GENEVA—A major challenge in drug development is to make effective new compounds available in the shortest amount of time, with the least possible risks to patients, and involving the fewest possible patients that will ensure solid and convincing data. That concept was expounded on by Jaap Verweij, MD, PhD, Chairman of the Department of Medical Oncology at Erasmus University Medical Center in Rotterdam, in his Michel Clavel award lecture here at the European Organisation for Research and Treatment of Cancer-National Cancer Institute-American Association for Cancer Research (EORTC-NCI-AACR) Symposium on Molecular Targets in Cancer Therapeutics.
Dr. Verweij titled his talk, “No Risk, No Fun,” a favorite expression, he noted, of Dr. Clavel, a French clinical pharmacologist who died of lymphoma in 1993 at age 47 and whose research had focused on improving Phase I trials.
An overly risk-averse and defensive approach does not yield optimal speed in drug development and the current strategy of the pharmaceutical industry is both too defensive and too aggressive, Dr. Verweij said. He spelled out several new approaches to streamline and speed up drug development in oncology, including better ways to develop protocols, setting the number of trial sites, knowing the country/region-specific requirements for drug approval, the timing of pharmacokinetic analyses, and maintaining close relationships and good communication between investigators and their pharmaceutical partners.
Site selection is also critical, both number and location. As a clinical trialist, Dr. Verweij said that fewer is better and that the number of sites should usually be limited to just one or two unless circumstances require more, such as situations where there are very few patients affected.
Because of often slower patient accrual, “the more sites you have in a study, the longer your study will take,” he said. “One committed investigator is worth more than a million sites.”
More sites, though, also means that each one has fewer patients, limiting the experience with adverse effects, and therefore possibly missing some. This last point raises an ethical issue in using too many sites, Dr. Verweij noted. “Phase I experiments should use sites that understand pharmacology, that have an urge to understand the pharmacology, that have an urge to identify potential toxicity issues.”
This is different from the situation for site selection for Phase III trials, which are mainly looking for efficacy in a larger number of patients. But in a Phase I study, the fewer patients enrolled at each site, “the less commitment, the worse the data you're going to give,” he warned.
Since people in different parts of the world can also vary in their pharmacogenetics, investigators must select sites accordingly. For example, Asians may react to and metabolize drugs differently from Westerners. Geographic regions also focus differently in their acceptance of toxicity. Studies done in Asia may have to be repeated in the United States and Europe, and vice versa.
Drug Formulations & Pharmacology
For studies with oral drugs, multiple tablet or capsule strengths should be available prior to starting a trial, Dr. Verweij said. He described a study he performed in which the drug was available as a 1 mg tablet, with dose-limiting toxicity seen in all patients at a dose of 2 mg, but the 1 mg dose had no toxicity.
There was a six-month delay to produce a 0.5 mg tablet needed for dose titrations. The drug company turned down the investigators' suggestion to also produce 0.25 mg tablets. When 1.5 mg doses were given to patients, there again was no toxicity.
“The recommended dose for this drug is 1.5 [mg] because we have no other choice, but I don't think it's the optimal dose,” Dr. Verweij said. Lack of a variety of doses at the beginning of the trial slowed it down by at least six months, and even then, a further lack of flexible doses thwarted efforts to find what may be a better dose.
Measuring and interpreting pharmacokinetics very early in a trial may make it much more useful by showing investigators how drug dosing and scheduling could be modified while the trial is still running. Dr. Verweij suggested getting pharmacokinetic results back in “a couple of days or a maximum couple of weeks,” which he called “real-time pharmacokinetics.”
Knowing that pharmacokinetics is not linear may be important in dose escalations because that would put patients at unacceptable risks of getting unacceptable toxicity, he said. On the other hand, if the pharmacokinetics are saturable, dose escalation may be useless. “That information should be available while you're running the study,” Dr. Verweij said.
He cited the example of GDC-0449, an investigational inhibitor of the hedgehog pathway that showed saturable kinetics with early doses and an extremely long serum half life. Once such kinetics are known, researchers can determine how often to sample the serum levels—i.e., less often for long half lives and more often and sooner for shorter ones.
Similarly, food effects should be investigated early, he said, giving the example of lapatinib, where a high-fat meal greatly increases the drug's bioavailability, a finding that was discovered after the earlier studies. The label now says that lapatinib should be taken at a dose of 1,500 mg while fasting.
It is possible that patients can achieve comparable blood levels with much smaller doses if they take it with food—a potential cost saving for the payer but an economic loss for the drug's manufacturer, Dr. Verweij said, noting that grapefruit juice is another food that can raise drug levels through its inhibition of the intestinal cytochrome P450 3A4.
An expanded cohort in a Phase I trial may be useful in terms of pharmacokinetics and toxicity monitoring, and it enables better insight into variation between subjects and provides support for recommending doses for further studies. Still, it can not replace an appropriately designed Phase II study, Dr. Verweij cautioned, and it should therefore be limited to a maximum of 10 to 15 additional patients.
‘Don't Cut and Paste’
When designing and writing a study protocol, Dr. Verweij warned “Don't cut and paste—and it happens every single day.”
He advised spending extra time on careful protocol design for each study and said it will be earned back in study performance.
Typical inclusion and exclusion criteria are unrealistic and look for patients who are atypical—for example, too healthy, having good prognostic scores, etc. “We need patients in good condition, but we don't need a marathon runner,” he said.
Even with prognostic scores, it is very hard to predict who will survive for some period. Too often, he said, protocols are written by non-physician industry representatives and are unrealistically conservative, without justification from animal toxicology.
To accrue subjects faster, trials might well liberalize inclusion criteria such as the number of permitted prior systemic treatments. Individual tolerance is important, and investigators should take into account the mechanism of action of the drug.
But Dr. Verweij advised not to cut back on performance status, warning that inclusion of patients with a performance status of 2 usually only leads to interpretation issues on adverse effects “and doubles the thickness of the CRF [case report form].”
Protocols should include starting doses, dose-escalation steps, provisions for modifications, toxicity endpoint definitions, and caution if using biomarkers as response criteria.
Flexible dose escalation would begin at dose level 0, a selected starting dose based on animal toxicology. De-escalation to level −1 decreases the dose by 25% to 100%. Escalation to dose level 1 and onward doubles the dose until any Grade 2 toxicity is observed.
Further escalation, he said, would occur in increments of 25% to 100% based on observed toxicity in previous cohorts and taking into account human pharmacokinetics, balanced against animal pharmacokinetics and exposure goals.
Stay in Touch with Each Other
Dr. Verweij noted that some speed can be gained by keeping all the investigators, especially the lead ones at each site, fully informed about what is going on at the other sites at all times. In addition, cohort sizes should be flexible instead of staying fixed at three per dose level. Three to four may eliminate delays in finding replacements should someone drop out, he said.
Finally, Dr. Verweij discussed biomarkers, defining them as a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention.
Biomarkers are of two types, he said: Predictive ones for toxicity or efficacy, measured before dosing; and pharmacodynamic ones that serve as a surrogate for toxicity or efficacy measured after dosing.
Cautioning that true validation of a biomarker usually comes only once it has been correlated with outcomes in Phase III studies, Dr. Verweij said not to overestimate the potential of a biomarker for early clinical decision-making.
In an interview after his talk, he referred to the example of epidermal growth factor receptor (EGFR) expression in lung cancer, a prognostic marker for survival: “That doesn't mean it's a predictive marker for the outcome of effect of EGFR inhibition. So we have to discriminate predictive markers for treatment outcome from prognostic markers because of all the failures that we've had with many of those markers that we chased.”
By the definition of a biomarker, he said Response Evaluation Criteria in Solid Tumors (RECIST) qualifies as one. However, “I am a strong believer that RECIST responses are totally irrelevant,” he said. “What I do believe is that absence of tumor progression is crucial.”
He said that rather than looking at tumor regression, he carefully monitors progressions to see if they can be stopped for a long time. “Why not just look at progression, because that's the only response—because progression is also a response—that leads to a change in your clinical behavior, in your clinical treatment?” Dr. Verweij proposed.
He concluded his lecture by summarizing his criteria for the optimal Phase I study:
* Center selection.
* Protocol design.