NEW ORLEANS—When the results of the C9011 randomized trial comparing fludarabine with chlorambucil in previously untreated patients with chronic lymphocytic leukemia (CLL) were reported in 2000, the fludarabine arm showed superior response rates and progression-free survival. The overall survival curves for both drugs, however, were superimposable.
But six years later those curves began to diverge, showing a statistically significant overall survival benefit in the fludarabine treated patients.
And now, eight years post-study, in an update of the North American Intergroup Study 9011 presented here at the most recent ASH Annual Meeting, investigators reported that 31% of CLL patients on fludarabine were alive, versus 19% for those receiving chlorambucil.
These outcomes are “somewhat philosophically intriguing,” said Kanti R. Rai, MD, Chief of the Division of Hematology-Oncology at Long Island Jewish Medical Center in New York, the principal investigator on the original trial and the long-term follow-up.
“If the influence [of a study drug] is superior in some effects but is not showing improved survivorship immediately, we should not give up on that arm,” Dr. Rai said after the meeting, in a telephone interview from Jaipur, India, where he spent the year-end holidays.
Dr. Rai said that when the C9011 trial began, fludarabine had been FDA approved as second-line treatment for CLL and chlorambucil was the only front-line choice—“This trial established fludarabine as front-line therapy as well.”
Interestingly, he added, the manufacturer never went through the Food and Drug Administration process for enlarging the approved indication for fludarabine from second line to front line after C9011.
“But it didn't seem to matter because the US-based hematologists and oncologists had already been using fludarabine as front line,” he said.
Dr. Rai said that the investigators were surprised, and somewhat disappointed at not seeing evidence of a fludarabine survival advantage in the 2000 report, which included five years of median follow-up.
“Now, with 10 additional years of follow-up, the outcomes are surprising again. We thought when patients went on second- or third-line therapy [after C9011 closed], those subsequent treatments would influence survivorship whether originally randomized to fludarabine or chlorambucil.
“But we have every reason to believe that if patients relapsed and needed other treatment, the menu of treatment available was identical for both study arms.”
During his oral presentation, Dr. Rai voiced some criticism about himself and about collaborators in C9011. Some of the original investigators, he said, were less than enthusiastic about collecting long-term follow-up data.
“You well know, the level of commitment goes down as the study becomes old. When the study is large and new, everyone has the questions fresh in their mind. But the questions fade from their memories, and so the importance of providing follow-up data becomes less important to them, and they become more involved in the new studies.”
He criticized himself as well, for not incorporating stricter follow-up rules in the protocol.
“At the time the study was planned, in 1988–89, we did not expect to get more data on a long-term basis, so we did not spell out in the protocol that we would require continued observations of the clinical state, year after year for 10 or 15 years,” Dr. Rai said.
“All we said was that we would collect survival data for 10 years on a year-to-year basis, because we did not think participating physicians would feel it relevant to follow patients after the active part of the study, patients who had been treated subsequently with other drugs.”
Dr. Rai said research leaders can be too cautious in asking participating investigators to supply more detailed information after the study's primary objectives have been met.
“My real criticism is that we investigators in the US have not been successful in entering the larger number of cancer patients on prospective clinical trials,” he explained during the interview. “Approximately 4% of eligible patients with any of the human malignancies are offered an appropriate trial and then accept and go onto that trial.”
But he also called some investigators and trial planners—including himself—prima donnas, “in the sense that when we start off [a clinical study] we all feel that in this disease we know what's going on, so there is nothing more to ask for or to learn, so we deliberately expect less and ask for less.
“In retrospect we kick ourselves and ask ‘why didn't we insist on asking for more information at yearly intervals?’”
He did applaud the approximately 80% of C9011’s participating investigators who continued reporting on survivorship, “people who had a serious sense of commitment to the cooperative groups’ study.”
“If you ask for a limited amount of data because you know from a practical point of view the doctors and their data monitors have limited time and funding, you become more and more restrictive in what follow-up data you ask for,” Dr. Rai said.
“My recommendation is, don't be restrictive—even if the data may sound like too much to ask for, we will regret it if we do not get that information. Important clinical data can become analyzable only if there is a quantitative high percentage of compliance in supplying that data. Think of all the tests you want your patients to go through following several years after the study, then insist on it and do not compromise on it,” he said.
He acknowledged that with more and more trials starting all the time, that demand becomes more difficult to comply with. But what I find time after time in conducting large randomized clinical trials, is that I have regretted every time that I have scrimped on testing in the follow-up period.”
C9011 Details, Old & New
Between 1990 and 1994, four cooperative groups collaborating in the C9011 trial, making up Cancer and Leukemia Group B, the Eastern Cooperative Oncology Group, the Southwest Oncology Group, and the National Cancer Institute of Canada-Clinical Trials Group, enrolled 509 eligible, untreated patients with symptomatic CLL.
Patients were randomly selected to received fludarabine (179 patients) or chlorambucil (193), or fludarabine-chlorambucil (137).
The combination arm was stopped early because of high morbidity and mortality.
From the last data collected during the trial, in June 1999, researchers reported that fludarabine provided significantly higher response rates and longer remission duration and progression-free survival than chlorambucil (Rai KR et al: NEJM 2000; 343:1750–1757).
In 2009, with eight more years of follow-up, 82% of the patients in the fludarabine group had died, compared with 89% of the chlorambucil group and 82% of the combination group. Median overall survival times were 63, 59, and 45 months, respectively.
The percentages of patients alive at eight years were 31% with fludarabine, 19% with chlorambucil, and 26% with the combination.
Reporting of second malignancies was required in the study, and in the 2009 report the overall incidences were 27 epithelial cancers (17%): nine on fludarabine, 11 on chlorambucil, and seven on the combination. These involved colon, lung, breast, prostate, pancreas, liver, bladder, and skin cancers (six squamous, two melanoma).
Seven therapy-related myeloid neoplasms were reported: six in the combination arm and one on fludarabine.
Richter's transformation to non-Hodgkin lymphoma was reported in 34 patients; prolymphocytic leukemia occurred in 10; Hodgkin lymphoma in six; myeloma in two; and hairy cell leukemia in one. These cases were distributed about evenly, with 18 on fludarabine, 18 on chlorambucil, and 17 with the combination.
Some of Dr. Rai's coauthors were asked for their opinion, but as of press time had not responded.