Olaparib

In Part 1 of this series, we began a discussion of a new, disruptive strategy for clinical trials of oncology drugs, which had been outlined in a Perspective by Drs. Johann S. de Bono and Alan Ashworth, and published in the 30 September 2010 issue of Nature.

This strategy, which these authors call the personalized medicine hypothesis-testing strategy, is aimed at testing targeted drugs that have been developed via biology-driven drug discovery. Such a strategy begins with a strong biological hypothesis that a particular altered molecular target is critical for the malignant phenotype of a particular cancer. Based on this hypothesis, drug discovery researchers develop both targeted drugs that are specific for these altered targets, and biomarkers that can be used to determine which patients have tumors that express the target, and thus are most likely to benefit from treatment with the drug.

Following preclinical studies, clinical researchers test the drug in patients whose tumors express the target, aiming for proof of mechanism and proof of concept in early clinical trials. This involves the use of rapid dose escalation and adaptive trial design. Following these early trials, the researchers go on to conduct Phase 3 clinical trials, aiming at registration. This strategy is designed to reduce clinical attrition and the time and cost of clinical trials, and to develop superior, targeted drugs that provide greater patient benefit (in terms of progression-free survival) than the typical new oncology drugs that reach the market.

In the de Bono and Ashworth article, the authors provide several examples of successful hypothesis-testing clinical trials using this strategy. In this blog post, we discuss three of these examples, one of which is a “classic” that should be familiar to most of you, another which we have discussed in previous articles on this blog, and a third example that is based on Drs. de Bono and Ashworth’s own research.

Imatinib (Novartis’ Gleevec/Glivec)

The “classic” example of the use of a personalized medicine hypothesis-testing strategy is the development of imatinib (Novartis’ Gleevec/Glivec).  This drug was originally designed as a specific inhibitor of the ABL tyrosine kinase, which is stuck in the activated conformation in the BCR-ABL fusion protein. BCR-ABL is the “driver” mutation in Philadelphia chromosome-positive chronic myeloid leukemia (CML). Imatinib was also found to be specific for two other tyrosine kinases, c-Kit and the platelet-derived growth factor receptor (PDGFR); these findings have led to the use of imatinib to treat other cancers, especially gastrointestinal stromal tumors (GIST). We discussed the role of Dr. Brian Druker (Oregon Health Sciences University in Portland) and Nicholas B. Lydon (then at Novartis) in the development of imatinib in an earlier blog post.

The 2001 published Phase 1 clinical trial of imatinib in CML led by Drs. Druker and Lydon, and clinician Charles L Sawyers, M.D. (Memorial Sloan-Kettering Cancer Center/Howard Hughes Medical Institute) is what Drs. de Bono and Ashworth called “a landmark paper” in the use of a personalized medicine hypothesis-testing strategy to demonstrate the efficacy and safety of a targeted oncology drug. The development of imatinib for CML was made possible by basic research that showed that the BCR-ABL fusion protein (which is generated as the result of the translocation that produces the Philadelphia (Ph) chromosome, the characteristic genetic abnormality of CML) alone was sufficient to cause CML, and that the tyrosine kinase activity of the ABL moiety of the protein was required for its oncogenic activity. Researchers then discovered a compound, imatinib, that was highly specific for BCR-ABL, c-kit, and PDGFR.

The Phase I clinical trial (which took place in 1999) was a dose-escalation trial of imatinib in 83 patients with chronic-phase CML in whom treatment with interferon-alpha had failed. The primary endpoint of the trial was the safety and tolerability of the drug; efficacy was a secondary endpoint. Imatinib was found to be well-tolerated, and a maximum tolerated dose was not identified in this trial. Complete hematological responses (defined by reductions in the white-cell and platelet counts) were seen in 53 of 54 patients who received 300 mg per day or more of imatinib; these responses typically occurred in the first four weeks after initiating treatment. Cytogenetic responses were defined by the percentage of blood cells in metaphase that were positive for the Ph chromosome, ranging from major responses (zero to 35% of Ph chromosome-positive cells) to minor responses (36-65% positive) to no response (over 65% positive). Of the 54 patients treated with doses of 300 mg or more, 29 had cytogentic responses, including 17 with major responses; seven of these patients had complete cytogenetic remissions (durable zero percent Ph chromosome positive).

Blood samples were taken to determine whether BCR-ABL tyrosine kinase activity had been inhibited by in vivo treatment with imatinib. The researchers observed dose-dependent inhibition of BCR-ABL tyrosine kinase activity. This constituted proof of mechanism of the drug, while the antileukemic activity of imatinib in the trial constituted proof-of-concept.

The researchers then conducted Phase 2 clinical trials, which confirmed and extended the results seen in Phase 1. The FDA approved imatinib in May 2001, less than three years after initiation of clinical trials. This rapid approval was made possible by the FDA granting imatinib a Fast Track designation and Accelerated Approval, which allowed approval of the drug based on Phase 2 trials using surrogate markers (in this case, cytogenetic responses).

As imatinib gained approval as frontline therapy for treatment of Ph chromosome-positive CML, resistance to imatinib became an important issue. Researchers found that this resistance was usually due to mutations in BCR-ABL that interfere with imatinib binding. Two companies therefore designed inhibitors that can bind to and inhibit these resistant BCR-ABL proteins and thus successfully treat imatinib-resistant CML–dasatinib (Bristol-Myers Squibb’s Sprycel) and nilotinib (Novartis’ Tasigna). This is an example of the use of reiterative translational studies to determine mechanisms of drug resistance, and the design of second-generation drugs to combat this resistance. This type of follow-up strategy was discussed in the de Bono and Ashworth article and in our previous blog post.

Only a few years ago, many industry commentators were of the opinion that the development of imatinib to treat CML was a unique case, and development of other personalized biology-driven drug discovery-based cancer medicines would not be successful. However, the examples discussed in the de Bono and Ashworth article (and elsewhere) show that that is not true.

Roche/Plexxikon’s PLX4032

The second example of successful use of the hypothesis-testing clinical trial strategy is the development of Roche/Plexxikon’s PLX4032 for metastatic melanoma. This compound is exquisitely specific for B-Raf carrying the V600E mutation B-Raf(V600E). This is the most common somatic mutation found in human melanomas, and is a “driver mutation” that is particularly critical for the malignant phenotype of human metastatic melanomas that carry the mutation.

We have discussed PLX4032 in three articles on this blog in 2010, published on March 2, March 10, and August 27.

As in the case of imatinib, researchers achieved proof-of-mechanism and proof-of-concept for PLX4032 in a dose-escalation Phase 1 trial in patients who were preselected for carriers of the B-Raf(V600E) mutation. The Phase 1 trial took place in 2008/2009. This was followed by an extension phase in which patients were given the maximum tolerated dose of the drug. Patients showed an 81% response rate (i.e, a partial or a complete response). The estimated median progression-free survival among all patients was over 7 months, as compared to less than 2 months in large numbers of advanced melanoma patients as determined by historical analysis. Oncologists had never seen such a dramatic response in treatment of metastatic melanoma.

PLX4032 is on an accelerated path to potential registration, and parallel Phase 2 and Phase 3 clinical trials are in progress in previously treated and previously untreated patients, respectively, all who have metastatic melanoma carrying the B-Raf(V600E) mutation.

Despite the dramatic regressions and increased survival seen in the Phase 1 trials, all the patients apparently eventually suffered relapses. As stated in the article on PLX4032 in the 30 September 2010 issue of Nature, researchers are therefore doing reiterative translational studies to determine the mechanisms of resistance to PLX4032 in cases of tumor regrowth after treatment with the drug. Proposed strategies include the development of combination therapies that include PLX4032 and other targeted drugs, immunotherapeutic agents, or chemotherapy. Given the promising efficacy and safety profile of PLX4032, researchers believe that the drug has the potential to enable the development of such combination therapies.

In conjunction with the early clinical trials of PLX4032, researchers developed a real-time polymerase chain reaction (PCR) assay to assess B-Raf(V600E) mutation status. The assay has the potential to be used as a companion diagnostic in treatment with PLX4032.  As stated in the 30 September article, researchers are assessing the reliability of the PCR assay In the ongoing concurrent Phase 2 and Phase 3 clinical trials of PLX4032.

A synthetic lethal therapeutic strategy using KuDOS/AstraZeneca’s olaparib

The third example of successful use of the hypothesis-testing clinical trial strategy is taken from Drs. de Bono and Ashworth’s own work. The therapeutic strategy in this example is fundamentally different from the cases of imatinib and PLX4032, both of which are exquisitely targeted drugs that inhibit specific mutated versions of oncogenes. Instead, this example involves the use of synthetic lethality in the design of an anticancer therapeutic strategy. Based on classic studies in yeast and Drosophila, synthetic lethality is defined as a situation in which mutation in either of two genes individually has no effect, but simultaneous mutation in both genes is lethal. In cancer, if one gene in a synthetically lethal pair is defective (and especially if this defect is involved in the malignant phenotype) targeting the other gene with a drug should be selectively lethal to the tumor cells but not to normal cells. If this works, it should result in a large therapeutic window for treatment with the drug.

Women with a germline mutation in one BRCA1 or BRCA2 allele have a high risk of developing breast and ovarian cancer; BRCA1 or BRCA2 carrier status in men also carries an increased risk of developing prostate cancer. Via the process of loss of heterozygosity, cells of carriers of loss-of-function mutations in BRCA1 or BRCA2 can lose the wild-type allele, resulting in cells that lack BRCA1 or BRCA2 function. The products of the two BRCA genes are both involved in the pathway for DNA repair via homologous recombination. Loss of a functional homologous recombination pathway results in the development of genomic instability that can lead to carcinogenesis. Moreover, since BRCA-negative tumor cells cannot repair their DNA via homologous recombination, they are dependent on an alternative pathway of DNA repair, which involves the enzyme Poly(ADP) ribose polymerase (PARP). Since the average cell must repair its DNA thousands of times a day, researchers hypothesized that BRCA-negative tumor cells should be uniquely vulnerable to drugs that inhibit PARP. In contrast, normal cells are able to utilize the homologous recombination pathway, and should not be affected by PARP inhibitors.

Alan Ashworth and his colleagues developed and published this synthetic lethality strategy for therapy of BRCA-negative breast cancer in 2005. They showed that cells deficient in BRCA1 or BRCA2 were about 1,000-fold more sensitive to a class of PARP inhibitors developed by AstraZeneca (AZ) subsidiary KoDOS Pharmaceuticals (Cambridge, MA) than cells with BRCA1 and BRCA2 function. Treatment of BRCA-deficient cells with the PARP inhibitors resulted in chromosomal instability and cell cycle arrest, followed by apoptosis. The efficacy and specificity of the PARP inhibitors for BRCA-deficient cells also carried over to in vivo studies in mouse models. These cell culture and animal studies constituted the generation of a strong hypothesis that this synthetic lethal therapeutic strategy would be useful in developing antitumor treatments for patients with BRCA-negative breast cancer.

In 2006 and 2007, Drs. Ashworth, de Bono, and their colleagues (including researchers from KuDOS and AZ) conducted a Phase 1, hypothesis-testing clinical trial of KuDOS/AZ’s potent, orally-active PARP inhibitor olaparib (AZD-2281; formerly known as KU-0059436). The study enrolled a total of 60 patients with a variety of types of solid tumors, including 22 who were confirmed BRCA1 or BRCA2 mutation carriers and one patient with a strong family history of BRCA-associated cancer but who declined mutation testing. The study was published in July 2009 in the New England Journal of Medicine. The trial was a dose-escalation study–the dose was increased from 10 mg daily for two of every three weeks to 600 mg twice daily. A reversible dose-limiting toxicity was seen in one of eight patients receiving 400 mg twice daily, and in two of five patients who received 400 mg twice daily. Based on these results, the researchers established 400 mg twice daily as the maximum tolerated dose. They then enrolled a new cohort of carriers of a BRCA1 or BRCA2 mutation; these patients received a dose of 200 mg twice daily.

As a Phase 1 trial, the primary objectives were to determine safety, adverse effects, the dose-limiting toxicity and maximum tolerated dose, and the pharmacokinetic and pharmacodynamic profiles. Once these were established, the aim was to test the hypothesis that patients’ BRCA1 or BRCA2 mutation-associated cancers would show an objective antitumor response to olaparib as a single agent. In terms of safety, adverse effects were generally mild. There were two patients deaths due to infectious disease that were deemed not to be drug related. There was also no difference in adverse effect profiles between known BRCA1 and BRCA2 mutation carriers and other patients.

The researchers established three types of biomarkers. The predictive biomarker was the presence of BRCA1 or BRCA2 loss-of-function mutations, as determined by standard sequencing methods in patients with a family history of BRCA-associated cancers. The pharmacodynamic biomarker was the inhibition of PARP enzymatic activity in peripheral blood mononuclear cells and in tumor biopsies taken before and after olaparib treatment, and the formation of double-strand DNA breaks in hair follicle tissue. The intermediate endpoint biomarker consisted of radiological determination of tumor shrinkage and biochemical tests for serum tumor markers.

Using the pharmacodynamic biomarker, the researchers showed that inhibition of PARP was over 90% in peripheral mononuclear cells in patients treated with 60 mg or more of olaparib twice daily. Determination of PARP activity in tumor biopsies before and after 8 days of treatment showed that drug treatment inhibited PARP in tumor tissue. Pharmacodynamic studies in samples of plucked eyebrow hair follicles showed that induction of formation of double-strand breaks occurred within 6 hours of olaparib treatment. These studies constitute proof-of-mechanism of olaparib in humans.

In studies to determine whether olaparib treatment induced antitumor responses, the researchers found that such responses only occurred in patients with confirmed BRCA1 or BRCA2 mutation carrier status, except for one patient who declined mutational testing but had a strong family history of BRCA mutation-related cancer. 23 patents who were confirmed or (in the one case) deemed to be BRCA mutation carriers were treated. Of these 23 patients, two could not be evaluated. Two of the remaining patients had tumors not typically associated with BRCA mutations, and neither received clinical benefits from drug treatment.

Of the remaining 19 patients (who had ovarian, breast, or prostate cancer), 12 exhibited clinical benefits from olaparib treatment, with either tumor responses (determined radiologically or via serum tumor markers) or stable disease for a period of four months or more. Nine BRCA carriers had a tumor response. Eight patients with advanced ovarian cancer had a partial response (determined by radiology), and six of these had a greater than 50% tumor response based on tumor marker assays. Of the three patients with advanced BRCA2 breast cancer, one had a complete remission lasting for over 60 weeks, and another had stable disease for 7 months. The other breast cancer patient, who had refused mutational testing, had a decline in metastases and an over 50% decline in serum tumor markers. The patient with BRCA2-related castration resistant prostate cancer has an over 50% reduction in PSA levels, and resolution of bone metastases. He had been participating in the study for over 58 weeks at the time of the cutoff date, and for more than 2 years since that date.

The above efficacy data constitutes proof-of-concept, and confirms the hypothesis that BRCA-associated cancers can be addressed by a synthetic lethal therapeutic strategy based on the use of the PARP inhibitor olaparib. Olaparib also has a satisfactory adverse effect profile, and lacks the toxicity typically seen with cancer chemotherapy. Since this Phase 1 clonal trial, AZ had taken olaparib into Phase 2 clinical trials in advanced BRCA-related breast and ovarian cancer. Olaparib has continued to demonstrate efficacy and a relatively mild adverse effect profile in these trials, as shown here and here, and as also discussed in a July 2010 Medscape article.

Dr. Ashworth and his colleagues noted that not all cancers in BRCA1 or BRCA2 carriers respond to olaparib. They hypothesize that different BRCA1 or BRCA2 mutations may result in different defects in homologous recombination, which may cause variations in sensitivity to PARP inhibition. Moreover, certain secondary BRCA2 mutations may restore BRCA function, which may cause resistance to PARP inhibition. They see the need to develop assays for homologous recombination proficiency, which might be used in reiterative translational studies to determine causes of resistance to olaparib.

Synthetic lethal therapy with PARP inhibitors such as olaparib may be applicable to other types of cancers that have defects in DNA repair by homologous recombination. These may include sporadic breast and ovarian cancers that acquire loss of function of BRCA1 or BRCA2 via somatic genetic or epigenetic events, and other sporadic cancers that develop loss of function (via somatic genetic or epigenetic events) of other proteins involved in the homologous recombination DNA repair pathway.

Dr. Ashworth and his colleagues have also shown that loss of function of DNA damage signaling proteins (e.g., ATM, ATR, CHK1, CHK2), and of Fanconi anemia proteins, can induce sensitivity to PARP inhibition. Loss of function in these pathways may be relatively common in other sporadic cancers. It will be essential to develop biomarkers for loss of function of these DNA repair proteins in order to design hypothesis-testing clinical trials to investigate the potential of olaparib (or other PARP inhibitors) to treat this broader class of cancers.

As show by these three examples–and the other examples discussed in the 30 September 2010 de Bono and Ashworth Perspective (see Box 5 in that article)–researchers have been using the personalized medicine hypothesis-testing strategy to develop exciting new oncology drugs to treat disease in specific classes of patients. However, except for the case of imatinib, all of the drugs are still in clinical trials and have not yet achieved registration, which is the real test of the success of this strategy. Moreover, as we discussed in the first article in this series, the personalized medicine hypothesis-testing strategy is a work in progress. For example, biomarker identification and qualification/validation, which is a critical need for further development and utilization of this new clinical trial strategy, is an early-stage area of science and technology. Nevertheless, the personalized medicine hypothesis-testing strategy for cancer drug development provides a means to extend biology-driven drug discovery into the clinic, to decrease the time and cost of clinical trials, and to develop anticancer drugs that should be superior to both conventional chemotherapy and to early-generation targeted drugs.

——————————————————————————————-

As the producers of this blog, and as consultants to the biotechnology and pharmaceutical industry, Haberman Associates would like to hear from you. If you are in a biotech or pharmaceutical company, and would like a 15-20-minute, no-obligation telephone discussion of issues raised by this or other blog articles, or of other issues that are important to  your company, please contact us. We also welcome your comments on this or any other article on this blog.

The 30 September issue of Nature included a major emphasis on translational research in cancer. Featured articles included an editorial, a Perspective, and a research report. There was also an online “Specials” archive on translational cancer research, containing many recent research reports, all with free access to nonsubscribers.

The theme of these articles was the development of novel strategies for accelerating the translation of research on cancer biology into safe and efficacious therapies.

The Perspective, entitled “Translating cancer research into targeted therapeutics”, by British researchers Johann S. de Bono, M.D., Ph.D. and Alan Ashworth, Ph.D., outlines a novel disruptive clinical trial strategy for accelerating the translation of biology-driven oncology drug discovery into the clinic, with an early determination of proof of concept. This new strategy is designed  to ameliorate the high levels of Phase 2 and Phase 3 attrition of cancer drugs, as well as to lower the cost of clinical trials and to shorten the time from preclinical studies to the approval and marketing of oncology drugs that successfully emerge from clinical trials. It also is designed to aid in the development of therapies that provide greater patient benefit (in terms of progression-free survival) than the typical new oncology drugs that reach the market.

We have discussed clinical trial strategies of this type in two of our publications–our 2009 book-length report Approaches to Reducing Phase II Attrition (available from Cambridge Healthtech Institute, and our 2009 article published in Genetic Engineering and Biotechnology News, “Overcoming Phase II Attrition Problem” (available on our website). The de Bono and Ashworth article provides a more detailed and specific presentation of this strategy with respect to oncology drug development, and provides several examples of its successful application.

In the traditional Phase 1/Phase 2/Phase 3 format of cancer drug development, clinical studies focus on treating patient populations with advanced cancers that have not been characterized in terms of their genetics and molecular biology. The trials culminate in large, pivotal randomized Phase 3 trials that typically last several years, and are aimed at regulatory approval. In most cases, new drugs that emerge from Phase 3 trials and win approval only improve survival by a few months. Patients who participate in Phase 1 and Phase 2 clinical trials usually derive little or no benefit, and a high proportion of drugs fail in Phase 2 or Phase 3. This clinical trial format determines how well a particular drug or drug combination works for the average patient. However, that treatment might not be the best for a given individual patient.

As basic researchers have advanced the study of molecular genetic pathways of cancer, drug discovery researchers have been developing targeted therapies for cancer. These drugs–such as monoclonal antibodies (MAbs) like trastuzumab (Genentech/Roche’s Herceptin) and kinase inhibitors like imatinib (Novartis’ Gleevec/Glivec) work by modulating specific biomolecules (e.g., overexpressed or mutated oncogenic proteins) that are critical for the malignant phenotype. Population-based clinical trials of unselected patients make little or no sense in developing targeted therapies. Instead, clinical researchers need to first select groups of patients whose cancers express the biomolecule to be targeted, and preferably whose cancers are driven by that particular biomolecule. This way of thinking leads to the formulation of a new clinical trial strategy, as outlined in the Perspective.

Drs. de Bono and Ashworth call this strategy a personalized medicine (or “stratified medicine”) hypothesis-testing approach. The first step in this strategy is to develop a strong biological hypothesis that a particular altered molecular target is critical for the malignant phenotype of a particular cancer. This hypothesis is usually generated as the result of laboratory and clinical studies. Researchers need to show that blocking of the function of the altered target results in a lethal or cytostatic effect in cancer cells that express the the target, but not in normal cells that do not. It is also preferred that resistance to agents that block the target is not easily gained.

In the discovery stage of this strategy, researchers need not only to identify and optimize clinical candidate drugs that modulate the target, but also biomarkers that can be used to identify patients whose tumors express the altered target and are therefore likely to benefit from treatment. These biomarkers are called “enrichment biomarkers”, and have the potential to become predictive biomarkers. (Predictive biomarkers may also be the basis for the development of companion diagnostics). It is also important to identify pharmacodynamic biomarkers (biomarkers that can be used to determine target occupancy by the drug) and intermediate endpoint biomarkers (which can assess antitumor activity of the drug–for example, radiological assessment of tumor regression). Identification and qualification/validation of biomarkers is a work in progress, and is a critical need for further development and utilization of the personalized medicine hypothesis-testing clinical trial strategy.

Once targets and drugs that modulate them have been identified, they must be validated in animal models. The issue of the inadequacy of current mouse models of cancer–mainly xenograft models in which human cancer cell lines are transplanted into immune deficient mice–to predict drug efficacy is important both in the traditional cancer drug development strategy and in the novel strategy discussed in the de Bono and Ashworth article. We have discussed development of improved animal models for cancer drug development in an earlier blog post. de Bono and Ashworth note that there is an urgent need to develop such improved animal models. Nevertheless, as discussed in the de Bono and Ashworth article, there are examples of the successful implementation of the personalized medicine hypothesis-testing strategy of cancer drug development that have used traditional animal models in the preclinical phase.

In the personalized medicine hypothesis-testing strategy, clinical trials have the same three phases as in traditional trials. However, the trials involved stratification of patients using biomarkers, such that clinical studies are done in patients whose tumors express the target of the drug. Trial design is also more flexible and adaptive, and is contingent on obtaining key clinical data. The trials focus on determining the following as early as possible:

• Proof of mechanism: determining a dose range and dosing schedule under which the drug achieves sufficient target occupancy for long enough, using biomarker-based pharmacodynamic assays.

• Proof of concept: Determining that once sufficient target occupant is achieved, the drug exhibits antitumor activity, as determined using intermediate endpoint biomarkers.

In first-in-human clinical trials, in addition to determining safety and tolerability and evaluating pharmacokinetics and pharmacodynamics as in traditional Phase 1 trials, researchers also pursue rapid dose escalation, until proof of mechanism is achieved, using the appropriate biomarkers. Researchers then move on to proof of concept hypothesis testing, at doses and dosing schedules (ideally, the maximum tolerated dose) that are sufficient to address the target for long enough to have a biological effect. Ideally, researchers should move seamlessly from determination of proof-of-mechanism to assessment of antitumor activity, via adaptive trial design and patient selection using enrichment biomarkers.

If the above early-stage strategy results in a strong determination of proof of concept, this provides the basis for moving on to Phase 3 trials in patients selected using enrichment/predictive biomarkers, with the goal of drug registration. Such Phase 3 trials should have a higher probability of success than traditional Phase 3 trials in unselected patient populations, with less than adequate demonstration of proof of concept in Phase 2.

However, in the personalized medicine hypothesis-testing strategy, there is also the need for reiterative translational studies, between the laboratory and the clinic and back to the laboratory. Such studies should be designed as early as possible in clinical development. For example, clinical trials might allow researchers to obtain tumor samples to determine mechanisms of drug resistance. Such studies might form the basis for generating further hypotheses that are relevant to reversing drug resistance, via such means as development of combination therapies or of second-generation drugs.

The personalized medicine hypothesis-testing strategy is a work in progress. However, as we shall discuss in Part 2 of this series, there are examples of its successful implementation. And this strategy provides a means to extend biology-driven drug discovery, arguably the most successful drug discovery strategy of the past decade, into early and mid-stage clinical trials, thus increasing the probability of clinical success.

Nevertheless, it must also be emphasized that our understanding of disease biology (especially cancer biology) is limited, thus limiting our ability to successfully carry out biology-driven drug discovery in all cases. However, as our understanding of disease biology grows–in an incremental manner–as the result of basic research mainly in academic laboratories, we should be able to utilize this research to develop novel, breakthrough treatments via biology-driven drug discovery and personalized medicine hypothesis-testing clinical trials.

The April 1, 2010 issue of The Scientist has an article, entitled “Building a better mouse”, on efforts of researchers to develop improved mouse models of cancer.

Current mouse models of cancer, mainly xenograft models in which human cancer cell lines are transplanted into immune deficient mice, are notoriously unpredictive of efficacy when oncology drug candidates are tested in them. This is a major factor in the high failure rate of oncology drugs in clinical trials. It is estimated that oncology drugs that enter human clinical trials have a 95 percent attrition rate, as compared to the 89 percent attrition rate for all clinical candidates. (Poorly predictive animal models are a major factor in the failure of clinical candidates in all therapeutic areas, but cancer models are particularly unpredictive.)

The Scientist article focuses on the ongoing “co-clinical mouse/human trials” now being led by Pier Paolo Pandolfi, MD, PhD (Director, Cancer and Genetics Program, Beth Israel-Deaconess Medical Center Cancer Center and the Dana-Farber/Harvard Cancer Center). Dr. Pandolfi and his colleagues have constructed genetically engineered transgenic mouse strains that have genetic changes that mimic those found in human cancers. These mouse models spontaneous develop cancers that resemble the corresponding human cancers. In the co-clinical mouse/human trials, researchers simultaneous treat a genetically engineered mouse model and patients with tumors that exhibit the same set of genetic changes with the same experimental targeted drugs. The goal is to determine to what extent the mouse models are predictive of patient response to therapeutic agents, and of tumor progression and survival. The studies may thus result in validated mouse models that are more predictive of drug efficacy than the currently standard xenograft models.

The human clinical trials being “shadowed” by simultaneous studies in mice include Phase III trials of several targeted therapies for lung and prostate cancer. Xenograft models in which tumor tissue from the patients have been transplanted into immunosuppressed mice are being tested in parallel with the genetically engineered mouse models. This two-year project represents the most rigorous test to date of how well genetically engineered mouse models of cancer can predict clinical outcomes.

Dr. Pandolfi started in the mouse cancer model field with his studies of acute promyelocytic leukemia (APL). Unlike humans, mice do not naturally develop APL. Chromosomal translocations, in which the gene for the retinoic acid receptor alpha (RARα) (located on chromosome 17) becomes fused to one of several partner genes (known as “X genes”) on different chromosomes, are involved in the causation of APL. In over 98% of cases of APL, RARα is fused to the promyelocytic leukemia (PML) gene, located on chromosome 15. In a relatively small percentage of cases, RARα is fused to other X genes. An example of one of these other genes is the promyelocytic leukemia zinc finger (PLZF) gene, located on chromosome 11.

In studies in the late 1990s, Dr. Pandolfi and his colleagues constructed transgenic mice that expressed either PML-RARα or PLZF-RARα transgenes, in a promyelocytic-specific manner. (Expression of these transgenes in every cell of a mouse embryo results in embryonic lethality, and their expression in all early hematopoietic progenitors results in impaired myelopoiesis but no leukemia; these transgenic mice are thus not informative with respect to APL. The researchers were able to model PML only by expressing the transgenes specifically and exclusively in promyelocytes.)

The promyelocytic-specific PML-RARα-transgenic mice exhibit abnormal hematopoiesis over their first year of life, and between 12-14 months of age 10% of them develop APL.The promyelocytic-specific PLZF-RARα transgenic mice also exhibit a long latency period, and a subset of these mice eventually develops a leukemia that has features of human chronic myelogenous leukemia (CML).

Importantly, the above transgenic mouse models were useful in designing therapies for human patients. The leukemias in both the PML-RARα-transgenic mice and in patients with the PML-RARα translocation were responsive to treatment with all-trans retinoic acid (ATRA) (Genentech’s Vesanoid, generics). However, both the PLZF-RARα transgenic mice and patients with APL bearing the PLZF-RARα translocation were not responsive to ATRA. APL patients who initially responded to ATRA developed resistance to the drug, as did the PML-RARα transgenic mice. Using the PML-RARα transgenic mice, the researchers found that a combination of ATRA with arsenic trioxide (As₂O₃) (Cephalon’s Trisenox) cured the mice of leukemia. This later proved to also be true for human patients with APL bearing the PML-RARα translocation. Thus a cancer that once was uniformly fatal now has an approximately 90% survival rate.

Leukemic mice with the PLZF-RARα transgene were not responsive to As₂O₃. However, later studies have indicated that histone deacetylase inhibitors such as phenylbutyrate, in combination with ATRA, may be effective in treating these transgenic mice. These drug combinations may therefore be effective in APL patients with the PLZF-RARα translocation.

The success of Dr. Pandolfi’s genetically engineered mouse model in designing an effective therapy for the major type of APL illustrates the potential power of improved mouse models for cancer. Of course, this is a special case, since researchers were able to use the model to design an effective therapy using already-approved drugs. In most cases, researchers use the models to develop novel therapeutic strategies for a particular cancer, which involves discovery and development of new drugs or design of clinical trials using experimental drugs that have yet to be approved. The “co-clinical mouse/human trials” being run by Dr. Pandolfi and his colleagues may result in additional validation of the power of genetically engineered mouse models of cancer, and may thus encourage their adoption by companies developing new oncology drugs.

Our recently published book-length report, Animal Models for Therapeutic Strategies, includes a case study on a genetically engineered model of pancreatic cancer. Pancreatic cancer is one of the most lethal of cancers. Although models bearing transplanted human pancreatic tumors (i.e., xenograft models) are sensitive to numerous chemotherapeutic agents, human pancreatic cancers are insensitive to the same agents. Using a genetically engineered mouse model of pancreatic cancer, researchers hypothesized that the reason for the insensitivity of human pancreatic cancer (and of tumors in the mouse model) is impaired drug delivery. Researchers have been using the mouse model to develop novel therapeutic strategies to enhance drug delivery and thus to achieve improved treatment of this disease.

Our 2009 book-length report, Approaches to Reducing Phase II Attrition, includes a case study on adoption of genetically engineered cancer models by industry. Most animal models designed to enable researchers to develop novel therapeutic strategies for complex human diseases are developed by academic researchers. This includes genetically engineered cancer mouse models. However, most drugs are developed by industry, not academia. Industrial researchers are hampered in their ability to develop successful new oncology drugs by the poorly predictive xenograft models. Genetically engineered mouse models of cancer may help biotechnology and pharmaceutical company researchers to be more productive in oncology drug development, provided the corporate researchers can adopt these animal models for use in their discovery research and preclinical studies. However, for several reasons, industry has not widely adopted these models.

Our report discusses the barriers to adoption of these models, large pharmaceutical companies that are beginning to adopt the models, and the biotechnology company Aveo Pharmaceuticals, whose technology platform is based on in-licensing genetically engineered mouse cancer models from its principals’ academic laboratories and developing new models in-house. Aveo uses its models in its own internal drug discovery and development, and also collaborates with several large pharmaceutical companies. Aveo thus serves as a means of technology transfer from academia to industry, including both to its own internal programs and to its partners. The article in The Scientist also discusses Aveo’s research on genetically engineered mouse cancer models, and their use in the company’s internal drug development programs.

During the week of February 22, 2010, the New York Times (NYT) ran a three-part series on a Phase I trial in 2008/2009 of a targeted therapy for metastatic melanoma, a disease that is almost always fatal within a year. The trial was led by Keith T. Flaherty, M.D. (then at the University of Pennsylvania in Philadelphia, and now at the Dana-Farber Cancer Center in Boston). The drug was PLX4032, developed by Plexxikon, which is co-developing the compound with Roche. PLX4032 is a kinase inhibitor, which specifically targets the V600E mutant of the B-Raf oncoprotein. This is the most common somatic mutation found in human melanomas. Researchers believe that B-Raf(V600E) is a “driver mutation” that is particularly critical for the malignant phenotype of human metastatic melanomas that carry the mutation. PLX4032 entered Phase III clinical trials in 2009.

The NYT series, authored by Amy Harmon, focused on the stories of several patients, and on the dogged efforts of Dr. Flaherty to help his patients and to prove the value of targeted therapy. Although the targeted kinase inhibitor imatinib (Novartis’ Gleevec/Glivec) produces complete responses in the majority of treated patients in the chronic phase of CML (chronic myelogenous leukemia) and long-lasting remissions in many of these patients, many researchers believe that this is a special case, and they cite evidence that targeted therapy, especially in solid tumors, almost never produces durable responses. But Dr. Flaherty pressed on with his quest to prove the value of targeted therapy, despite this skepticism.

A key point in the story was when the original formulation of PLX4032, at the highest dose that patients could absorb, produced neither adverse effects nor clinical responses. Because of his belief in targeted therapy, and in this particular drug, Dr. Flaherty convinced Roche to reformulate the drug to enable patients to absorb a higher dose. With the higher doses of the drug made possible by the new formulation, the researchers saw dramatic clinical responses in the great majority of patients whose tumors contained B-Raf(V600E). Responses lasted an average of nearly 9 months, a dramatic breakthrough in treatment of metastatic melanoma.

As the series ended, Dr. Flaherty was working with his colleagues and the pharmaceutical industry to find ways to enable the testing of combination therapies of targeted drugs (including PLX4032) that might result in long-lasting remissions in patients with metastatic melanoma. Meanwhile, Plexxikon and Roche have taken PLX4032 into Phase II clinical trials and now into Phase III.

The NYT series is essentially a human-interest story. I commend it to all researchers, executives, and consultants in the industry whose work does not involve contact with patients, since creating products that can help patients is what our work is all about.

Dr. Flaherty reminds me, and others who have commented on this story, of Brian J. Druker, M.D. at the Oregon Health Sciences University in Portland. It was Dr. Druker’s efforts, centered on helping patients and proving the value of targeted therapy, that was the driving force behind the development of imatinib (Novartis’ Gleevec/Glivec). Without this effort (conducted in collaboration with biochemist Nicholas B. Lydon, then at Novartis), the whole field of kinase inhibitors for targeted therapy of cancer would not have emerged. Dr. Flaherty, as well as several other oncologists, is continuing this worthy tradition.

As pointed out to me by a leading Boston-area academic researcher in a cancer-related area, the NYT series did not give credit to the academic researchers who identified the role of B-Raf in cancer, and especially the role of B-Raf(V600E) in human melanoma. (For that matter, it did not credit the Plexxicon researchers who discovered PLX4032.) She said that the series sounded as if only one person, Dr. Flaherty, was responsible for the development of PLX4032. Moreover, the development of imatinib was made possible by decades of academic research on the target of the drug, Bcr-Abl, a fusion protein formed as the result of a chromosomal translocation. Drs. Druker and Lydon thus were not solely responsible for the development of imatinib either.

The academic researcher has a point. However, some industry commentators take a contrary point of view, downplaying the role of academic researchers in the drug discovery/development process and giving most of the credit to industry.

For years, we have taken the point of view that biology-driven drug discovery and development (arguably the most successful drug discovery/development strategy in the post-genomic era) requires the contributions of both academia and industry, and that more effective collaboration between academia and industry would result in more effective drug discovery and development. (See also my 2005 letter to the editor of BusinessWeek.)

It is basic research, usually in academic laboratories, that has resulted in the very best validated targets. Basic research on a particular target typically takes years or even decades (as in the case of Bcr-Abl). Many of the breakthrough drugs that have emerged in the past 10-15 years (as well as numerous promising pipeline drugs now in clinical testing) were made possible by this research. In contrast, large-scale “target validation” testing in industry more often than not results in targets whose role in normal physiology and in disease is poorly understood. This is an important cause of clinical attrition in drug development.

Nevertheless, it is industry, not academia, which uses this basic research to create drugs. In particular, it is industry that bears the enormous economic risk of drug development, especially of late-stage clinical trials.

Translational researchers, who are involved in taking the results of academic research and/or of discovery research in industry, and translating them into therapies that benefit patients, are—or should be—a key component of the drug discovery-development process. Drs. Druker and Flaherty are two outstanding examples.

However, at least some sectors of academia (and of governmental policy-makers and the media) are suspicious of the type of closer industry-academic collaboration that is needed to produce more effective translation of basic and drug-discovery research into the clinic. An editorial in the 25 February issue of Nature notes that there has been criticism of the recent hiring of William Chin, Lilly’s senior VP for discovery and clinical research, to be the executive dean for research at Harvard Medical School. The critics charge that strong research collaborations between academia and industry will inevitably result in conflicts of interest. The Nature editorial supports institutional policies that require disclosure of links between academic researchers and industry, but deplores the views of influential critics who believe that any collaboration between academic researchers and industry “corrupts” the academic research enterprise.

In addition to Nature, some leading academic researchers say that it is time for industry and the academic medical community to fight back against the critics, rather than appeasing them with ever more restrictive conflict-of-interest policies. These researchers note that the main purpose of medical research is not to publish scientific papers, but to translate this knowledge into therapies that benefit patients. This requires effective collaboration between academia and industry. We agree.

In the December 15, 2009 issue of Neurology, a research report by Stephen Salloway and his colleagues at the Butler Hospital and Brown University (Providence, RI) and an editorial by Dan Kaufer and Sam Gandy (University of North Carolina at Chapel Hill) focus on a Phase II multicenter placebo-controlled clinical trial of Elan/Wyeth’s bapineuzumab (AAB-001) in patients with mild to moderate Alzheimer’s disease (AD). (Wyeth is now part of Pfizer.) (A subscription is required to read the full text of both of these articles.) Bapineuzumab is a monoclonal antibody (MAb) drug that is specific for amyloid-β (Aβ) peptide. The dominant paradigm among AD researchers and drug developers is that the disease is caused by aberrant metabolism of Aβ, resulting in accumulation of neurotoxic Aβ plaques. This paradigm is known as the “amyloid hypothesis”.

The overall result of the study by Salloway et al. was that there was no difference in cognitive function between patients in the drug-treated and the placebo groups. However, the study did not have sufficient statistical power to exclude the possibility that there was such a difference. About 10% of patients treated with the agent also experienced vasogenic edema (VE), which was reversible. (Cerebral VE is the infiltration of intravascular fluid and proteins into brain tissue, as the result of breakdown of the blood-brain barrier.)

Retrospective analysis of the data suggested that bapineuzumab-treated patients who were not carriers of the apolipoprotein E epsilon4 allele (ApoE4) showed improved cognitive function as compared to placebo treatment, and that they had a lower incidence of VE than ApoE4 carriers. The ApoE4 polymorphism is the only known, well-characterized genetic risk factor associated with the development of late-onset AD. Of the three common isoforms of ApoE, ApoE3 is the most common, followed by ApoE4 and ApoE2, respectively. Unlike ApoE4, the ApoE2 allele appears to protect against development of AD. Some researchers estimate that allelic variations in ApoE may account for over 95% of AD cases.

In the study by Salloway et al., nearly two-thirds of the AD patients carried one or more ApoE4 alleles; thus only the remaining one-third of patients appeared to show positive effects of bapineuzumab treatment according to the retrospective analysis. However, the idea that the drug is efficacious in ApoE4 noncarriers is only a hypothesis, which will require prospective clinical trials to confirm. Elan and Pfizer are now conducting large Phase III clinical trials of bapineuzumab, which have prospectively segregated enrollment into ApoE4 carrier and noncarrier groups.

The hypothesized association of ApoE4 noncarrier status of AD patients with bapineuzumab efficacy and safety has been used as a case study in workshops on stratified medicine sponsored by the FDA, MIT, and industry partners in 2009 and 2010. You can read about the October 2009 workshop here. The most recent workshop was held at MIT on January 19, 2010. In these workshops, two case studies were discussed: the use of diagnostic tests for the HER2 receptor in identifying breast cancer patients who are likely to benefit from treatment with trastuzumab (Genentech/Roche’s Herceptin), and the bapineuzumab/ApoE4 case. The HER2/ trastuzumab relationship is well known and well characterized, and is considered to be a paradigm of stratified medicine. This contrasts with the bapineuzumab/ApoE4 association, which remains a hypothesis pending the results of the Phase III prospective clinical studies.

A growing minority of researchers is skeptical that the amyloid hypothesis is sufficient to account for AD pathogenesis in all stages of the disease or in various disease subpopulations, and they are investigating other pathways that may contribute to the disease, either in combination with the amyloid pathway or as alternative mechanisms. We have discussed alternative hypotheses for AD pathogenesis in a 2004 article published in Genetic Engineering News (available on our website), and in book-length reports published by Cambridge Healthtech Institute in 2006 and in 2009.

The search for alternative hypotheses takes on added urgency because of the clinical failure of several AD drugs that had been designed based on the amyloid hypothesis. These include Neurochem’s (now Bellus Health) Alzhemed (3-amino-1-propanesulfonic acid) and Myriad Pharmaceuticals’ Flurizan (tarenflurbil), both of which failed in Phase III clinical trials. Based on the overall results of the Phase II trial of bapineuzumab, most researchers and industry commentators would add bapineuzumab to the list, unless the stratified Phase III trial shows that the drug is significantly efficacious and safe for ApoE4 noncarriers.

Since ApoE4 carrier status is such a prominent risk factor for developing late-onset AD, might ApoE4 itself be a target for drug discovery in AD? Drs. Kaufer and Gandy suggest that such an approach might be fruitful, whatever the outcome of the Phase III trial of bapineuzumab. Several academic laboratories have been investigating mechanisms by which ApoE4 may be involved in the pathobiology of AD. You may read two recent papers on this subject here and here. ApoE4 may contribute to AD pathogenesis via multiple mechanisms, including by causing synaptic deficits and mitochondrial dysfunction in neurons, and by inducing endoplasmic reticulum stress leading to astrocyte dysfunction.

Given the prominence of ApoE4 expression as a risk factor for AD, the study of the mechanistic basis of ApoE4’s role in AD pathobiology needs greater attention. Hopefully, this research will lead to the development of novel therapeutic strategies for AD.