Jae Hong Lee, MD, MPH, National Center for Health Research, July 9, 2002
Two years after the FDA’s controversial approval of saline breast implants, the FDA Advisory Committee on breast implants held an unusual meeting to review the latest research on implant complication rates on July 9, 2002.
In 2000, saline breast implants were approved by the FDA despite a 70% rate of serious complications among breast cancer reconstruction patients during the first three years and 40% among augmentation patients. The Advisory Committee expressed strong concerns about the high complication rate but recommended approval on the condition that the manufacturers conduct long-term safety studies.
Meanwhile:
• The number of teens getting saline implants has more than doubled.
• The number of women getting breast augmentation is at an all-time high
(more than 200,000 in 2001).
• Government-sponsored research has shown an increased risk of cancer and
deaths among women who had saline or silicone breast implants for more than
8 years.
A 4-year criminal investigation of one of the manufacturers has attracted Congressional attention.
This following testimony was presented at the public comment period at the Advisory Committee meeting on Tuesday, July 9, 2002. [Dr. Lee’s testimony has been slightly modified after the hearing to reflect additional information he received about the manufacturers’ data that was not available before the statement was given].
When physicians are asked about the safety of breast implants, many — perhaps most — will focus their thoughts on the reported systemic illnesses that spurred significant controversy and not a small amount of litigation about a decade ago. An Institute of Medicine (IOM) report — Safety of Silicone Breast Implants — was commissioned by Congress to examine the issue and concluded that insufficient evidence existed to implicate silicone breast implants as a cause of these systemic illnesses. Many physicians and surgeons now misleadingly and inappropriately cite the study as definitive proof that breast implants are “safe.”
A more careful reading of the IOM report reveals something quite different. Rather than a comprehensive declaration that silicone breast implants were safe, the report noted that the “frequency of local and perioperative complications has been substantial in both augmentation and reconstruction of the breast” and that reoperations and local complications were “the primary safety issue with silicone breast implants.”
This issue remains important for several reasons. First, more and more women are receiving breast implants. According to statistics from the American Society of Plastic Surgeons (ASPS), more than 219,000 breast augmentation procedures were performed in 2001 in the United States alone. In the last two years, over 6,000 girls 18 years old or younger have undergone breast augmentation. There may now be over 2 million American women with silicone breast implants.
Second, breast augmentation is not a life-saving or life-sustaining procedure. In order to justify a completely elective cosmetic procedure, the risks to the patient must be clearly outweighed by the benefits.
Third, the cost of reoperations and local complications associated with breast implants can be significant for both the patient and society. Many health plans do not cover the medical costs associated with cosmetic breast implant complications — forcing many patients to cover these costs out-of-pocket. For example, Aetna — one of the nation’s largest health insurers — does not cover the cost of removing ruptured saline implants in patients who have previously undergone cosmetic breast augmentation. It should be noted that in 2001, more than 62,000 breast implant removal procedures were performed in the U.S.
The Inamed and Mentor Post-approval Studies
In light of the serious concern regarding reoperations and local complications, the post-approval studies recently released by Inamed and Mentor are extremely disappointing. Both studies are poorly designed and executed, and most likely underestimate the cumulative incidence of complications. I will discuss just a few of the more glaring weaknesses of the studies.
A major problem shared by both the Inamed and Mentor studies is the failure to follow-up on patients who have had all their breast implants removed prior to the 3 or 5-year study interval. One can argue that these patients — among the earliest to have all their implants removed — are the most important group of patients to study. Excluding follow-up data from these patients introduces an unacceptable post-entry exclusion bias into both studies. Since it seems likely that these patients were experiencing complication rates higher than those who did not have all their implants removed, the overall effect of this exclusion bias may be to seriously underestimate the cumulative incidence of certain complications — e.g. additional operations performed to improve the frequently unsatisfactory aesthetic result after complete breast implant removal.
It is also legitimate to question the use of Kaplan-Meier risk estimates in both studies. One essential condition for using the Kaplan-Meier method is the assumption of independence between censoring and outcome.1 In other words, one must assume that the rate of local complications and reoperations in the excluded patients was similar to those who remained in the study. Since the excluded patients had all of their initial implants removed early — presumably due to complications in nearly all cases — this is not a reasonable assumption to make. Using the Kaplan-Meier method in this situation clearly results in an underestimation of the cumulative complication rate, especially as time progresses and the more complication-prone patients and implants are removed from the study cohort — a fundamental flaw in both studies. It will be essential to compare the characteristics of the patients and implants who remain in the studies beyond 5 years to those who had all their implants removed at earlier time intervals.
When Inamed conveniently excludes the breast implant-removal patients from further follow-up, they inflate their follow-up rate for patients. For example, they calculated an “expected follow-up” reconstruction cohort of 175 by excluding 11 patients who had died and 51 patients who had all their initial implants removed. As a result, their follow-up rate for reconstruction patients was reported as 80% (or 140 out of 175). As mentioned earlier, it is highly inappropriate to completely exclude those implant-removal patients from the study cohort. If those patients are left in, the expected follow-up cohort consists of 226 patients — not 175 — and the actual follow-up rate for reconstruction patients is a clearly inadequate 62%.
There are also signs that the Inamed study is statistically underpowered. For example, Inamed suggests that because most of the confidence intervals between 5-year and 3-year risk rates overlap, the true rates may not be higher at 5 years. They imply that it may not be necessary to follow complications with overlapping 3 and 5-year confidence intervals in subsequent years. A more reasonable interpretation of this data is that the study is too small and underpowered. Since even a small difference of 3% represents over 6,000 breast augmentation patients, the post-approval studies must be large enough to detect even small differences in risk rates.
As bad as the Inamed study is, the Mentor study is even worse. The Mentor report acknowledges that “low response rate and the differences between the responders and the non-responders is a major limitation” of their study. I would go further and state that most of the data presented in the Mentor study is scientifically worthless due to the disturbingly low follow-up rates.
Perhaps the only compelling result in the Mentor study is their Focus Group data. They reported that breast “augmentation women expressed a degree of cynicism regarding the extensive list of complications” found in the patient brochure and “concluded that the occurrence of complications was exaggerated and not to be taken at face value.” This finding strongly suggests that informed consent is not an adequate solution for the unacceptably high complication rates reported for saline breast implants.
Unjustifiable Complication Rates
Setting aside for the moment the very serious methodological problems of both studies, we find little comfort in the reported data. Even taken at face value, the complication rates are unacceptably high. For instance, Inamed reports 5-year reoperation rates of 26% for augmentation patients and nearly 45% for reconstruction patients. As a physician with surgical training, I find these complication rates to be completely unjustifiable for a purely elective cosmetic procedure.
What Should Be Done?
The poor quality of the studies clearly indicate that both Inamed and Mentor have failed to fulfill all the conditions of their premarket approval agreement. Furthermore, the reported complication rates in both studies reinforce the IOM’s concern about local complications and reoperations. These companies seemingly lack the motivation to perform adequate and appropriate safety studies as long as they can market and sell their products as “FDA approved.”
So what should be done?
At the very least, the FDA should require a black box warning on all advertisements noting that the manufacturer has not adequately completed required 5-year safety studies. The panel should also recommend that patient consent forms include a statement noting the lack of reliable long-term safety data. Another option would be to temporarily suspend general distribution of these implants until better safety data become available.
The Inamed and Mentor studies leave many questions unanswered but do make clear one important fact: right now, there are no adequate data demonstrating the safety of saline breast implants.
1 Szklo M and Nieto FJ. Epidemiology: Beyond the Basics. Aspen Publishers, Inc. Gaithersburg, MD. 2000
Jae Hong Lee, MD, MPH, is the Senior Medical Policy Analyst of the National Center for Health Research. Dr. Lee has professional experience in basic science research, clinical patient care, clinical research, and public health, and works on a variety of health issues for women and families.
Dr. Lee’s work in research began in college, where he performed studies in psychology and electron microscopy. During medical school, he pursued two years of full-time basic science research in neurology and pathology under fellowships from the American Heart Association and the NIMH/Measey Foundation. After completing a surgical internship at Yale, Dr. Lee performed clinical research for five years at UCLA, focusing his studies on traumatic brain injury and cerebrovascular disease. His work has been published in respected peer-reviewed journals and presented at national and international meetings.
Dr. Lee received his Bachelor of Arts degree with a major in biophysics from The Johns Hopkins University. His medical degree is from the University of Pennsylvania School of Medicine. Most recently, he completed a Master of Public Health degree with a concentration in quantitative methods at the Harvard University School of Public Health.