2 mutations, but for a wide range of DNA mutations found in cancer, PARP acts as a backup.

This led to the New England Journal of Medicine study. This is from Johann de Bono’s group in the Royal Marsden. They had 49 men with prostate cancer who had failed docetaxel and either abiraterone or enzalutamide. These were treated with olaparib, 400 mg twice a day. And the overall response rate was 33%. By itself alone would be a rather stunning outcome. For those who have the usual apparent, it’s really an easy drug to use, and well tolerated.

What’s interesting is that 16 of the patients had DNA repair defects, and the response rate in those patients was 88%. Let that sink in, these are post-Taxotere widely metastatic higher aggressive prostate cancers. If they had DNA mutations, 88% response rate.

The two most commons DNA repair defects were BRCA2 where seven out of seven patients responded, 100% response rate. The other is the ataxia telangiectasia mutation where four out of five responded.

I actually got wind of these results, in the spring at ASCO and so July 17th I started my first patient on this drug and he was in hospice care, BRCA-2 mutations, extensive long and liver involvement, peritoneal seeding with ascites equivalent to a five month pregnancy, and diffused evasion of his pancreas which is unusual in this disease. He was in hospice care and I would guess he probably had about 7 to 10 days to live. He’s doing fine now, out of hospice, thinking about going back to work. So this is very impressive.

So in the New England Journal administrative paper, they had a 33% incidence of DNA repair mutations in advanced disease. Is this an accident of that study or are they really that common? Well, there are two datasets we can go to. One is a Cell paper which is a wonderful compendium of what we know about DNA changes in prostate cancer. And in that trial BRCA1, and BRCA2, and ATM were mutated in 19.3% of the patients. With Caris, we’ve analyzed 118 patients. We presented this at ASCO-GU. And we had 11% with BRCA2 and 11% with ATM mutations. So 20% to 30% of post Taxotere patients may be candidates for PARP therapy, an oral drug associated with close to 90% response rate. So it’s hard for me to believe this won’t be an important advance in the management of those patients. In addition to being very well tolerated, this drug does not depend on testosterone presence or absence to work.

So we want to talk about drugs with independent mechanisms of action. Here’s a drug with a high single agent efficacy and doesn’t overlap with the side effects or mechanism of action of non-hormonal therapy drugs, so it forms an interesting bases for developing combinations for these agents.

Clinically these patients, the BRCA2 patients will often also respond to cisplatinum, and there is a DNA repair gene that predicts platinum-responsiveness, ERCC1. And in our dataset, 75% of these DNA mutant patients lack ERCC1 and would be protected to be very sensitive to cisplatinum. So suddenly a very bad prognostic group of patients has a range of potential successive therapies.

Now another way of looking at this is to look at the morphology and molecular events. This goes from looking at a single mutation to pathology, and here we delve into issues that have been mentioned earlier. The presence of neuroendocrine differentiation and epithelial and mesenchymal transition.

This is a trial from a paper by Beltran,  a young investigator who’s been doing great work in this field. I’d like to walk you through this. Adenocarcinomas can go to a neuroendocrine cell that is not dividing, growing or moving. And although she shows this as a one way picture, they can move back and forth, so these neuroendocrine cells can act as a sanctuary site. They don’t divide, so cytotoxic agents don’t attack them. They’re radiation resistant, RTOG is interesting, dataset on how this converge radiation resistance. But when–and testosterone removal can trigger this, you add testosterone back, they become adenocarcinoma cells. So this is a form that the cells can persist in through hormonal therapy, radiation, and other treatments.

The other neuroendocrine cells are the small cell, adenocarcinoma cells and these are quite different. These are metabolically active rapidly growing and spreading cells. And the origins of these are controversial, something there’s a direct move from adenocarcinoma to small cell. I think the more likely move is from these transdifferentiated cells to the small cell because it’s a smaller molecular change that needs to take place. To go from this to this involves a huge range of genetic alterations that are unlikely probabilistically. And so these are highly aggressive and can kill rapidly. These can act as a sanctuary site.

Well this story has been around for a long time. The endocrine conversion was first described in the late 1980s, but there are two presentations at ASCO-GU in 2015 that I think really put this in perspective. The first one was by Eric Small and the West Coast Dream Team, and they had patients progressing on abiraterone and enzalutamide and they biopsied tumor sites.

Now you might think gosh, hasn’t this been done before? But it’s been very hard to biopsy bone metastasis and get enough material from molecular characterization, and the success of the West Coast Dream Team rest in part on solving the problem of getting enough cancer out of a bone biopsy for molecular characterization. Thirteen percent had pure neuroendocrine biology; 26% were a mixture of adenocarcinoma and neuroendocrine cells; 35% were pure adenocarcinoma; and 26% were a new histology called intermittent atypical carcinoma that molecularly looked like a transition between adenocarcinoma and the neuroendocrine cancers; so a total of 65% of the biopsies showed some element of neuroendocrine biology.

So from this and the rest of the information we know, it is apparent that neuroendocrine differentiation plays a major role in limiting the success of hormonal therapy, and as we’ve gotten more successful in attacking testosterone, this has become more of a clinical problem. And effectively sets a limit of what we can do with standard hormonal therapy and puts a limit on drugs like enzalutamide and abiraterone.

What are the characteristics of the small cell carcinoma? Again, a very nice set of work. Gene methylation was extensive in these cells. I think it’s interesting that gene methylation turns out to be important in newly diagnosed patients as a field effect, at the very beginning of the diseases of prognostic factor, and here it raises its ugly head as a major change in the formation of small cell carcinoma of the prostate.

RB was lost in 70% and TP53 in 66.6% of the biopsies. But loss of RB1 and TP53 have been known for quite a while to be a very poor prognostic sign in prostate cancer. In addition to showing genetic evidence of neuroendocrine differentiation, these cells also showed the changes that are associated with epithelial to mesenchymal transition.

Now the first time I heard that term, I scratch my head and said, what the hell is that? It turns out to be something that many carcinomas do and so you have here the primary tumor, and these are adenocarcinoma cells. They lack motility, they adhere to one another, to E-cadherin, so they can’t move. The adenocarcinoma cells cold undergo transition to mesenchymal cells, and mesenchymal cells lose their cell to cell adhesion and gain motility. And this is the form that invasions of blood stream and spreads. Well, when the cancer reaches its metastatic site, it escapes, and there’s the tissue, and does the reverse, it goes from mesenchymal to epithelial. And this isn’t true for just prostate cancer, but for a wide range of other cancers, for example, a big event in cancer of the pancreas.

If neuroendocrine differentiation limits the effectiveness of hormonal therapy, we really only have two options. We either have to prevent neuroendocrine differentiation, or we have to selectively kill the neuroendocrine cells.

Well, if you just look at the gene changes in the neuroendocrine cells, there are multiple gene changes. And it’s very much looking at the cornfield or best of the corn rows. How do you take that complexity and reduce it to a simple enough view of what’s going on that you can attack it therapeutically. And I’ve always been a fan of Occam’s Razor and then among competing hypotheses, the one’s with the fewest assumptions should be selected.

And so I pick up this, it turns out that cyclic AMP, a common second messenger from many metabolic events can trigger neuroendocrine differentiation. My team at NCI first showed this in 1992, and then in 1994 when I went to University of Virginia, I was on the thesis committee for Mike Cox and he picked this as a subject. And he went on to show that in fact cyclic AMP activation of protein kinase A was all that you needed to to cause neuroendocrine differentiation, and they could turn it on and off the neuroendocrine state by turning on and off protein kinase A. They then went and took cells constitutively active for protein kinase A and mixed them with adenocarcinoma cells and put them in a mouse. And that mixed population did much better than adenocarcinoma alone showing that it played a role in proliferation. And then went on to show that also it cause hormone resistance. Protein kinase A, cyclic AMP, sufficient to cause neuroendocrine differentiation.

Well, there are many other signals that can cause neuroendocrine differentiation of prostate cancer, and this is a list, a partial list of the various things that can cause adenocarcinoma of the prostate to become neuroendocrine. And you notice androgen withdrawal is on the bottom. Each one of these increases cyclic AMP intracellularly so the hypothesis that cyclic AMP is driving it fits all the known data.

How is cyclic AMP generated? I don’t know about you but often when I see a slide like this, I just groan. But again, our goal is simplicity here and focus.

Cyclic AMP is generated with the heterotrimeric G cell surface receptors and these are a bunch of the different ligands that are evolved. Things as simple as epinephrine, gaba, light, odor, colors. And the receptor is linked up with three proteins of  called heterotrimeric and there are a variety of eight subunits. The subunit involve the neuroendocrine transformation as A3, and it activates adenyl cyclase to generate cyclic AMP and then activates protein kinase A. Cyclic AMP is generated by adenyl cyclase and destroyed by a phosphodiesterase. Well, in the transition from hormones, sensitive to hormone resistant prostate cancers, adenyl cyclase goes up, protein kinase A subunits change in the prostate cancer, those with greater metabolic activity, and protein kinase A activity goes up in all these different model systems.

How does an elevation of acyclic AMP and activation of protein kinase A cause neuroendocrine differentiation? There’s a transcription factor called SNAIL. It’s a stick finger transcription suppressor, and cyclic… the protein kinase say phosphorylation of serine 11 and 92 and this is sufficient to cause neuroendocrine differentiation. Also the same event is sufficient to cause epithelial to neuroendocrine differentiation. So this makes a very nice clean package, the same molecular events that triggered neuroendocrine differentiation particularly the epithelial to mesenchymal differentiation. So the model then becomes very simple, elevated cyclic AMP, activation of protein kinase A, activation of snow by phosphorylation, and resulting in neuroendocrine differentiation and epithelia to mesenchymal transition.

Below I’ve given you references to several of the key papers that I can’t discuss the role of SNAIL in this process. The attraction of this is we have a single target, to select a drug for that would be specific for two of our major problems limiting therapy.

The question is, how do you target this? Well the challenge in cancer drug development is you don’t want to kill the patient at the same time you kill the cancer. And so you’ll have to work to get tissue, organ specific or cancer specific biology, and so this is a review I wrote where I went through what we knew about cyclic AMP dependent signaling in prostate cancer and one of the potential attack points. And the conclusion of that is the most attractive site is adenyl cyclase. So when you’re dealing with a signaling system that is important for every tissue in your body, and most normal physiologic functions, you have to find a way of limiting its effectiveness. And the classic model of this is epinephrine, the epinephrine, norepinephrine receptor. And of course the solution there was to discover the multiple epinephrine receptors, alpha, beta, and all the different subunits. And then the drug industry identified targets for each of the sub receptors. What you want is heterogeneity. It turns out that adenyl cyclase exist in 10 different isoforms each with limited tissue distribution.

One, adenyl cyclase 10 is limited to the prostate and to sperm. And the drugs have been developed for that, and it is cytotoxic to prostate cancer cells. Of course this is in early development. Unfortunately, adenyl cyclase 10 doesn’t activate protein kinase A. It uses a different system. Forms 1 through 9 activate protein kinase A. So the next step is to find out which isoforms are classic for prostate cancer.

So in stepping back now the big picture, with complete remissions of prostate cancer, how do we get from where we are to the world I’m trying to paint? Well, one of the major problems that clinical trials do not often capture the frequency of complete remissions, and rarely the durability of those remissions. Anyone who’s been in a large prostate clinical will have some patients who enter durable complete remissions so we know that can happen but we don’t know how to study them, in part because the clinical trials don’t capture that information. And it is not an accident, the FDA doesn’t require it. It’s expensive and very hard work in a randomized control trial to do a quality assessment of complete remissions.

I think the second conclusion is this DNA repair mutation story is just spectacular, and may well revolutionize a group of hormone refractory prostate cancer patients.

The final conclusion is that the neuroendocrine differentiation and epithelial to mesenchymal transition represents a key barrier to complete remission and likely the durability of our remissions, and we have to find some way of dealing with that.

AUDIENCE: Guidelines say that if you are a Gleason 7 or above, at the breast, ovarian, pancreatic or prostate cancer, you should be tested for BRCA. Now upfront, right now, because it’s unclear what that means in your prognosis for prostate cancer, the real benefit of that testing is for your sisters and daughters and etc., etc. Just like if you have breast cancer, you get tested for your family, etc. But as Dr. Myers has eloquently reviewed that olaparib story and there are lots of other PARP inhibitors coming, the best predictor of response to that class of drugs looks like it’s BRCA2 mutation. So it may be indirectly or after you’ve been diagnosed with prostate, knowing your BRCA status, going down the line can help tailor your therapy potentially.[su_spacer]

MYERS: Well, the thing to realize is the frequency of these mutations and the biopsy specimens far exceed the germline mutation frequency. So these are emerging then the course of the disease and ATM is also almost as frequent as BRCA2. The ATM DNA repair pathway works in series with CHEK2 and so you really could add the two together. If you add ATM and CHEK2 mutations, they’re actually more common than the BRCA2. The reason I’ve liked working with Caris is they can on a single tumor specimen do an entire DNA repair spectrum. And the cost is low enough that we’ve been able to get insurance coverage for most of our patients.[su_spacer]

CRAWFORD: So the germline mutations if I remember like 1% or less, is that right? So this is like your 18 to 20?[su_spacer]

MYERS: Yes. I would also say these patients are clinically different. It’s the most aggressive of the hormone refractory patients. I mean how many cases of pancreatic invasion have you seen? These patients become hormone resistant in six to seven months. Lung and liver involvement, brain, other tissues, extensive, so right now in the clinic, any time a patient comes in looking like that, I suspect a somatic or germline mutation and we do the testing.[su_spacer]

AUDIENCE: Snuffy, do you think that, especially for the BRCA1 and 2 story, that we give chemotherapy, right, and that what’s left is an angry prostate cancer which may be more amenable to PARP inhibitors than trying to move it up in this space.[su_spacer]

MYERS: Yes, I think there’s strong selector pressure. We know, the lethal phenotype is loss of P10, RB1 and P53, and DNA mutation frequency will go through the roof in that subset.[su_spacer]

CRAWFORD: You mentioned that one of the things that they merge was the gene methylation?[su_spacer]

MYERS: Yes.[su_spacer]

CRAWFORD: Is that something you could reverse? I know some methylation you can or is that some… possible target for therapy?[su_spacer]

MYERS: Well actually the original version of this talk had a section on that. But I couldn’t fit it in the timeline. Actually going back to 1990s we’ve done a lot of work on gene methylation, and acetylation, and there are candidate agents, nothing has really emerged that can be taken to the clinic. 5-azacytidine and 5-deoxy-azacytidine, two leukemic drugs reverse DNA methylation. So in 1992 or 1993 we did a clinical trial of 5-azacytidine in hormone refractory prostate cancer but we used the leukemic dose and it’s myelosuppressive and quite toxic. Since then, they’ve shown that low sub-toxic infusions of 5-azacytidine or 5-deoxycytidine can reverse DNA methylation. This isn’t done with other cancers, but someone needs to now go ahead and do the clinical trial. And if I had to pick the drug I would pick 5-aza-deoxycytidine, 96 hour infusions, sub-lethal dose and see if you can reverse DNA methylation of prostate cancer. That trial hasn’t been done. Histone acetylation, there’s a wide range of drugs that have been developed for that SAHA is one. We worked in phenylacetate and phenylbutyrate and they are histone deacetylase agents. And they had some clinical activity but we couldn’t figure out how to make it worth doing further work on it.[su_spacer]

Interestingly, DNA methylation is affected by diet, in the lab and in the clinic, soy isoflavones can affect DNA methylation. Resveratrol and a related compound, pterostilbene can also reverse DNA methylation by altering a micro RNA. But with DNA methylation we have excellent evidence of its importance in the biology of the cancer, but not something that we can take to the clinic yet.