Study: Breast cancer implant study suggests links with illness but has serious flaws

IN-DEPTH REVIEW OF RESEARCH

Study background:

In the early 1990s, numerous anecdotal and case reports of silicone breast implants causing connective tissue disorders, cancer and autoimmune diseases resulted in a voluntary moratorium of their use. Subsequently, a new generation of silicone implants were approved in 2006.

As a condition of that approval, the FDA ordered breast implant manufacturers Allergan and Mentor to follow implant patients for 10 years to identify common complications and rare events. The FDA shared results of these studies, which involved nearly 100,000 patients in a report published in 2011.

In addition to the FDA’s 2011 report, an extensive literature review in 2016 by Balk and colleagues of more than 5,000 citations concluded that case reports and anecdotes did not conclusively associate breast implants with any health outcomes, including connective tissue disease, cancer, neurological disorders or reproductive problems.

Researchers of this study wanted to know:

The long-term safety and efficacy of breast implants.

Study design:

The subset of data used for this study include information from the following patients from the original post-approval studies:

• Silicone implants: 41,342 patients from Allergan; 41,975 patients from Mentor
• Saline implants: 15,646 patients with Allergan; 1,030 patients from Mentor

Allergan patients with Natrelle® silicone gel or saline implants were recruited from February 2007 to March 2010 from 873 sites.  Mentor patients with saline or MemoryGel® silicone implants were recruited from February 2007 to July 2009 from 2,342 sites. Collected clinical outcome data included deaths, rare adverse events (e.g., connective tissue disease) and local complications. For each manufacturer, participants were assessed at 3 to 4 postoperative visits at predefined intervals, and by annual remote (mail, telephone, internet) patient questionnaires. 

back to top

Study findings:

The conclusions of this study cannot be deemed as scientific or reliable, based on the way data was collected, the analysis method, and the overall bias involved.

Compared to the general population, patients with silicone implants had a statistically significant, more than double incidence of the following conditions:

• Rheumatoid arthritis (observed only in Mentor implants)
  • 0.05% per person per year in the general population
  • 0.3% per person per year in women with implants (nearly 6-fold more)
• Sjorgen syndrome (observed only in Mentor implants)
  • 0.007% per person per year in the general population
  • 0.057% per person per year in women with implants (nearly 8-fold more)
• Scleroderma (observed only in Mentor implants)
  • 0.006% per person per year in the general population
  • 0.04% per person per year in women with implants (7-fold more)
• Melanoma (observed only in Mentor implants)
  • 0.021% per person per year in the general population
  • 0.08% per person per year in women with implants (nearly 4-fold more)

For patients with silicone implants, incidence rates were statistically significant, but less than double that of the general population for the following:

• Overall cancer diagnosis (in patients with either Mentor or Allergan implants):
  • 0.41% per person per year in the general population
  • 0.16% per person per year in women with Allergan implants (less than the general population)
  • 0.64% per person per year in women with Mentor implants (1.5-fold more often)
     
• Neurological disorder (in patients with either Mentor or Allegan implants)
  • 0.23% per person per year in the general population
  • 0.036% per person per year in women with o with Allergan implants (less than the general population)
  • 0.36% per person per year in women with Mentor implants (1.5-fold more often)
     
• Multiple sclerosis (observed only in Mentor implants)
  • 0.25% per person per year in the general population
  • 0.4% per person per year in women with Mentor implants (1.7 fold more often)
     
• Myositis (observed only in Mentor implants)
  • 0.008% per person per year in the general population
  • 0.15% per person per year in women with Mentor implants (1.8-fold more often)
     
• Regardless of the manufacturer, women with silicone implants had lower rates of fibromyalgia. Patients with Mentor implants had lower rates of lung cancer.

back to top

Limitations:

There are important and significant problems with this study.

The data used was inadequate for testing rare events.

The authors of this study performed their analysis on data that was known to be problematic. Although their original post-approval studies enrolled a sufficient number of patients (99,993), other experts warned that using physician-reported or patient-reported data was inadequate, especially for studying rare events, and should not be used to make statistically significant conclusions. The study follow-up was also poorly conducted.

Only a subset of the data was evaluated.

Ethical, accurate study procedure calls for objective analysis of all data in a study. Yet, the authors of this study “cherry-picked” only certain data to include and subsets of the results data to calculate rates of adverse effects. This “picking and choosing” of data skews observations and results. It is unacceptable scientific process without the standard stringent internal validation.

A poorly matched control group was used.

The number of adverse events in this study’s subset of patients was compared to controls.  The authors chose the general population rates reported in the literature as controls, rather than control participants who participated in this study. From a scientific perspective, this is highly irregular, since the study test group and the control group from the general population may differ in ways (e.g., environmental exposure to chemicals or differences in ethnicity) that are not related to the presence or absence of breast implants. A more reliable control would be a comparison of rates in the same study of harmful events between women with silicone implants and women with saline implants. In fact, the authors had this data, but they did not use it.

Different data collection methods were used for different test groups.

This study violated another basic rule of scientific methodology: that the method of data collection should be the same among groups that are compared. This study’s comparison of Allergan and Mentor data was highly inappropriate, since they were collected by different methods. Study authors used rates of adverse events that were calculated from physician-reported data in the Allergen study of 25,219 patients. In contrast, rates of adverse events were calculated from patient-reported data (without physician confirmation) for the Mentor study of 8,437 patients. Using patient-reported data for connective tissue diseases or other conditions without clinical confirmation is unreliable. These diseases can vary considerably in presentation and may require physician expertise or laboratory tests to confirm a diagnosis.

Poor follow-up rates increase the likelihood of biased responses.

Rates of patient follow-up were poor. The 2-year follow-up data included 60.5% of Allergan patients, but just 24% of Mentor patients. Even fewer (20%) of women participated in the 7-year follow-up. Poor follow-up rates like these can introduce substantial bias. One reason is that patients without symptoms or concerns are less likely to follow-up, while patients with symptoms or concerns may be more likely to keep their follow-up visits and/or respond to questionnaires. Despite this obvious shortcoming, which was worse with the Mentor study, the authors heavily base their conclusions on that data.

The potential for statistical error exists.

Because large data sets inherently have variations, a subset of data can almost always be randomly found to fit a given pre-formed hypothesis. Objective studies do the opposite: they collect and analyze reliable data and come to objective conclusions.

In one of the two accompanying editorials, Amy S. Colwell, MD, and Babak Mehrara, MD, state that:

“This [type of statistical error] may arise from deliberate academic dishonesty or as an honest mistake. Type I [statistical] errors have a pernicious effect because they cause patient/physician anxiety and lead to medical errors and flawed conclusions.”

Nor did the current analysis evaluate other patient data such as geography, lifestyle, UV light exposure, smoking, alcohol use, cancer history, weight, depression, medications or reproductive history that might impact this study’s outcomes. Failing to control for the potential of statistical and other confounding factors means the statistical conclusions of the current study are untrustworthy.

Results of the study are illogical.

For unknown reasons, the current study emphasizes the patient-reported data from 8,437 Mentor patients, while virtually ignoring the physician-reported data from the larger group of 25,219 Allergan patients:

• While the authors emphasize the increased risk of rheumatoid arthritis in patients with Mentor implants, they ignore data that suggests a protective effect in patients with Allergan implants.
   
• The authors also emphasize an increased rate of scleroderma and Sjogren’s syndrome in patients with Mentor implants, again ignoring their data showing no reported cases of scleroderma or Sjogren’s syndrome in patients who have Allergan implants.
   
• The authors emphasize the statistically increased risk of neurological disorders for patients with silicone breast implants.  However, they fail to highlight that while Mentor implants are associated with a statistically significant increased risk, Allegan implants are associated with a statistically significant decreased risk. This suggests that Allergan implants, but not Mentor implants are protective for neurological disorders.
   
• Mentor implants were associated with increased melanoma risk while no cases of melanoma were reported in patients with Allergan implants.

Binita S. Ashar, MD, FDA Divisional Director and author of the other editorial that accompanied this study, states:

“Unfortunately, as the authors acknowledge, they failed to account for methodological differences between studies, inconsistencies in the data, differential loss to follow-up, confounding and other potential sources of bias.”

Conflicts of interest are not reported

Importantly, the authors report no financial conflicts of interest. Annals of Surgery conforms to the disclosure policy published by the International Committee of Medical Journal Editors, instructing authors to “report all sources of revenue paid (or promised to be paid) directly to you or your institution on your behalf over the 36 months prior to submission of the work.” According to the Propublica online database Dollars for Docs, Mark W. Clemens, MD and senior author of the study received $64,013 from Allergan in 2016 for promoting its Natrelle and Seri products. In 2015, he received $134,782 from Allergan. Awareness of possible bias is essential for proper evaluation of study conclusions.

back to top

Conclusions:

Social media often amplifies fears about breast implants.  As a result, increasing numbers of women see plastic surgeons for evaluation of systemic symptoms of “breast implant illness” that they attribute to their breast implants. However, breast implant illness is not a medical diagnosis. Currently, no simple test confirms the relationship between implants and a patient’s symptoms. Despite the claims of the authors of this scientifically flawed study, no sufficient scientific evidence supports a link between breast implants and autoimmune-related disorders or other diseases. Breast implants (primarily implants with textured surfaces) are linked by evidence to BIA-ALCL, although this is extremely rare. Other known, non-disease risks associated with breast implants include capsular contracture, reoperation, removal, and implant rupture. These remain the most common problems that occur with breast implants.

Posted 2/21/19

(back to top)