Journal of the National Cancer Institute, Vol. 95, No. 4, 254-255,
February 19, 2003
© 2003 Oxford University Press
Dosing Study Seen As Victory for Clinical Trials, Mathematical Models
Rabiya S. Tuma
A recent Cancer and Leukemia Group B (CALGB) clinical trial was by all accounts a major success; by simply decreasing the interval between chemotherapy treatments from 3 weeks to 2 weeks, clinicians saw a 26% proportional improvement in disease-free survival and a 31% proportional improvement in overall survival in women with node-positive breast cancer following surgery.
But for many researchers the importance of the trial goes beyond finding an improved method of treatment. It is also a proof of concept, of the Norton-Simon Regression model specifically, and more generally a clear demonstration that mathematical modeling has a place in cancer research and clinical trial design.
Trial Design and Results
The clinical trial, CALGB 9741, used a two-by-two factorial design to simultaneously test whether giving drugs sequentially or in combination made a difference in outcomes, and whether a 3-week interval in chemotherapy administration (conventional regimen) yielded different results from a 2-week interval (dose-dense regimen).
A total of 2,005 patients enrolled in the trial between September 1997 and March 1999. All the women had node-positive breast cancer and had either had a mastectomy or lumpectomy prior to enrollment. The eligibility criteria were relatively broad "to include patients that would be seen in normal practice," said Marc L. Citron, M.D., from Albert Einstein College of Medicine, New York, and lead investigator on CALGB 9741.
All patients received doxorubicin, cyclophosphamide, and paclitaxel; the arms of the trial differed by drug scheduling, not in dosage or drugs (see box). Patients in the dose-dense arms of the study also received granulocyte-colony stimulating factor (G-CSF), or filgrastim, to prevent neutropenia.
Among patients who received dose-dense therapy (arms II and IV), 82% were disease free after 4 years, compared with 75% of patients treated on the conventional 3-week schedule. The women who received the dose-dense treatment also had better overall survival at 3 years, with 92% of patients still alive versus 90% of those treated on 3-week schedule.
Citron noted during his presentation that, although G-CSF adds cost to the regimen, the toxicities were manageable in all four arms. The details of the study will be published this spring in the Journal of Clinical Oncology.
"I think it is a very important study," said Marc E. Lippman, M.D., chair of the Department of Internal Medicine at the University of Michigan in Ann Arbor, "though it is still early days for three reasons. The follow-up is still reasonably brief, it certainly hasn’t been confirmed in another study, and finally, it is possible, though not necessarily likely, that the differing dose density could result in unforeseen later toxicities, like leukemia or something we don’t know about."
But, Lippman emphasized, the study was well-designed, well-executed, "and, very importantly, biologically plausible." After all, the trial results were predicted by the Norton-Simon Regression Hypothesis and the mathematical model on which it is based. That, Lippman speculated, means "it’s not just the play of chance" as is the case with many empirically designed trials.
Mathematical Models
The Norton-Simon Regression Hypothesis, which was first put forward more than 25 years ago, states that the rate of regrowth between treatments will be proportional to the rate of tumor growth. Because tumors do not grow in a simple exponential manner but rather follow a Gompterzian growth pattern, with faster growth early on and slower growth as the tumor gets larger, then if tumors are given less time to regrow between treatments, the likelihood of cure is improved. In other words, said Larry Norton, M.D., from Memorial Sloan-Kettering Cancer Center, New York, "because there is less regrowth between therapies, the overall regression is increased."
"This would be the first time where someone has taken a mathematical model, tested it in a clinical trial and found that it is right," said David A. Cameron, M.D., from Western General Hospital in Edinburgh. "As someone who is a modeler, that is very very interesting, because . . . by and large, people who stand up and talk about models to clinicians tend to be looked at as ‘yes, but it is just a nice idea’."
But this is not just a question of self-justification. Both Norton and Cameron said that using mathematical models can improve trial design and help researchers identify which variables should be tested and which are likely to provide the most benefit to patients. Using mathematical modeling, researchers can test a large number of ideas very quickly, which can then be confirmed in the laboratory or in small clinical studies. With such a strategy, said Cameron, one might be able to avoid the expense and difficulty of large negative trials and streamline the development of positive ones.
Already Norton and his team at Memorial Sloan-Kettering are working to see if they can develop better applications of the concept tested in CALGB 9741. They are looking to see if dose-dense therapy improves the clinical outcome in prostate and lung cancer, and they are testing whether they can achieve another 30% improvement by decreasing the treatment intervals from every 14 days to every 10 days, as the model predicts they will. The important thing, said Norton, is that the concept of dose-dense therapy has been validated. "And it shouldn’t escape anyone’s notice that this protocol is shorter and less toxic," concluded Norton.
Similar articles found in:
