Other Candidate Predictive and Surrogate Markers of Antitumor Activity

Slide 1. Other Candidate Predictive and Surrogate Markers of Antitumor Activity

I will try to sum up some elements of what is going on in the development of erlotinib, looking at specific predictive and surrogate markers from previous trials.

Slide 2. Identifying Predictive and Surrogate Markers for NSCLC

When trying to identify predictive and surrogate markers for non-small cell lung cancer, we must first remember that prognostic factors are not necessarily positive or negative predictive markers. We also have to remember that predictive markers can be specific to a certain therapy or nonspecific to the type of therapy that is being used.

Surrogate markers differ by the fact that they have to be measurable after the start of therapy, which is in itself something that is very difficult in most clinical instances, and they should encompass a clinically relevant endpoint of the treatment.

Slide 3. HER1/EGFR: Prognostic Factor?

Coming back to our target, human epidermal growth factor receptor 1 (HER1)/epidermal growth factor receptor (EGFR), is it really a prognostic factor? Here is a study that was reported recently in a review by Sridhar in which they looked at patients who had surgery for non-small cell lung cancer, looked at the prognosis of those patients, and looked at HER2 and HER1 expression by immunocytochemistry (IC).

In this study it was not possible to demonstrate very effectively that HER1 was a good predictive marker of recurrence; but, when combined with HER2 expression, it seems to bring out some differences in the possibility of recurrence in these groups. Remember the number of patients in each of the groups remains quite small until now.

Slide 4. Factors Identified by Microarray: NSCLC Recurrence Post-Surgery

Are there factors that can be identified on microarray that can help us? Here is a study done by Wigle looking at more than 3000 genes in patients who had surgery for non-small cell lung cancer and looking at the recurrence pattern and the type of genetic pattern that they had.

Of the more than 3000 genes analyzed in this study, 16 genes were identified as prognostic indicators; 15 of them demonstrated statistically different expression levels between the good and the bad prognosis groups. Some of the genes that were identified were overexpressed, some were underexpressed, meaning that it is not just a fact of overexpression on certain molecules, but also of suppressor genes.

Slide 5. Factors Identified by Microarray: NSCLC Recurrence Post-Surgery (Continued)

This is the same study looking at the good prognosis group and the bad prognosis group. In this context, there was a molecular prognosis that was independent of the histologic subtype, and there was also no prognosis. The prognosis was independent of the HER1 status or expression at the messenger ribonucleic acid (mRNA) level.

Slide 6. Factors Identified by Microarray: NSCLC

Here is another study published in 2002 in Cancer Research by Sugita in which they were looking at genetic factors that could help identify non-small cell lung cancer. In this study, looking at metastatic tissue from non-small cell lung cancer and cell lines, they were able to identify 20 genes that were mainly related to testis-like and melanoma-like antigens. None of those genes is similar to the ones that we had in the previous study by Wigle, meaning that there is a large heterogeneity in non-small cell lung cancer, which is something we knew for a long time, but it can also bring different expression of different genes.

Slide 7. Target Characteristics: Effect of Level of HER1/EGFR Expression?

What we have now is the expression level that we know for non-small cell lung cancer by IC. It is unclear until now if there is a good correlation between expression of EGFR and prognosis. We know this association to be quite weak for non-small cell lung cancer, whereas, it seems to be much stronger for head and neck cancer, cervical cancer, ovarian cancer, and bladder cancer.

This type of analysis might be confounded by the methods that we are using to test HER1, and it is mainly immunohistochemistry (IHC) in those studies. When looking at different methods to look for area under the curve (AUC) or for epidermal growth factor (EGF) overexpression, there might be differences. I am looking at possible reverse transcriptase-polymerase chain reaction (RT-PCR) methods that can bring differences compared with AUC methods. We know that to be true for HER2 as well.

Slide 8. Statistical Considerations

Let's look at some statistical considerations when looking at predictive and surrogate markers of response or of antitumor activity. We know that there are recognized chemotherapeutic agents for non-small cell lung cancer. Until now, after more than 30 years of treating these types of cancer, there is no universally recognized, measurable biologic predictor or marker of response. The only selection basis we have for the new agents that we are using is the response rate, which in itself is the only surrogate marker we are able to use right now for non-small cell lung cancer.

We know that there is a survival advantage for combination therapy and that there is no demonstrated advantage for any combination that has been used compared with the others.

Slide 9. Statistical Considerations: Targeted Therapy

When looking at targeted therapy, there are some statistical questions that come to mind. Is it applicable to a whole population of patients with the same histologically defined disease, or do we have to look at molecularly defined disease when we are trying to do a trial with these types of agents? Is it possible to demonstrate differences in efficacy with conventional trials without selecting patients based on the presence of a specific therapeutic target? Is expression of a potential target sufficient evidence to warrant specific therapy?

Slide 10. Correlation With Breast Cancer: Herceptin Therapy

Let's try to correlate this with what we know for breast cancer and the development of Herceptin (trastuzumab). Trastuzumab was tested in metastatic breast cancer patients when there was proof of overexpression of HER2. If we try to apply the development of trastuzumab to the whole population of metastatic breast cancer patients without selecting for HER2, and if we make a hypothesis based on the results of the registration trial -- the hypothesis being that there is no disadvantage from receiving trastuzumab for HER2-negative patients and a survival of 20 months for HER2-negative patients -- and if we were to run a 750-patient trial with about 30% of patients with HER2 overexpression, the trastuzumab-treated group would have an overall survival of 21.6 months and the standard therapy group would have an overall survival of 20 months. That would not reach the level of statistical significance. It would require about 2500 patients to demonstrate a significantly and clinically relevant difference between the groups.

Slide 11. Herceptin Therapy for HER2+ NSCLC

Here is what we know about one of the target therapies that was done for non-small cell lung cancer overexpressing HER2 as defined by AUC.

In this randomized phase II trial, there was no difference between the group treated with just cisplatin and gemcitabine compared with the group treated with cisplatin, gemcitabine, and trastuzumab. No difference in response rate, in time to progression, and in survival.

In this study, only 20% of tested cases had overexpression of HER2; it was defined by AUC. Only 2% of patients presented overexpression of HER2 caused by overamplification of the gene, as proved by fluorescent in situ hybridization (FISH). Notice that 5 of the 6 patients who had HER2 overexpression caused by gene amplification had a response to the therapy containing trastuzumab.

Slide 12. Cell Regulation: Complex Molecular Interactions

We are stuck with this complex molecular interaction, knowing that we have a receptor (the HER1) that can be responsible for activation of different signal pathways that lead to cell proliferation. But HER1/EGFR is not the only one of the molecules in the receptor under the cell's surface that could be responsible for cell proliferation. We know that to be true for insulin-like growth factor (IGF), c-kit, platelet-derived growth factor (PDGF) receptor alpha and beta, and also other receptors.

Slide 13. Impact of Increased Tyrosine-Kinase Activity

We have either a model in which a large number of different molecules could be responsible for an exponential amplification of the signal pathways, or there could be a model in which only a few of those tyrosine kinases (TKs) are responsible for the proliferation of the cell. In this concept of defining which one of those is really important, we are trying to define if there might be a clonal event that is responsible for it.

Slide 14. Cell Regulation: Complex Molecular Interactions

We know that to be true for metastatic breast cancer with HER2 overexpression caused by gene amplification in which trastuzumab is a very good and active agent. We know that to be true for chronic myelogenous leukemia (CML) in which bcr-abl expression is increased because of a translocation. We know that to be true for gastrointestinal stromal tumor (GIST) in which there is CD117 overexpression caused by a mutation of c-kit. A specific therapy with a target that is very specific brings a very good response.

Do we know that to be true for non-small cell lung cancer? Not until now. We have to try to restrict this large number of possible targets and see if HER1/EGFR can be an adequate target with a possible clonal event leading to its overexpression.

Slide 15. Predictive and Surrogate Markers

In trying to define predictive and surrogate markers of response, we first have to try to characterize better the HER1 molecule and the downstream molecules that are activated by HER1. We can try to do that in a predictive mode, in which defining specific elements of HER1 in downstream molecules can help us predict whether or not there is going to be a better clinical outcome for those patients. We can also look at surrogate markers in the clinical trials and in clinical practice -- surrogate markers that can target dephosphorylation or demonstration of reduction of downstream activity, which can help us in trying to modify the regimen and try to then correlate that with a better clinical outcome. This would lead to validation of a marker and to future clinical trials.

Slide 16. Opportunities to Identify Predictive/Surrogate Markers of Activity

There are opportunities to identify predictive and surrogate markers of response of antitumor activity based on tumor characteristics and also based on host characteristics.

Slide 17. Target Characteristics: Should We Evaluate Activation Status as Well as Expression?

If we look at the target characteristics, there might be elements that are different regarding the ligands -- their concentration or level of expression regarding the molecule itself -- and HER1, whether or not there are mutations or a polymorphism that could explain differences in activity of erlotinib or other EGFR inhibitors.

If the dimerization part of HER1 is different, is it always a dimerization with HER1 or a dimerization with other partners making differences in the possibility to respond to therapy? We also know about looking at whether or not the phosphorylation status of HER1 is predictive in itself.

Slide 18. Target Characteristics: Effect of HER1/EGFR Gene Alterations?

Looking again at the target characteristics, we know that probably a good proportion of patients with non-small cell lung cancer present a mutation of EGFR. This leads to constitutively active TK, that is known to be very common for glial tumors, but we do not know exactly what percentage of patients present this mutation for non-small cell lung cancer. In different reports it is stated to be between 15% and 30%, but most of the analyses have been done by IHC and not necessarily by RT-PCR.

There is very little known about the polymorphisms and the other mutations of EGFR that can also influence the possibility to respond to an EGFR inhibitor and about the causes of gene amplification or other types of genetic elements that can lead to overexpression of the molecule.

Slide 19. The HER Signaling Pathways

Regarding the signaling pathways, knowing which type of dimerization is present at the cell's surface could help predict which type of pathway is specifically activated in a cell and is responsible for the proliferation of this tumor cell.

Looking at the other elements, downstream molecules, and looking for specific mutation of ras or other molecules can also explain why there is a possibility of response to EGFR inhibitors or no possibility of response to these types of agents.

Slide 20. Inhibition of Purified HER1/EGFR Phosphorylation by Erlotinib

There is inhibition of EGFR when treated with erlotinib. Looking at that in an in vivo setting has already been done and demonstrated that, in many instances, the tumor does seem to have a dephosphorylation of the HER1/EGFR when treated with erlotinib, but it is not the only target; we have to demonstrate that there is also a reduction in the signaling pathways.

Slide 21. Inhibition of Downstream Signaling Intermediaries in Erlotinib-Treated Tumors

This is from Hidalgo, a study presented at the 2001 American Society of Clinical Oncology (ASCO), in which he demonstrated in the head and neck cancer population that there was not only EGFR dephosphorylation, but dephosphorylation of extracellular signal-regulated kinase (ERK) and p70s6K, demonstrating that the inhibition of HER1/EGFR was also responsible for the reduction of the activation of the signaling pathways.

Slide 22. Determinants of Acquired Resistance to HER1/EGFR-Targetd Therapy

We also have to try to determine if there are determinants of acquired resistance to EGFR inhibitors. There are probably molecular determinants of these types of resistance.

Slide 23. Cytotoxicity of Erlotinib Against HN5 Cells Progressively Resistant to Erlotinib

There is an ongoing study that was reported by Perez-Soler this year at ASCO in which HN5, a head and neck cell line, was cultured with increasing concentration of erlotinib and then tested for the potential of erlotinib to still be active for those cell lines.

With different passages and with increasing concentration of erlotinib being present in the cell culture, there is an increase of the inhibitory concentration of 50% (IC₅₀) and development of a resistance.

Slide 24. Expression of HER Receptors in HN5 Cells Progressively Resistant to Erlotinib

This resistance was associated with a downregulation of HER1, HER2, and HER3 molecules in this setting of a cell line being cultured with an increasing concentration of erlotinib.

Slide 25. Effect of Erlotinib on Akt Activation in HN5 Wild-Type (WT) and Erlotinib-Resistant Cells (AE)

It was also correlated with an increase in Akt and phosphorylated Akt in the cell lines of head and neck.

Slide 26. Determinants of Acquired Resistance to Erlotinib

So, acquired resistance might be a very important surrogate marker to look for. We know that an acquired resistance to erlotinib might be caused by a downregulation of members of the HER family. Whether or not there is a positive or negative selection of cells that express HER molecules is something that will be looked at further. Also, there is an overexpression of genes that are involved in signaling pathways, probably caused by the activation of other genes that are present at the cell surface.

Slide 27. GRG-1: Gefitinib Resistant Gene 1

Not only could there be acquired resistance, but there can be an innate resistance to HER1 inhibitors. This was reported by Natale at ASCO 2003 in which he did screen for numerous genetic alterations. He identified the GRG-1 gene, which is responsible for the expression of a cell-surface protein that is closely related to HER2. In his clinical setting of people treated with gefitinib, he was able to demonstrate that there was an inverse correlation between GRG mRNA expression and the possibility of survival for those patients.

Slide 28. Host Factors of Response to HER1/EGFR Inhibition

What about host factors we have to look for? There are adverse events, which is the rash that can be looked at as a possible surrogate marker of response with a HER1 inhibitor, but let's also look at some elements regarding pharmacokinetics.

Slide 29. Steady-State Plasma Erlotinib/OSI-420 Concentration Predicts Overall Survival

This is from a large phase II head and neck patient trial that was done with patients with recurring disease treated with erlotinib. There is a correlation between the possibility of higher concentration being linked with an increase in survival. This has been proven not only for the active agent, erlotinib, but also for its major metabolite, OSI-420. In this case, there was a correlation between higher concentrations of the drug and the possibility of an increased survival.

In this trial there was also a correlation between rash and increased survival, but -- probably due to the small number of patients or because there is no correlation -- we were not able to demonstrate a correlation between the plasma levels that were obtained and the presence of a rash. So, it could be that they are 2 different elements or they could be related. We were not able to show that with the limited number of patients we looked at.

It is possible that those characteristics could be responsible for differences in response.

Slide 30. Host Characteristics

We have to look at the determination of elements that modify pharmacokinetics (PK), mainly P450 polymorphisms (especially 3A4 and 4A5, for which a large number of polymorphisms are known) and analyze comedication (chemotherapy and other comedication), even though, in most phase I studies until now, there has not been any demonstration that chemotherapy is responsible for very noted differences in PK.

Slide 31. Erlotinib Marker-Identification Program

In the context of all the trials that are going on with erlotinib, there is a marker identification program. We are going to be able to look at a large number of candidate markers -- based on the large number of clinical samples that are being obtained from TALENT and TRIBUTE, the large phase III trials; from BR.21, the phase III trial for patients with recurring disease; and from a phase II trial, doing biopsies before and during therapy -- to look at possible surrogate markers and looking specifically at signaling pathways.

Looking at a different possibility of analysis of those markers, either by RT-PCR or microarrays, we hope to be able to look at specific markers, either predictive or surrogate, that can help design better and more specific trials and that can get the response of HER1 inhibitors in this context.

Slide 32. Summary

In summary, there are numerous opportunities to identify candidate and surrogate markers and predictive markers of activity. Further research is necessary to understand the components and interaction of HER signaling pathways more fully. And the large collection of an analysis of tumor samples from the trials being done with erlotinib will certainly help in that process.