Results from a retrospective study conducted by researchers at Dana-Farber Cancer Institute have found that women 63 years of age or younger with newly diagnosed breast cancer experienced more chemotherapy-related serious adverse effects than previously reported by large clinical trials. These adverse effects may be responsible for increased patient suffering and higher health care expenditures, the researchers said. The study was published in the August 16th issue of the Journal of the National Cancer Institute (2006;98:1108–1117).
“Although serious adverse effects from chemotherapy may be more common than previously estimated, our results do not suggest that chemotherapy does not work or that it should not be offered to women with breast cancer,” said primary investigator Michael J. Hassett, MD, MPH, a researcher at the Center for Outcomes and Policy Research at Dana-Farber Cancer Institute and an Instructor of Medicine at Harvard Medical School.
“In deciding whether or not to have chemotherapy, women with breast cancer should discuss the options with their doctor to determine what the best option for their individual case is.”
The coauthor of an accompanying editorial, John K. Erban, MD, Director of the Breast Cancer Program at Tufts-New England Medical Center, said that clinicians should be alert to the possibility that infrequent side effects may be more common than suggested by published studies.
He and Joseph Lau, MD, Professor of Medicine in the Division of Clinical Care Research, also at Tufts-New England Medical Center, noted that younger patients are frequently offered chemotherapy in the adjuvant setting, and though improvements in survival may be below 5%, women will often accept chemotherapy for as little as a 1% to 2% survival advantage.
This decision to undergo chemotherapy may stem from previous clinical trials, which have reported that chemotherapy-related serious adverse effects are uncommon in breast cancer patients.
Clinical Trials Focused on Survival
In addition, the editorial said, clinical trials are usually powered to assess endpoints such as disease-free and overall survival rather than the toxicities of interventions, a factor that may account for the seemingly rare occurrence of serious adverse effects in previous trials.
“It is important to remember that randomized trials are constructed to detect a difference in the effectiveness of a treatment, which may be 20 percent or more from the standard, rather than to detect a difference in rates of rare side effects, which may occur at one to two percent,” Dr. Erban said in an interview.
“Thus, the sample size is really not chosen to accurately detect infrequent events. Investigators have an easier time recording single events like response. Therefore, a fever, a pain, an emergency room visit, and so on may not get fully recorded.”
“The development of cancer therapeutics is efficacy-driven,” said Andy Trotti, MD, Director of Radiation Oncology Clinical Research at H. Lee Moffitt Cancer Center of the University of South Florida.
Dr. Trotti's article “The need for adverse effects reporting standards in oncology clinical trials” was referenced in Drs. Erban and Lau's editorial.
“Cure is the over-riding focus, prompting trial designs and advanced methods to precisely measure and report even small benefits,” Dr. Trotti said.
“Since toxicity is expected—and considered largely unavoidable—complementary methods and standards to report and compare safety profiles have not kept pace. One must view safety data with much less confidence since the numbers are, by nature, only rough estimates. The Hassett article serves to increase our awareness of inherent limitations in the safety reporting process.”
Using a medical information database, Dr. Hassett and his colleagues A. James O'Malley, Juliana R. Pakes, Joseph P. Newhouse, and Craig C. Earle examined the insurance claims filed between January 1998 and December 2002 by individuals with employer-provided health insurance.
A total of 12,239 women younger than 64 years of age with newly diagnosed breast cancer were identified: 4,075 of these patients received chemotherapy during the 12 months following the initial breast cancer diagnosis whereas 8,164 did not.
These two cohorts of chemotherapy and non-chemotherapy patients were further whittled down into two groups of 3,526 patients each.
Diagnostic codes for eight chemotherapy-related adverse effects were identified.
These serious adverse effects, which were chosen by researchers based on their association with chemotherapy in previous clinical trials, included: infections or fever; neutropenia or thrombocytopenia; anemia or transfused packed red blood cells; nausea, emesis, or diarrhea; dehydration or electrolyte abnormality; malnutrition; constitutional symptoms or nonspecific complications of treatments; and deep venous thrombosis or pulmonary embolus.
A comparison was then made between the two groups based on the following three criteria: Total hospitalizations for all causes, hospitalizations or emergency room visits for adverse effects typically related to chemotherapy, and health care expenditures. All statistical tests were two-sided.
Fifty-one percent of the breast cancer patients in Dr. Hassett's study had at least one hospitalization or ER visit in the year following diagnosis. Chemotherapy recipients were more likely to be hospitalized, or to visit the ER than non-chemotherapy recipients were (61% vs 42%, respectively).
Furthermore, women receiving chemotherapy were more than three times more likely than non-chemotherapy patients to be hospitalized, or visit the emergency room, due to serious adverse effects (16% vs 5%, respectively).
Chemotherapy-Related Serious Toxicity
Fever or infection was the most common chemotherapy-related serious adverse effects associated with hospitalizations or emergency room visits, affecting 8.4% of chemotherapy recipients; neutropenia or thrombocytopenia was the second most common, affecting 5.5% of patients.
The remaining chemotherapy-related serious adverse effects, in descending order, were dehydration or electrolyte disorders (2.5%); nausea, emesis, or diarrhea (2.4%); anemia (2.2%); constitutional symptoms (2%); deep venous thrombosis or pulmonary embolus (1.2%); and malnutrition (0.9%).
Health Care Costs
Chemotherapy recipients incurred larger incremental expenditures than their non-chemotherapy counterparts did. Among chemotherapy recipients, the costs associated with hospitalizations or emergency room visits for chemotherapy-related serious toxicities were $1,681 per person per year.
In contrast, non-chemotherapy patients who experienced such problems incurred a yearly cost of $410 a year; a difference of $1,271.
Ambulatory encounter costs were $17,617 more for chemotherapy recipients than non-chemotherapy recipients.
Most striking to Dr. Hassett and colleagues, they said, were the high costs incurred by patients who experienced chemotherapy-related serious adverse effects. In comparison with non-chemotherapy recipients, hospitalizations, prescriptions, and prescription co-payments cost $12,907, $1,908, and $120 more per patient per year, respectively.
Dr. Erban offered several possible reasons Dr. Hassett's study found a higher frequency of chemotherapy-related toxicities in breast cancer patients than had been previously observed in larger clinical trials.
“Clinical trials may select a healthier population of subjects to study.” In other words, the population studied might not be truly representative of the general population. These clinical trials may have also underreported the events that actually occurred or may have ‘prevented’ some events by more careful attention to the administration of protocol treatment and monitoring.
“Finally, infrequent events may have been reported as zero, when the true confidence interval may have been two to three percent, depending upon the sample size [the so-called rule of 3/n].”
Trial Design: A Closer Look
In their editorial, Drs. Erban and Lau raised some questions about the methodology of the study, specifically the homogeneity of the data in regards to ethnicity and socioeconomic status.
“The current study design seems inadequate to fully assess the true cost of treatment when applied generally to the population,” they said. “The claims data are dependent upon the nature of the insurance coverage and biased by what populations were covered. Thus, the data may have excluded populations who were under-insured, or of a particular socioeconomic group.”
Also of concern was that the researchers did not comment on the question of whether toxicity incidence and reporting may differ among ethnic groups, the editorial said.
Dr. Hassett acknowledged that the study could not take into account all socioeconomic and ethnic variables.
“Only women who had employer-sponsored health insurance were included in the database,” he said. “Women who had no health insurance, or had coverage provided though Medicaid, were not included.” Therefore, poor women were likely not represented in our study.
“Furthermore, the database used for our study did not record race or ethnicity, so we cannot say much about the racial or ethnic distribution of the study cohort or how closely it represents the population as a whole. A prospective, population-based assessment of the frequency and nature of treatment-related adverse effects would be ideal, but may be impractical,” he concluded.
An Important Trial Nonetheless
Despite such methodological issues, Dr. Erban said that the study was very important, especially since clinical trial data only provide approximations of risk and may underestimate the likelihood of a patient experiencing chemotherapy-related serious adverse effects (SAEs).
“Trial data present a likely minimum frequency of side effects, rather than an exact estimate,” he said. “To improve on the current situation we should simply pay more attention on reporting SAEs in a timely fashion and, perhaps, continue reporting on possible adverse effects once the primary study has concluded.”
“Measuring toxicity is much more complex than measuring a single efficacy endpoint,” Dr. Trotti said. “Survival, progression, and response are basically simple binary endpoints, and are easily expressed on a limited scale [0–100%)”.
“But harm is measured on various relative scales, and in aggregate has no clear upper limit, especially in oncology.” More than 1,000 individual adverse event terms—forms of tissue injury or symptoms—are now recognized by the NCI's Common Terminology Criteria for Adverse Events [NCI-CTCAE], and each has four grades. Many patients experience multiple high-grade events and recurring episodes of events.
“Regrettably, we have not fully developed the methods and standards to collect, analyze, and report such complexity. Today, therapeutic gain is by necessity a highly individual impression, often lacking information on the full scope and magnitude of risk,” Dr. Trotti said.
“As cancer treatments become more effective, increasing emphasis is placed on understanding and reducing short- and long-term toxicity burden. The availability of less toxic treatment options—for example, targeted therapies—will make traditional cytotoxic agents less appealing. I think this will help drive further improvements in safety endpoint reporting.”
Patient Monitoring Tools
Dr. Erban gave an example of two tests that focus the use of therapeutic drugs on patients who will benefit from treatment while avoiding side effects in patients who will not.
“Estrogen receptor testing and HER2 gene amplification are two examples of tissue testing that guide therapy for some and eliminate the use of drugs such as tamoxifen and Herceptin for others,” he said.
More complex models—or “patient-specific predictive instruments,” he said, such as the Oncotype DX breast cancer diagnostic assay—may be used to guide complex treatments such as chemotherapy and supportive interventions such as growth factors.
The study by Dr. Hassett and his colleagues convinced Dr. Erban that physicians should be especially vigilant in monitoring patients for long-term adverse events that may result from newer molecular therapeutics, he said.
“In the case of trastuzumab and aromatase inhibitors, the use of these agents over time may change the frequency of rare or infrequent events, and thus careful monitoring is reasonable even for these apparently safe and effective agents. In addition, we are just entering the era of sequential use of novel and effective agents, which is not something easily tracked by current study design.”
Dr. Hassett emphasized that his study did not provide any new data on long-term serious toxicities associated with non-cytotoxic therapies. Rather, the study was solely focused on the short-term adverse effects of cytotoxic chemotherapy medications.
“Whatever the mechanism of action of newer therapies, long-term data for these agents do not exist,” he said.
To better help breast cancer patients make informed decisions about whether or not to undergo chemotherapy, Dr. Hassett recommended that clinicians improve efforts to estimate the frequencies of treatment-related serious adverse effects.
Strengthening standard reporting requirements and improving surveillance of chemotherapy-related serious toxicities during clinical trials might help physicians' present better information to patients about their treatment options.
“One of the main take-home lessons of the Hassett study is that when you apply therapies outside of the highly controlled trial setting you are bound to get somewhat different results,” Dr. Trotti noted. “We've known this to be true of efficacy outcomes for some time, but this is one of the first large studies to document this on the toxicity side.”