|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2002 American Society for Clinical Oncology
Early Detection of Toxicity and Adjustment of Ongoing Clinical Trials: The History and Performance of the North Central Cancer Treatment Group’s Real-Time Toxicity Monitoring ProgramByFrom the Divisions of Medical Oncology and Biostatistics, Mayo Clinic, Rochester, and The Duluth Clinic, Ltd, Duluth, MN; and Iowa Oncology Research Association, Des Moines, IA. Address reprint requests to Richard M. Goldberg, MD, Division of Medical Oncology, Mayo Clinic, 200 First St SW, Rochester, MN 55905; email: goldberg.richard{at}mayo.edu
ABSTRACT: Prospective clinical trials are the gold standard for evidence-based methodology used to support changes in the practice of medicine. Clinical researchers, regulatory agencies, payers, and the public embrace the conduct of phase I, II, and III clinical trials as integral to improving patient care. The National Cancer Institute (NCI) funds a number of cooperative oncology groups to conduct such clinical trials in the United States. In order to protect enrolling patients, the NCI requires expedited reporting to allow rapid identification of severe side effects on NCI-sponsored clinical trials. However, chemotherapy drugs frequently cause predictable side effects, the rapid reporting of which would potentially overwhelm the system. This article describes the development and documents the performance of a real-time toxicity reporting system implemented by the North Central Cancer Treatment Group. The goal of this system is to supplement the currently required NCI adverse event monitoring procedures and to permit study teams to identify the need to modify ongoing clinical trials. The system has proven its value in the monitoring of phase II and III trials, including trial N9741, a three-arm, phase III, advanced colorectal cancer chemotherapy study exploring combinations of irinotecan, oxaliplatin, and fluorouracil. We believe the methods described present opportunities for improving patient safety in clinical research.
CLINICAL TRIALS are powerful tools for collecting evidence to provide a basis for changes in medical practice. Depending on sponsorship, trial conduct frequently involves a complex interplay between investigator teams, pharmaceutical companies, scientific committees, regulatory and funding agencies, patients, and the media. Once completed, trial results may be used to determine new drug indications and, through governmental collaboration with industry and insurance carriers, to bring new treatments to the bedside. Answering clinically relevant questions about the efficacy and safety of new drugs or drug regimens frequently requires the formation of collaborative groups to allow timely enrollment of a sufficient number of patients. Conducting clinical trials at multiple and geographically dispersed sites presents logistical challenges. Systems must be constructed to ensure treatment compliance, standard assessment of tumor measurements, accurate and timely reporting of data, confirmation of safe practices, and compliance with regulatory and ethical standards. The National Cancer Institute (NCI) competitively evaluates and funds cooperative oncology groups to conduct cancer clinical trials in the United States. These groups, under NCI supervision, have established methods for addressing issues related to trial conduct through investigator training and performance audits. The NCI has developed a system that cooperative groups must use to rapidly identify individual episodes of severe and unexpected side effects on clinical trials that it sponsors. Rapid notification occurs through the filing of an adverse drug reaction form that must be sent to the NCI. However, by their nature, chemotherapy drugs frequently cause predictable side effects, such as neutropenia or nausea, the immediate reporting of which could overwhelm the system. Therefore, for commercially available agents, the current NCI requirements for expedited notification are limited to unexpected life-threatening or lethal toxicity or "an increased incidence of a known toxicity." It is unlikely that any one investigator could identify an increased incidence of a known toxicity from their individual experience in treating patients. Clearly, toxicity and drug-activity monitoring is and should be a multitiered process. At the treating facility, the patient, their health care practitioners, and protocol-related clinical research associates generate and record data. That information is then collated and reported to the centralized statistical center. There, a study team including a study chair, data monitors, statisticians, and other physicians working together examine the data and periodically report on it to various stakeholders. These interested parties include the individuals entering patients onto the trial, local institutional review boards, and, in some cases, a group-wide external data safety and monitoring board (EDSMB). The role of the EDSMB in monitoring trials has been described elsewhere.1 Centralized regulatory officials at the NCI and the United States Food and Drug Administration also participate in toxicity monitoring. An essential element to facilitating this multitiered safety review process is timely submission and review of individual patient adverse event data. This article describes the development and documents the performance of a real-time toxicity reporting system implemented by the North Central Cancer Treatment Group (NCCTG) in 1998. This mechanism focuses on (1) rapid delivery of data from the local setting and (2) rapid communication of this information to the study team to ensure that pertinent information is available to the study investigators to permit intervention if needed. The approach adopted is analogous to the use of gauges with a red line for monitoring the performance of a machine. Our system has proven its value in phase II and III trial monitoring. On the basis of early toxicity reports, we have made midstudy protocol modifications in seven studies to date. These results underscore the need to dispense with the concept of clinical trials as a scripted effort that automatically produces a product and adopt the viewpoint that this is a dynamic process requiring timely and continuous monitoring. Studies will require midcourse adjustments in order to evaluate new treatment strategies while prioritizing patient safety. We hope that the NCCTG experience can serve as a parable to identify potential mechanisms to improve trial conduct, regulatory monitoring, and patient care.
The NCCTG is one of the cancer clinical trials group funded by the NCI. Its trials focus on treatment of cancer and treatment-related symptoms through a network of physicians, nurses, pharmacists, research associates, and other professionals located at more than 200 treating sites under the aegis of 23 member institutions. NCCTG’s mission is to provide state-of-the-art oncologic care to patients in their own communities. Data collected at each local treating institution are forwarded for centralized computer entry and analysis at the Statistical and Operations Office located at the Mayo Clinic in Rochester, MN. The NCCTG real-time toxicity monitoring program was created as the result of an experience on a single phase II chemotherapy study of a commercially available agent in upper gastrointestinal cancer. The postfacto discovery that six patients were hospitalized for life-threatening toxicity among nine patients entered at six different treating locations in this study illustrated the need for such a system. This situation became apparent to the study team after the trial’s principal investigator was notified of one episode of severe toxicity requiring hospitalization through a telephone call from a concerned investigator. The principal investigator then inquired about the experience of the three patients treated at a different center and discovered that each one had been hospitalized after first-cycle toxicity. All four patients had experienced diarrhea leading to dehydration, a dose-limiting toxicity of the drug being studied. Because this side effect was expected, immediate reporting had not been mandated by the protocol. On this discovery, patient accrual was suspended, immediate data submission from the treating center and entry at the NCCTG central office occurred, and physicians treating study patients were alerted. In this process, two additional patients with severe first-cycle toxicity were identified. On the basis of this information, the NCCTG instituted a protocol amendment to reduce the drug dose for future patients on the study. When the study reopened at the lower dose, rapid reporting of any severe toxicity occurring in this study was instituted. On reopening, the rapid reporting system provided data that prompted a second dose reduction, after which toxicity was reduced to acceptable levels. The experience on this trial identified the need for a systematic approach to real-time toxicity monitoring of life-threatening or lethal toxic events. In this trial, the toxicity occurred despite the fact that the agent and the administration schedule under study had been the subject of a number of previous trials. There was considerable published data, including phase III studies, reporting on its activity and toxicity in a different disease site. At the time of initiation of the study in 1997, the chemotherapy agent used was commercially available. Accordingly, investigators enrolling patients onto this trial followed the standard NCCTG operating procedure of submitting only unexpected life-threatening or fatal toxicity rapidly, with other adverse event data submitted to the operations office within 2 months of the time that the patient visit occurred. Data entry at the operations office required up to an additional month after receipt of the report from the treating institution. Such a time lag presented the possibility that adverse events could go unrecognized while new patients continued to be placed at risk.
As a result of this experience, in April 1998 the NCCTG implemented a real-time data monitoring system. This system was designed for expedited identification and tracking of life-threatening or lethal toxicities (ie, graded 4 or 5 by NCI common toxicity criteria, version 2) or any hospitalizations occurring in patients enrolled onto trials.2 The simple, one-page form that we have developed is reproduced in Fig 1.
This mechanism applies both to events reported through the usual NCI-mandated expedited reporting system and to those events that would not typically be reported under that system. For events reported through the usual NCI system, the mechanism consists of the entry of selected information into a database within 24 hours of the receipt of the report. For grade 4 events, grade 5 events, and hospitalizations for which the usual NCI system does not apply, the NCCTG requires the completion and faxing of a single page to the NCCTG operations office within 5 days of the study site’s knowledge of the event(s). Information from this form is also entered into the same database as that used for the NCI-mandated events, also within 24 hours of receipt. Each night, an automated computer program e-mails the study chair, study statistician, and data monitor to notify them of any event(s) that was reported the previous day and to report a cumulative summary of all events observed on that study. Under this system, the current median interval from the local treatment team awareness of an event to the NCCTG study team notification is 4 days.
The policy and procedures described above were implemented in the NCCTG in 1999. After an initial learning curve, compliance with the policy has been excellent. In 2000 and 2001, the median time between a certified research assistant learning of a reportable event and entry of that event into the NCCTG database was 3 days. Compliance is not perfect. The interval from the time that a certified research assistant becomes aware of an event to the time the event is entered into the database exceeds 15 days in approximately 25% of cases. This indicates the potential for further reduction in the time to database entry through education and training. However, we believe that these results are an excellent improvement in the timeliness of data submission when compared with the norm observed before its implementation. These results indicate that such a system can be implemented in the setting of a community-based clinical trials organization.
In the 3 years that the NCCTG system has been in place, it has detected unexpectedly high toxicity rates leading to immediate protocol modification(s) in six phase II trials (Table 1). The rates of any grade
The system is useful for monitoring trials in cancers regardless of stage, site of origin, or sponsoring group. These points are illustrated by the fact that the fourth and fifth studies to be modified involved patients with stages III and IV non–small-cell lung cancer. Finally, a trial of chemotherapy in advanced neuroendocrine carcinoma was modified to specify a lower starting dose level when 72% of patients (eight of 11) experienced grade 4 neutropenia during their first treatment cycle. All of these studies were continued with acceptable toxicity after protocol-specified drug dose reductions. To answer phase III questions, it is often necessary to recruit patients from centers affiliated with multiple cooperative groups. Implementing standardized procedures among groups with different governance and standard operating procedures potentially magnifies the logistical difficulties inherent in phase II single-group trials by several orders of magnitude. Nevertheless, the NCCTG did implement components of the real-time toxicity monitoring system among all participants of the phase III intergroup advanced colorectal cancer trial N9741. Intergroup trial N9741 is a phase III advanced colorectal cancer chemotherapy study exploring combinations of irinotecan (CPT-11), oxaliplatin, and fluorouracil (5-FU) initiated and led by NCCTG. By design, study N9741 compared five experimental regimens with a control arm of 5-FU and leucovorin popularly known as the Mayo regimen for the treatment of advanced colorectal cancer (Fig 2). The Mayo 5-FU 425 plus leucovorin (LV) 20 mg/m2 5-day regimen given every 4 to 5 weeks was an established therapy for which activity and toxicity profiles were well documented. The five experimental arms of N9741 included two containing CPT-11 and two containing oxaliplatin, all given with 5-FU and LV. The sixth arm prescribed oxaliplatin plus CPT-11.
Varying experience was available on the five experimental regimens at the initiation of study N9741. For two of the regimens, only phase I or phase I and limited phase II data existed (sequential CPT-11 followed by 5-FU/LV3 and CPT-11 plus oxaliplatin4). For the oxaliplatin + 5-FU/LV bolus regimen, data from a 113-patient trial among patients with 5-FU–refractory colorectal cancer were communicated to the investigators by pharmaceutical company representatives. The other two regimens (weekly CPT-11 + 5-FU/LV [the Saltz regimen5] and oxaliplatin + 5-FU/LV [the de Gramont regimen6]) had been tested in phase III trials. Because of the limited experience with two of the six original regimens, in study N9741 we implemented elements of the real-time toxicity monitoring system into this intergroup effort. We required the supplemental toxicity reporting form for NCCTG members but not for non-NCCTG participants. All expedited toxicity reports that were submitted, ie, both the NCI-required adverse drug reaction reports for unexpected events as well as the supplemental NCCTG reporting forms for expected events, were immediately entered into the NCCTG database. This entry generated the nightly e-mail reports as described, which provided the study team with adverse event data on virtually a real-time basis. The real-time toxicity reporting system led to a decision to close two of the six arms of the study early due to excessive toxicity. On the CPT-11 sequential arm3 of the trial, five deaths were observed in 61 patients (an 8% lethal toxicity rate). Because the weekly CPT-11 regimen5 seemed to be less toxic, and there seemed little likelihood that the sequential regimen would supplant it due to the toxicity concerns, this regimen was eliminated from the randomization schema. The second arm to be closed early due to toxicity was the oxaliplatin + 5-FU/LV bolus regimen. On this regimen, there were three deaths among 16 patients (a 16% lethal toxicity rate). Consequently, the dose of oxaliplatin was dropped from 130 mg/m2 to 100 mg/m2 and that of 5-FU from 320 mg/m2 to 260 mg/m2. Despite this, a fourth patient died after enrollment of 31 additional patients; this arm was terminated as excessively toxic. Concurrently, in March 2000, data from a trial that randomized patients with previously untreated metastatic colorectal cancer to the Saltz CPT-11 + 5-FU/LV regimen, the Mayo 5-FU/LV regimen, or CPT-11 alone were analyzed.5 The data from this trial and data from a similar7 European trial showed statistically significant advantages in survival, response rate, and time to progression for the addition of CPT-11 to 5-FU/LV. These data were presented to the Oncologic Drug Advisory Committee of the U.S. Food and Drug Administration and led to the approval of CPT-11 as part of the three-drug combination chemotherapy for first-line advanced disease. This led the investigator team at NCCTG and the gastrointestinal intergroup that oversees NCI-sponsored trials in gastrointestinal cancer to delete the 5-FU/LV arm of N9741. After those modifications, trial N9741 continued as a three-arm study, as illustrated in Fig 3. In February and March of 2001, the study team was again alerted to an imbalance in the number of deaths attributed at least in part to toxicity in the CPT-11 + 5-FU/LV arm of N9741. Earlier we conducted a literature review of the treatment-related death rates reported in selected large advanced colorectal cancer chemotherapy trials, to set the threshold that would trigger possible study intervention. A treatment-related death rate of approximately 1% was consistently noted in previous large trials in advanced colorectal cancer.5-9 The study team used that 1% death rate to set the "red line," which would signal a need for intervention in the trial. Because of inherent variations in monitoring a dynamic process, we determined that we would intervene if the treatment-related death rate became high enough that its associated 95% confidence interval no longer included 1%. In April 2001, two lethal events occurred in quick succession at a time when nearly 900 patients were on study, raising the treatment-related death rate on the CPT-11 + 5-FU/LV arm to 3.1% (nine of 293 patients; 95% confidence interval, 1.4% to 5.8%).
The NCCTG study team reviewed the research records on each of these patients, noting that patients who died soon after treatment experienced several consistent toxicities. Eight of the nine deaths occurred within the first cycle of treatment. Dehydration from diarrhea, nausea, and vomiting or thrombotic events led to hospitalization in most cases. These events were subsequently reviewed by an external panel (see10,11 for further details). By happenstance, the eighth and ninth treatment-related fatalities occurred within 3 working days of a scheduled meeting of the NCCTG EDSMB. After review, the EDSMB voted to suspend accrual to N9741 pending further review of the data. Patients in the midst of treatment were informed of these findings, and their doses were immediately reduced if they were in their first treatment cycle. Patients currently on the trial were asked to sign a revised consent form indicating their understanding of the potential risks of continued treatment. Consent forms for new enrollees were revised to reflect the additional potential risks. In addition to the actions taken on study N9741, the NCCTG EDSMB directed the study team to alert the group chair and the chair of the EDSMB for the Cancer and Leukemia Group B (CALGB) of these findings. CALGB was coordinating a trial using the Saltz regimen in an adjuvant trial in patients with resected stage III colon cancer patients. Review of data from the CALGB study, performed by the CALGB EDSMB, indicated an imbalance in 60-day all-cause mortality between the two treatment arms of that study. At the time this information became known, the study had reached its planned accrual goal and was, therefore, closed to patient accrual. N9741 was reopened for patient accrual with a reduced initial dose level on the weekly CPT-11 + 5-FU/LV regimen, with a very acceptable toxicity profile. This chain of events, facilitated in large part by real-time access to toxicity information, has resulted in a heightened awareness of the inherent toxicity in the treatment of advanced colorectal cancer in general and in new guidelines for the use of the Saltz regimen in particular.10 In conclusion, the NCCTG real-time toxicity monitoring program identified unexpectedly high rates of toxicity, including lethal events in phase II trials in patients with esophageal cancer, gastric cancer, cholangiocarcinoma, lung cancer, and neuroendocrine cancer within NCCTG and Mayo Clinic Cancer Center studies. A modified version of the system designed to be workable across cooperative clinical trials groups also identified toxicity problems in a phase III intergroup trial for treatment of advanced colorectal cancer. Once recognized, this series of events led to suspension of accrual and revision of the treatment programs in each trial. We have provided examples illustrating the utility of the real-time toxicity reporting system. In each instance, the rapid discovery of the toxicity problem allowed the trial to continue until its planned accrual or to be terminated early because of excessive toxicity. We believe the methods described in this article present opportunities for improving the monitoring of cooperative group, single-institution, and pharmaceutical company–sponsored clinical research, and ultimately for further safeguarding the safety of the patient participants. First and foremost, we believe that a real-time toxicity monitoring program such as the one that we have designed should be standard operating procedure for groups conducting clinical trials in oncology. Second, study teams need to meet regularly to review the data generated. The real-time access to data allows the clinical trial to become a more dynamic process, where midcourse corrections may allow trials or even treatment regimens to succeed where they previously may have failed due to excessive toxicity. The recruitment of patients to research studies is critical to advancing medicine. Every patient who chooses to enroll onto a clinical trial makes that decision for personal reasons. While some hope for the newest advance so that they can get the best possible therapy, others are motivated principally by altruism. The protection of patients currently enrolled onto a study, and those who have yet to enroll, is critical. We believe that real-time toxicity monitoring, made possible by modern technology, has the potential to significantly bolster this protection.
1. Ellenberg S, Geller N, Simon R, et al (eds): Practical issues in data monitoring of clinical trials. Stat Med 12:414-616, 1993
2. Sargent DJ, Goldberg RM, Mahoney MR, et al: Rapid reporting and review of an increased incidence of a known adverse event. J Natl Cancer Inst 92: 1011-1013, 2000 3. Fonseca R, Goldberg RM, Erlichman C, et al: Phase I study of CPT-11/5-FU/leucovorin. Proc Am Assoc Cancer Res 38: 76, 1997 (abstr) 4. Wasserman E, Goldwasser F, Ouldkaci M, et al: CPT-11/oxaliplatin (L-OHP) every 3 weeks: An active combination in colorectal cancer (CRC). Proc Am Assoc Cancer Res 39: A2190, 1998 (abstr)
5. Saltz LB, Cox JV, Blanke C, et al: Irinotecan plus fluorouracil and leucovorin for metastatic colorectal cancer. N Engl J Med 343: 905-914, 2000
6. de Gramont A, Figer A, Seymour M, et al: Leucovorin and fluorouracil with or without oxaliplatin as first-line treatment in advanced colorectal cancer. J Clin Oncol 18: 2938-2947, 2000 7. Douillard JY, Cunningham D, Roth AD, et al: Irinotecan combined with fluorouracil compared with fluorouracil alone as first-line treatment for metastatic colorectal cancer: A multicentre randomised trial. Lancet 355: 1041-1047, 2000[CrossRef][Medline]
8. Goldberg RM, Hatfield AK, Kahn M, et al: A prospectively randomized North Central Cancer Treatment Group trial of intensive-course fluorouracil combined with the l-isomer of intravenous leucovorin, oral leucovorin, or intravenous leucovorin for the treatment of advanced colorectal cancer. J Clin Oncol 15: 3320-3329, 1997 9. Leichman CG, Fleming TR, Franco MM, et al: Phase II study of fluorouracil and its modulation in advanced colorectal cancer: A Southwest Oncology Group study. J Clin Oncol 13: 1303-1311, 1995[Abstract]
10. Rothenberg ML, Meropol NJ, Poplin EA, et al: Mortality associated with irinotecan plus bolus fluorouracil/leucovorin: Summary findings of an independent panel. J Clin Oncol 19: 3801-3807, 2001
11. Sargent DJ, Niedzwiecki D, O’Connell MJ, et al: Recommendation for caution with irinotecan. fluorouracil, and leucovorin for colorectal cancer. N Engl J Med 345: 144-145, 2001 Submitted March 6, 2002; accepted July 30, 2002.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2002 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
4 toxicity both before and after study modifications are given in 
