Kelly Musings

It is not unusual for drugs to fail Phase 3 pivotal studies after showing success in Phase 2 studies. In fact, it is the norm rather than the exception. And this reality is not limited to experimental drugs … almost every major anti-cancer drug on the market today has failed at least one pivotal study in attempts to expand its approved uses.

The reasons for these failures are legion, but essentially come down to an inability to reproduce in a large sample size the same strength of clinical benefit found in a smaller sample size.  Moving from a relatively small number of patients in a well-controlled setting of one or several clinics (a Phase 2 trial), to the less well-controlled setting of many clinics, often over a range of disparate countries, with a far greater variety of clinical variables (a Phase 3 study), places a challenge on the statistical power which causes most drugs to fail at that final hurdle.

Pivotal studies involving many hundreds or even thousands of patients are not carried out on a whim.  Large studies such as these don’t proceed unless all concerned (sponsor, investigators, ethical review committees, FDA) are satisfied that there is sufficiently robust evidence from Phase 2 studies to suggest that a Phase 3 study has the potential to deliver a positive outcome.

That was the case with OVATURE. The Phase 2 clinical evidence pointed unequivocally to a beneficial effect (including complete and partial tumour responses) in about 30% of patients with platinum-refractory ovarian cancer. No other drug had shown the promise being displayed by phenoxodiol in being able to restore the benefit of chemotherapy in end-stage ovarian cancer. That was why OVATURE received the support and encouragement of the FDA and senior oncologists worldwide.

That background makes the outcome of OVATURE so perplexing. OVATURE didn’t just fall short of meeting a required statistical target, it appeared not to deliver any benefit at all. And that looks suspiciously like a failure of trial design rather than a failure of the drug per se.

So what went wrong? As the person who coordinated the design of OVATURE, I have both a sense of responsibility for the failure and a desire to find the answers. This posting is intended to be the start of a discussion and thought process to which I hope others will contribute. No doubt the same questions are being asked within Marshall Edwards, but the Company won’t have a monopoly on insights into the problem, so a broader discussion might prove to be helpful.

I am not sure whether MEI will continue with phenoxodiol given the limited patent life remaining on the drug, plus the development of later-generation drugs with greater power and greater utility than phenoxodiol. However, phenoxodiol shares a common pharmacological profile to the larger family of isoflavonoid drugs, and for that matter, issues regarding the appropriate use of phenoxodiol are almost certainly going to be pertinent to the entire family of drugs. Hence the need for this forensic review and discussion.

The finger of blame is being pointed at the dosage formulation given that this is the one essential difference between the Phase 2 and 3 studies. In the Phase 2 study, phenoxodiol was administered intravenously. In OVATURE, this was changed to the oral dosage form. The reasons for this change in formulation have been canvassed previously …… the ability to deliver steady-state drug levels in blood, the ability to dose patients over a longer term, the avoidance of cyclodextrin-related side-effects, and greater patient convenience.

With the benefit of hindsight it is easy now to criticise such a change so late in the day, but in the context of the time, it was not seen as such a major change. By 2005, the Company was committed to seeking marketing registration for a wide range of clinical applications, including monotherapy treatment for prostate cancer, for which long-term dosing using an oral dosage form was the only practical option. The decision to commit to a single formulation across all clinical applications was only taken after we were satisfied that the oral dosage form delivered equivalent bio-availability to the intravenous dosage form.

Even in the setting of restoration of chemo-sensitivity as in OVATURE where phenoxodiol treatment only needed to be short-term, there was a sense that the long-term, steady-state blood levels possible with the oral dosage form gave the patient an additional two-pronged benefit. As a chemo-sensitiser, phenoxodiol only needed to be present contemporaneously with carboplatin dosing. The intravenous dosage form was perfectly well suited for such short-term delivery, but  the oral dosage form was perfectly capable of delivering the same outcome. But where the oral dosage form offered a clear advantage was its ability to go on being used, long after the carboplatin treatment, where there was the potential for it to deliver an ongoing monotherapy anti-cancer effect.

That, I believe, is the mistake that we made. Not that we changed dosage forms, but that we confused monotherapy and combination therapy outcomes – two entirely different pharmacological effects of phenoxodiol that look now to be incompatible.

Before coming back to this important point, we need first from a purely forensic approach to canvass all potential implications of the change from intravenous to oral dosage form. To that extent, three aspects come to mind in the first instance.

 (a)  Cyclodextrin increases tumor penetration.  The one outstanding compositional difference between the oral and intravenous dosage forms is the presence of cyclodextrin in blood in association with the intravenous dosage form.

The question then being, is the presence of cyclodextrin in blood somehow enhancing PXD uptake by tumor tissue? The answer, I believe, is, highly unlikely.

Release of drug from the large sugar molecule appears to be relatively rapid post-injection, and relatively complete, with  efficient separation of PXD from its carrier occurring within 30 seconds of injection.

Once separated, could the co-presence of cyclodextrin somehow enhance tumor penetration by PXD? This would be highly unlikely given the physico-chemical nature of the cyclodextrin molecule. A very large molecule, it doesn’t leave the bloodstream, so the notion of its presence somehow enhancing contact between PXD and its molecular target seems remote.

A second possibility concerns the fact that not all of the injected PXD leaves the carrier post-injection. Between about 5-10% of PXD remains within the carrier some hours following injection, and although the fate of this drug has not been confirmed, it seems likely to follow the same fate as the cyclodextrin molecule and be excreted via the kidneys. For the same reason that it is difficult to see any biological interaction between released PXD and cyclodextrin because of the size of the cyclodextrin molecule, it is equally difficult to see any retained PXD having any biological role within the body.

In summary, it is difficult to see the presence of cyclodextrin as contributing to any additional benefit of the intravenous dosage form.

(b)  Differing conjugation profile.     Isoflavonoids are chemically reactive compounds. That is, they are prone to attach themselves (conjugate) to a range of chemicals (eg. salts) and large molecules (eg. sugars and proteins) after entering the bloodstream. This reactivity is the result of them expressing hydroxyl (OH) groups as the following structure of phenoxodiol shows. The chemical reactivity stems from hydroxyl groups being inclined to form hydrogen bonds with other reactive chemicals and molecules.

As well as being chemically active, hydroxyl groups also are hydrophobic, meaning that compounds that express such groups are inclined to be insoluble in water.  Isoflavonoids are highly insoluble in water.  The biological outcome of forming conjugates is to convert the isoflavone into a water-soluble form that facilitates its transport in the blood, its movement out of the bloodstream into target tissues, and its ability to be excreted in urine via the kidneys.

The great bulk (>95%) of phenoxodiol in the bloodstream is conjugated either to the sugar, glucuronide, or to sulphate(SO₄) ions. With two hydroxyl sites per phenoxodiol molecule, 5 different basic conjugated forms are possible – mono-glucuronide, di-glucuronide, mono-sulfate, di-sulfate, and a blend of glucuronide on one OH group and a sulfate on the other.

Apart from making phenoxodiol water-soluble, this process of conjugation also eliminates the compound’s anti-cancer activity. That activity relies in part on the overall 3-dimensional structure of the isoflavonoid molecule (significantly altered by the addition of a conjugated chemical) and in part on the hydroxyl groups being intact. Restoration of that activity occurs outside of the bloodstream under the action of the enzymes glucuronidase and sulfatase which cut the conjugates, restoring the free phenoxodiol molecule. All human tissues express these deconjugating enzymes.

Phenoxodiol also binds very weakly to plasma proteins (mainly albumin), and it is believed that a very small proportion is present in blood in this form and reaches tissues in that form. The bond between the two compounds is readily broken and does not require enzymatic action. We found it difficult to quantify the extent of this form of conjugation, but estimated it to be in the order of between 1-5% of the total amount of drug in the bloodstream.

This conjugation story has two points of relevance when it comes to how the drug is administered. The first is that there may be a difference in the proportion of different conjugates depending on the route of administration. When given orally, the primary site of conjugation is the gut wall, with the liver then providing secondary processing.  In the case of intravenous injection, all conjugation takes place in the liver. We were not aware of any major differences in the conjugation profile between the two routes of administration, however our methods of analysis of the formed conjugates were not subtle enough to be confident that there were not minor differences, particularly in the level of free, unconjugated phenoxodiol that was present bound to plasma protein. The suspicion remained that the intravenous dosage form may have been delivering a slightly higher proportion of protein-bound drug.

Which brings us to the second point of relevance of the dosage form. The whole phenomenon of drugs such as phenoxodiol being conjugated by the body to facilitate transport, only to be released in the free, active form in the end tissues under the action of deconjugating enzymes, is a normal phenomenon pertaining to normal tissues. The big unknown is what happens in the case of cancer tissue, which is the very point in using phenoxodiol. Is cancer tissue just as capable of deconjugating phenoxodiol as normal tissue? There was then, and still is, no certain answer to that question. To the best of my knowledge, the broad family of cancer tissues have yet to be mapped for their content of glucuronidase and sulfatase as have normal tissues.  The point being that if certain cancer tissues do not contain deconjugating enzymes, then conjugated drug is going to be ineffective. In that circumstance, any anti-cancer activity would be dependent on the presence of free, protein-bound phenoxodiol, in which case the intravenous dosage form may have provided a clear advantage over the oral dosage form.

However, there was enough evidence to allay this concern when it came to the design of OVATURE. First, ovarian cancer tissue had been shown by others to contain normal levels of glucuronidase and sulfatase activity. That was a highly reassuring finding. It didn’t mean that every case of ovarian cancer would behave in that way, but it suggested that most, and probably enough for our purposes, would. Second, we had conducted a study in mice bearing xenografts of human prostate cancer where we had dosed them orally with radioactive-labelled phenoxodiol. That study showed that free phenoxodiol accumulated within the tumour tissue, indicating that conjugated phenoxodiol being borne in the bloodstream was being deconjugated by the tumor tissue and accumulating in an active form within that tissue. Third, we had experienced some dramatic anti-cancer responses in patients being dosed orally with phenoxodiol in a compassionate use study. The complete responses in a number of those patients reassured us that the oral dosage form was bio-available and perfectly capable of delivering a potent anti-cancer effect.  

(c)  Monotherapy vs chemo-sensitising effect.     With the wisdom of hindsight, I think our mistake was to confuse these two effects. In a nutshell, the way we used phenoxodiol in the Phase 2 ovarian cancer study was the appropriate way as a chemo-sensitiser, whereas in OVATURE it was administered inappropriately as if we were intending to obtain a monotherapy effect.

In phase 2, phenoxodiol was injected on two consecutive days, the rationale being that this primed the cancer cells with phenoxodiol immediately prior to the addition of cisplatin or paclitaxel. In vitro studies had shown that there was no benefit to be gained in terms of restoring chemo-sensitivity by continuing to expose cancer cells to phenoxodiol after the platinum drug was added, so we left the course of treatment at phenoxodiol (Day 1, morning), phenoxodiol (Day 2, morning), cisplatin/paclitaxel (Day 2, afternoon).

In OVATURE, phenoxodiol was administered orally on a continuous basis that delivered steady-state blood levels before, during and after the platinum challenge. At the time that OVATURE was being designed, there was the thought that this approach would deliver the same chemo-sensitising effect that we saw in the Phase 2 study by priming the cancer cells with phenoxodiol prior to the second drug challenge, but that ongoing exposure to phenoxodiol would potentially deliver a secondary benefit of a monotherapy anti-cancer effect. What the outcome of OVATURE suggests is that the two effects are mutually incompatible, and that the monotherapy effect nullified any chemo-sensitising effect.

All the data that we have accumulated to date on the pharmacology of phenoxodiol points to the fact that this drug only partially fits conventional wisdom on how to use anti-cancer drugs. The drug is conventional to the extent that it targets a product of a mutant gene (the tNOX protein) and in this respect parallels the action of Gleevec  and Herceptin. By inhibiting that gene product in a dose-response way,  the metabolic activity of the cancer cell is progressively disrupted to the extent that the cell eventually dies. This direct, dose-response effect appeared to translate into the clinic when in two small Phase 2 studies in late-stage prostate cancer and early-stage cervical cancer, we achieved a dose-response anti-cancer effect when phenoxodiol was used as a monotherapy. All that fits with the conventional wisdom that the more drug you can manage to get to the molecular target, the more likely it is to inhibit that target.

The departure from conventional wisdom comes when the drug is used to restore chemo-sensitivity. Three data sets stand out in this regard.

(a)    The first is the laboratory work done using phenoxodiol to restore sensitivity in highly resistant cancer cells to platinums, taxanes and gemcitabine. Small (sub-lethal) doses of phenoxodiol were highly effective at restoring sensitivity, so much so that the doses of the cytotoxics could be lowered to extraordinarily low levels without loss of efficacy. The cytotoxics retained efficacy down to almost homeopathic levels (10^-6). There was evidence also that with both phenoxodiol and the cytotoxic drug, the smaller the dose of either agent,  the more effective the restoration of chemo-sensitivity. That is an entirely unconventional notion.

(b)   The second piece of data concerns in vitro studies showing that restoration of chemo-sensitivity by phenoxodiol to taxanes is a ‘bystander’ effect, meaning that it can be transferred to other cancer cells that are phenoxodiol-naive.

(c)    The third piece of data is the clinical observation that restoration of sensitivity of ovarian cancer to taxanes occurred following exposure to phenoxodiol, but only in the absence of concurrent phenoxodiol (up to 3 weeks following phenoxodiol therapy) at the time of challenge with the taxane.

It is anyone’s guess what is happening here, but after 15 years of dealing with this drug and experiencing all its various twists and turns, here is what I think.

By the very definition of using phenoxodiol to restore chemo-sensitivity, you don’t need to or even want to kill the cell. You are simply looking to modify the cell sufficiently to render it susceptible to the killing effect of a cytotoxic drug.  That means that it not only isn’t necessary to saturate the molecular target, it looks like it is vital that you don’t.

We don’t know the nature of the interaction between phenoxodiol and its molecular target. We know that it inhibits its ability to moderate the NOX redox pump at the cell surface, and probably also in mitochondria. The ability of the cell to transfer hydrogen ions across both the plasma and mitochondrial membranes is lost, leading both to inhibition of the sphingomyelin pathway, a build-up of NADH within the cytoplasm, and a loss of ATP production.

Exactly how phenoxodiol interacts with the tNOX assembly is not clear. It could be as simple as blocking a critical receptor site (as do Gleevec and Herceptin).  Or it could be that it is rendering the tNOX regulating protein inoperable through a structural change. My theory is that phenoxodiol induces structural changes in the tNOX regulating protein. At high drug doses, this effect is lethal, with tNOX function being completely abrogated, leading to irreversible damage to the cancer cell.

From a chemo-sensitising perspective, however, that structural change is noon-lethal on its own. The ‘bystander’ effect that we have observed, suggests that the structural change to the tNOX regulating protein is both permanent and transferable, in much the same way that prions are formed, by-passing the genome and endowing a learned effect on the tNOX proteins of other cancer cells not previously exposed to phenoxodiol. That, if true, takes us into an entirely new field anti-cancer drug pharmacology. OVATURE was designed according to conventional wisdom of how anti-cancer drugs work….the lesson to be learnt from that is that new rules need to be written for this class of drug when being used as a chemo-sensitiser.

Whatever rules need to be re-written for phenoxodiol in this regard, they almost certainly will apply also to triphendiol and NV-128. And I strongly suspect that the rules will need to be fine-tuned for each class of cytotoxic to be employed (viz. platinums, taxanes and gemcitabine).

As disappointing as the outcome of OVATURE was for everyone, this is far from being the death-knell for the technology. Rather, it serves as a lesson that we need to learn from. Once learned, the technology still holds the extraordinary promise that many of us know lies yet to be fulfilled.

This entry was posted Sunday, June 27th, 2010 at 5:39 pm and is filed under Uncategorized. You can leave a response, or trackback from your own site.