Dr. Jeffrey Shuren, Director
Center for Devices and Radiological Health
Food and Drug Administration
Silver Spring, MD 20993
RE: FDA Needs to Provide Breast Implant Patients and Physicians with Unreported Industry Data about Quality of Life, Connective Tissue Symptoms, Rupture Rates per Patient, and Other Complications
Dear Dr. Shuren:
At the FDA Advisory Panel meetings in 2003 and 2005, FDA provided and presented data from the Breast Implant Core studies regarding connective tissue disorder (CTD) signs and symptoms (S/S) and Quality of Life measures. The data indicated poorer outcomes for patients after implants. Neither the previous data nor any follow-up data were presented nor discussed in the FDA’s June 2011 summary nor at the August 2011 meeting of the General and Plastic Surgery Devices Advisory Panel, when those same studies were summarized. Moreover, in direct contradiction to the 2005 FDA scientific summaries of data from the Core Studies, the 2011 FDA summary erroneously stated that there was no evidence of breast implants influencing connective tissue diseases.
We are writing to request that the data reported by FDA in 2005 be included in an updated version of the 2011 summary that is available on the FDA web site, and to ask for an explanation about why these important data were not included in the 8-year and 10-year Core Study data summaries provided to the Advisory Panel or presented in recent summary documents on the FDA web site.
At the August 31, 2011 meeting, Dr. Dana Casciotti asked that question during the public comment period. Dr. Marinac-Dabic told Dr. Casciotti that FDA would respond later that day (pg. 371 of the transcript), but no response was ever provided at the meeting or subsequently.
At the same FDA Advisory Panel Meeting on August 30, 2011, Dr. Diana Zuckerman pointed out that both Mentor and Allergan had reported higher complication rates at 3- and 4-year follow-up according to the FDA analyses of their data in 2005, compared to their cumulative 8- and 10-year complication rates reported in 2011. This makes no sense, since cumulative complication rates should be higher in longer-term studies than shorter-term studies. Dr. Zuckerman also pointed out that rupture rates were mistakenly reported “by implant” rather than by patient in the FDA 2011 summary. Five months later, that error has not yet been corrected.
In this letter, we provide those previously reported data on Connective Tissue Disease Signs and Symptoms, Quality of Life, rupture rates, and other complications from FDA 2003 and 2005 public documents. Information in quotations and italics are quoted from FDA documents, and all data and summaries are based on FDA documents. When available, we also provide data from FDA documents presented in 2011, and show how those comparisons indicate inaccurate reporting in 2011.
Connective Tissue Disease Signs and Symptoms
According to the FDA Summary Panel Memorandum from the FDA review team of the Inamed data, dated March 2, 2005, comparing data from before and two years after getting implants, there was an increase in “5 out of 8 of the CTD S/S [connective tissue disease signs and symptoms] categories for augmentation patients, even after adjusting for age.”
The FDA memo pointed out that Inamed claimed that age accounted for these changes, which the memo stated “is incorrect” (see page 57). That document is on the FDA web site here.
In the memo summary on page 59, FDA concluded (entire section is directly quoted, with key sections highlighted):
“• In evaluating whether the increases in CTD S/S from baseline were related to age, the increases in the following CTD categories occurred despite age:
o Augmentation: Other, Skin, General, Muscle, Joint
o Reconstruction: Other, Skin, General
o Revision: Other, Muscle, Joint.
• The responses to fatigue questions increased by 3% from baseline; however, the prevalence of FM S/S reporting was within the range reported for the U.S. population.
• No statistical associations were found for CTD S/S reporting and implant rupture or patient satisfaction; however, lack of statistical association could have been due to the sample size rather than the lack of a relationship.
• Patients in the MRI Cohort and patients with unresolved complications tended to report higher frequencies of CTD S/S and FM S/S.
• When compared to the patients having undergone saline-filled breast implants, the increase in CTD S/S were not significantly different for patients having silicone gel-filled breast implants.
• Without a control/comparison group of patients without implants followed for the same duration of follow-up and with similar demographic characteristics, the interpretation of these data is difficult.”
In addition, at the April 2005 meeting, FDA panel member Brent Blumenstein, a statistician, indicated that the Inamed analysis was inadequate to evaluate changes in signs and symptoms when age was controlled. According to the FDA transcript, Blumenstein stated, “The model was incorrectly fit….The bottom line is that while there exists data addressing this connective tissue disease signs and symptoms there’s no inference possible at this time. We are still in a state of ignorance with respect to whether there are notable changes in connective tissue disease signs and symptoms following an implant.”
The FDA Summary Panel Memorandum from FDA’s Mentor PMA Review Team, March 2, 2005, is presented in tables and text, as shown below (for original, see http://www.fda.gov/ohrms/dockets/ac/05/briefing/2005-4101b1_Tab-1_fda-Mentor%20Panel%20Memo.pdf )
On Page 62, the FDA memo states “The tables below summarize the cumulative incidence through 3 years of any individual sign/symptom and the two most commonly reported individual sign/symptoms for each indication. Joint pain was one of the top two most commonly reported CTD sign/symptoms for all three surgical indications.”
| Augmentation | Number with sign/symptom reported | Cumulative Incidence |
| Any Sign/Symptom | 53 | 10.3% |
| Numbness of Hands | 13 | 2.7% |
| Joint Pain | 13 | 2.6% |
| Reconstruction | Number with sign/symptom reported | Cumulative Incidence |
| Any Sign/Symptom | 44 | 21.5% |
| Numbness of Hands | 17 | 8.0% |
| Joint Pain | 10 | 4.5% |
| Revision | Number with sign/symptom reported | Cumulative Incidence |
| Any Sign/Symptom | 39 | 21.3% |
| Numbness of Hands | 14 | 7.3% |
| Joint Pain | 11 | 6.0% |
Although the tables do not compare pre and post-tests, on page 63 the FDA team discusses how the incidence increased significantly:
“For augmentation patients, the following CTD signs/symptom increases from baseline were statistically significant and, therefore, were due to reasons other than aging: fatigue, exhaustion, joint swelling, frequent muscle cramps, joint pain, combined fatigue, combined pain, and combined fibromyalgia. Generalized aching was nearly statistically significant at p=0.0641. For reconstruction patients, the GEE analysis yielded no statistically significant results. Therefore, for the reconstruction population, it can be concluded that the increases noted in CTD signs/symptoms from baseline were due to aging. The symptom of joint pain, however, was nearly significant at p=0.0579 for reconstruction patients. For revision patients, the GEE analysis results indicated statistically significant findings for fatigue, generalized aching, and combined fatigue. Therefore, for this population, the increases from baseline in fatigue and aching were due to reasons other than aging.”
Although the FDA did not mention it in their 2005 memorandum, the lack of statistical significance for the Mentor reconstruction sample may have been related to the smaller sample size. According to the table on page 51, only 251 reconstruction patients were included in the study, less than half the number of augmentation patients (551 women).
Statistician Brent Blumenstein presented additional analyses on Mentor augmentation patients for the FDA meeting and concluded that the Mentor data were superior to previous CTD studies on breast implants, saying (from the FDA transcript for April 13, 2005) “What we have here is a well-designed study with subjects serving as their own control. And that these findings are consistent, that is the specific list of signs and symptoms and classes of signs and symptoms are consistent with our a priori expectations, and we have adjusted for age. And also, it is important to remember that these signs and symptoms changes track well with some of the quality of life changes that we have seen. So, in my opinion, I find these results disquieting.” Dr. Blumenstein suggested FDA require additional studies, advising “Find patients who have these high symptom scores and study these women, plus a sample of women with low scores, in a supplemental study. You might even think about assaying tissues if you could get that. Assess their lifestyles….put them under intense follow-up. Find out what is going on with these women, because there is something going on here.”
It is worth noting that of the several men and women on the published list of 2011 Advisory Panel members who had participated in previous FDA Advisory Panel meetings on breast implants, Dr. Blumenstein was the only member who had consistently recommended against approval at the 2005 meeting. Despite being listed as a Panel member for the 2011 meeting, however, Blumenstein was apparently excluded from that meeting. His exclusion may help explain why these issues of missing Signs and Symptoms data and Quality of Life data were not raised at the 2011 Panel meeting. His exclusion also raises questions about bias in the final roster of 2011 panel members.
Quality of Life
Despite claims that breast implants would improve self-esteem and quality of life, most of the data presented in 2005 FDA summaries do not support those claims.
In summary, for Inamed augmentation patients, 12 quality of life scores differed significantly in the pre-test and post-test. Nine of the 12 (75%) were worse in the post-test. For Inamed revision patients, 9 of 9 (100%) that differed significantly were worse in the post-test. For reconstruction patients, only two scores were significantly different in the post-test, and both showed improvement in physical functioning, which probably reflects the fact that many of these women were being treated for breast cancer at the pre-test and their quality of life was better as cancer survivors two years later.
The FDA summary points out that for both Mentor and Inamed, the patients showed higher scores on quality of life than the general population, in the pre-test and post-test. This indicates that the women who had breast implant surgery are not generally women suffering from low self-esteem or poor quality of life.
Inamed (Allergan)
Here are the details from the FDA Summary Panel Memorandum from FDA’s Inamed PMA Review Team, March 2, 2005 (http://www.fda.gov/ohrms/dockets/ac/05/briefing/2005-4101b1_tab-1_fda-Inamed%20Panel%20Memo.pdf)
“With respect to the Health Status Questionnaire (SF-36 and MOS-20), the core augmentation cohort….There were small, statistically significant declines in some subscales of these measures in breast implant recipients over time. However, the 2-year values for the augmentation cohort were generally numerically higher than normative values for the general female population” (page 71).
Although the FDA summary does not mention it, most of the significant differences showed lower scores on quality of life in the post-test. Nine of 12 were worse for augmentation patients and nine of 9 were worse for revision patients.
Quality of Life measures include the SF-36, which measures 8 health concepts: physical functioning; role-physical; bodily pain; general health; vitality; social functioning; role-emotional; and mental health. The 8 scales can then be collapsed into two summary scales with the first 4 scales comprising the Physical, and the last 4 scales comprising the Mental Health.
Inamed Augmentation Patients
All Statistically Significant Changes are as follows:
- SF-36 Role Emotional: Significantly worse in post-test
- SF-36 Role Physical: Significantly worse in post-test
- SF-36 General Health: Significantly worse in post-test
- SF-36 Social: Significantly worse in post-test
- SF-36 Vitality: Significantly worse in post-test
- SF-36 Mental Health: Significantly worse in post-test
- MOS-20 Health Perceptions: Significantly worse in post-test
- MOS-20 Mental Health: Significantly worse in post-test
- Tennessee Self-Concept Scale: Physical Self: Significantly better in post-test
- Body Esteem-Total Score: Significantly better in post-test
- Body Esteem-Sexual Attractiveness: Significantly better in post-test
- Body Esteem-Physical Condition: Significantly worse in post-test
- Scores on the Rosenberg Self Esteem Scale were worse in the post-test, but the difference was not statistically significant.
Allergan Reconstruction Patients
- SF-36 Role Physical: Significantly better in post-test
- MOS-20 Physical Functioning: Significantly better in post-test
Inamed Revision Patients
- SF-36 Role Emotional: Significantly worse in post-test
- SF-36 General Health: Significantly worse in post-test
- SF-36 Social: Significantly worse in post-test
- Mental Health: Significantly worse in post-test
- MOS-20 Health Perceptions: Significantly worse in post-test
- MOS-20 Mental Health: Significantly worse in post-test
- Tennessee Self-Concept Scale Physical Self: Significantly worse in post-test
- Rosenberg Self-esteem Scale: Significantly worse in post-test
- Body Esteem-Physical Condition: Significantly worse in post-test
Mentor
Below are the data from the FDA Summary Panel Memorandum from FDA’s Mentor PMA Review Team, March 2, 2005 (http://www.fda.gov/ohrms/dockets/ac/05/briefing/2005-4101b1_Tab-1_fda-Mentor%20Panel%20Memo.pdf)
Similar to the Inamed findings, when there were statistically significant changes from pre-test to post-test for Mentor patients, almost all were worse in the post-test compared to the pre-test. For augmentation patients, scores on physical health and mental health were significantly worse, scores on the Rosenberg self-esteem scale were better, and there was no change on the Tennessee self-concept scores or body esteem scale. For revision patients, scores on physical health, mental health, body esteem and Tennessee self-concept scale all were worse in the post-test, and there was no change in the Rosenberg self-esteem scale. No scores were better in the post-test. For reconstruction patients, there were no significant changes on any of the scales.
The data below are not as detailed as the Inamed data, because the FDA memo did not provide as much specific information. However, it includes differences in scores that were provided by the FDA.
Mentor Augmentation Patients
- Physical Health: Significantly worse in post-test (1.0)
- Mental Health: Significantly worse in post-test (1.1)
- Tennessee self-concept scores: No significant change
- Body Esteem scale: No significant change
- Rosenberg Self-Esteem Scale: Significantly better in post-test (0.6)
Mentor Reconstruction Patients
- Physical Health: No significant change
- Mental Health: No significant change
- Tennessee Self-Concept Scale: No significant change
- Body Esteem Scale: No significant change
- Rosenberg Self-esteem Scale: No significant change
Mentor Revision Patients
- Physical Health: Significantly worse in post-test (1.8)
- Mental Health: Significantly worse in post-test (2.5)
- Tennessee Self-Concept Scale: significantly worse in post-test (6.6)
- Body Esteem Scale: significantly worse in post-test (5.0)
- Rosenberg Self-esteem scale: no significant change
FDA also noted the following about the literature review on Quality of Life information (provided by Mentor):
- Page 70: “…the literature does not provide strong scientific support that breast implants have measurable psychological and psychosocial benefits for women seeking breast augmentation.”
- Page 73: “Literature that adequately evaluates the short-term or long-term psychological or psychosocial benefits of breast implants as a reconstructive procedure utilizing appropriate control group was not provided by Mentor.”
Cumulative Complication Rates
Allergan Patients at 3 Years and 10 Years
The long-term safety data on breast implants provide data on the cumulative complication rates for a wide range of risks, including health problems and cosmetic complications. These data provide important information that patients can use to determine if the risks of implants outweigh the benefits. Either health risks or cosmetic complications can result in additional surgeries to remove and/or replace breast implants.
At the August 2011 FDA Advisory Panel meeting, the FDA presented data on 8-year complication rates for Mentor and 10-year complication rates for Allergan. Rates at 3 and 4 years had previously been made available at the FDA’s 2005 Advisory Panel meetings, but were not presented in 2011 at the FDA meeting or in the FDA’s summary report. However, the earlier data are still available on the FDA web site, and when we compared the complication rates, we discovered that numerous reported complication rates decreased over time for both companies’ data. This shows problems with the data, since the complication rates are reported to be cumulative and should therefore stay the same or increase over time.
The 3-year complication rates for Inamed/Allergan were reported in this FDA official memorandum in 2003: http://www.fda.gov/ohrms/dockets/ac/03/briefing/3989b1_clinical-summary-memo.pdf
For Allergan Primary Augmentation Patients, the following complications were reported to be lower at 10 years (N=455) than at 3 years (N=494). That raises questions about the accuracy of reporting. The FDA reported that the samples sizes decreased because some patients died and some withdrew from the study. However, in order to maintain accurate reports of complications, patients who reported problems earlier in the study should not be eliminated from the later sample. We ask that the FDA explain whether some of the patients with earlier complications were excluded from the 10-year sample, and if so, under what circumstances and how did that affect complication rates? Accuracy problems are more obvious when the reported percentages decreased over time, but even those where the percentages increased may be inaccurate.
Swelling: 23% went down to 9%
Scarring: 8% went down to 4%
Seroma: 3% went down to 2%
Ptosis: 3% went down to 2%
For Allergan Primary Reconstruction Patients, the following complications were reported to be lower at 10 years (N=98) than at 3 years (N=221). That again raises questions about the accuracy of reporting, and whether patients with complications were excluded from the 10-year sample.
Swelling: 16% went down to 7%
Redness: 6% went down to 2%
Seroma: 5% went down to 2%
Malposition: 5% went down to 2%
Bruising: 4% went down to 1%
Skin Rash: 3% went down to 2%
Ptosis: 1% went down to 0%
Mentor Patients at 3 Years and 8 Years
For Mentor Primary Augmentation Patients, the following complications were not reported at 8 years but had been reported at 3 years, for a sample described as 551 patients in 2005 and in 2011. Since all complications experienced by 1% or more of the sample were supposed to be reported, the absence of these reported complications at 8 years raises questions about the accuracy of reporting, and whether patients with complications were excluded from the 8-year sample even though the sample sizes reported were identical.
Hypertrophic Scarring: 6% at 3 years, not reported at 8 years
Ptosis: 2% at 3 years and not reported at 8 years
The above 3-year Augmentation Patient data are reported in the 2005 FDA Summary Panel Memorandum on page 52.
http://www.fda.gov/ohrms/dockets/ac/05/briefing/2005-4101b1_Tab-1_fda-Mentor%20Panel%20Memo.pdf . The below 3-year Reconstruction Patient data are on page 53.
For Mentor Primary Reconstruction Patients, the following complications were not reported at 8 years (N=251) but had been reported at 3 years (N=251). That again raises questions about the accuracy of reporting, and whether patients with complications were excluded from the 10-year sample even though the reported sample size is identical.
Hematoma extrusion: 2% at 3 years and not reported at 8 years
Necrosis: 1% at 3 years and not reported at 8 years
Asymmetry: 7% at 3 years and not reported at 8 years
Ptosis: 7% at 3 years and not reported at 8 years
Hypertrophic scarring: 6% at 3 years and not reported at 8 years
Redness and skin rash were reported for Allergan but not for Mentor.
In addition to the clear errors, where the cumulative percentage reporting a complication decreased or disappeared from 3 years to 8 years, these problems raise questions about the accuracy of all the 8-year data, even those where the percentages increased. For example, while all experts agree that asymmetry increases over time, especially for reconstruction patients, Allergan asymmetry increased from 15% at 3 years to only 23% at 10 years. Other increases in complications were also rather modest compared to patient reports: Mentor reconstruction patient reoperations increased from 26% at 3 years to 39% at 8 years and removal from 13% to 23%. Would those numbers have been higher if the reporting was more accurate, or if patients who had problems had not been dropped from the final sample? For example, the Allergan 10-year sample is less than half the size of the 3-year sample. Are women who had their implants removed being removed from that sample and therefore biasing the results? These questions are important for the Inamed and Mentor data.
Rupture Rates
On page 10 of the FDA memorandum dated June 2011 (linked above), the implant rupture rate is reported per implant. Those same percentages are included in Table 4 (page 53), where all the complication rates, including rupture data seem to be reported per patient, since the sample sizes are listed as the number of patients not the number of implants. This is incorrect, since the percentage of women with ruptured implants is much higher – approximately twice as high – as the percentage of ruptured implants.
Based on the MRI cohort in the 2011 summary, the Allergan rupture rate per implant is 10% of augmentation implants and 27% of reconstruction implants. No rupture rates per patient are reported. At 4 years, the Allergan rupture rate per patient was almost exactly twice the rupture rate per implant, according to the March 2, 2005 FDA summary of Inamed (see link on page 3 of this document, pages 21-22). Similarly, the Mentor rupture rate at 8 years was 14% per implant for augmentation patients and 14% for reconstruction patients, but no data per patient were reported. At 3 years, the Mentor rupture rate per patient was approximately twice that per implant, according to the March 2, 2005 FDA summary of Mentor (see link on page 5 of this document, page 77).
The use of per implant rupture rates is misleading for patients. When a patient’s implant ruptures, surgery is required to remove it, regardless of whether one implant is ruptured or both are ruptured. The risks of surgery and costs to the patients are very similar whether one or two implants are involved. Therefore, patients want to know what the chances are of having a ruptured implant, not the percentage of implants that rupture. It is certainly important for women considering implants to know if the rupture rate is more than half the primary reconstruction patients and 20% of primary augmentation patients at 10 years.
An accurate rupture rate per patient is especially important in light of the FDA panel’s discussion about how often women with silicone implants should be advised to undergo breast MRIs to check for leakage. It is widely acknowledged that most women do not undergo regular MRIs to check for breast implant leakage, but one would expect that to change if patients (especially reconstruction and revision patients) realized how likely it was for their implants to rupture within 10 years.
The importance of per patient rates is true for all complications. Patients care whether they will have breast pain more than they care whether the pain will be in one or both breasts. If complications occur that require surgery, that is an important piece of information, whether or not one or both implants are involved. We ask FDA to clarify whether each of the complication rates reported is per implant or per patient.
The somewhat lower rupture rates for Mentor may reflect a better product or may reflect the different time frames for the analysis (3 vs. 4 years, 8 vs. 10 years). Or, given other problems with the Mentor data, the data differences raise questions about the accuracy and generalizability of the data because of the high drop out rate.
Summary
In conclusion, we are concerned about missing information and inaccurate or misleading data that were provided to the FDA Advisory Panel on breast implants and that are included (or missing) in the June 2011 summary on the FDA web and in other recent FDA materials.
- Missing information about signs and symptoms of connective tissue disease from 2005 and the most recent follow-up data, and an inaccurate summary of earlier findings
- Missing information about poor outcome on Quality of Life measures from 2005, and missing follow-up data
- Inaccurate cumulative complication rates for at least some outcomes
- Inaccurate and misleading presentation of rupture data, which should be presented per patient rather than per implant.
- Questions about which complications are presented per implant or per patient and questions about whether women with complications at 3 or 4 years were excluded from the 8- and 10-year follow-up samples.
We expressed concern about these omissions and errors during the FDA Advisory Panel Public Comment Period in August 2011, and are very disappointed that the FDA has not corrected any of these problems in the last 5 months. We would appreciate a response to this letter in writing, and also ask for a meeting to discuss these issues.
Sincerely,
Diana Zuckerman, Ph.D.
President
Cc: The Honorable Barbara Boxer
The Honorable Rosa DeLauro
The Honorable Dianne Feinstein
The Honorable Olympia Snowe