Comments on FDA Safety and Innovation Act Section 907 Report

National Center for Health Research, November 20, 2013

Division of Dockets Management
Food and Drug Administration
5630 Fishers Lane, Room 1061 (HFA-305)
Rockville, Maryland 20852

Comments of the National Research Center for Women & Families and the Cancer Prevention and Treatment Fund
on FDA Safety and Innovation Act Section 907 Report

Docket no. FDA-2013-N-0745-0001

The National Research Center for Women & Families strongly supports the requirement of the Food and Drug Administration Safety and Innovation Act (FDASIA) for an action plan to include demographic subgroups in clinical trials and data analysis. Greater diversity in clinical trials, analyzing subgroup data, and reporting the results and explaining the implications in product labels and MedGuides will shed light on which medical products are safe and effective for which demographic subgroups, including racial and ethnic minorities.

The Section 907 report, which outlines the recent status of demographic subgroup’s inclusion in clinical trials and data analysis, creates a baseline measurement that indicates much room for improvement. Although there have been improvements in the last decade, this report is replete with examples (described in detail below) where lack of adequate demographic subgroup reporting not only hindered this analysis itself, but also obscures medically important information. Even when subgroup information has been collected, if crucial subgroup data are not explicitly included on labels, providers and patients may incorrectly assume that “no news is good news,” when in reality the drug or device may not have been adequately tested or analyzed for their subgroup.

Our major criticisms of the report are that, as reflected in the Executive Summary:

  1. The inclusion of demographic subgroups is considered a measure of success, rather than acknowledging the importance of having large enough subgroups to analyze separately.
  2. The lower prevalence of disease in those subgroups is given as justification for inadequate numbers of patients even when those diseases are very prevalent.
  3. The summary optimistically states that “We also found that FDA shares this information with the public in a variety of ways.”  We are disappointed with the small amount of useful information that is made public, particularly on the label, which is the easiest route to communicate with doctors and patients.

We respectfully urge the FDA to incorporate our comments and recommendations into the Action Plan.

On the matter of inclusion, although demographic subgroups were frequently included in the clinical trials described in the report, they were rarely analyzed separately. In many cases, their numbers were too small to be meaningfully analyzed as separate subgroups. There is little value in including subgroups if they are not analyzed separately to produce useful information on the safety and effectiveness of the products for these subgroups. Including subgroups that are so small that their data cannot be meaningfully analyzed renders their participation useless from a scientific point of view. In addition, the report sometimes generalizes the findings in ways that are not precisely accurate; for example, many studies had very limited diversity and yet the Executive Summary states that “For approved drugs and biologics, the extent to which patients were represented in clinical trials by age and sex tended to reflect the disease indication studied.”

In addition to minimizing the importance of subgroup analysis, that statement ignores the fact that there are many examples where a subgroup has a lower incidence of the disease, but also has much worse outcomes when they get the disease. Breast cancer is one such example: the disease is less prevalent among African Americans, but African American women tend to get a more aggressive form of the disease and their survival rate is lower.

If a treatment is substantially less effective in a particular subgroup, and that subgroup is too small to analyze separately, the results of the study will not provide the crucial information to warn that subgroup that the medical product is ineffective. On the other hand, if a treatment is more effective for some subgroups that are not analyzed separately, that crucial information would also be unavailable.

For example, all (100%) of the patients in the melanoma trials were white (See Table 1-3), but many reports before 2011 indicate an increasing incidence among non-white populations. These include a 20% increase in melanoma among Hispanic men and a 60% increase among non-Hispanic black women in Florida.[1]Even more concerning, studies suggest melanoma is more likely to be diagnosed at later stages in minorities than in non-Hispanic whites, and has a poorer prognosis.[2], [3], [4] By not conducting trials that analyze Hispanics separately, there is no opportunity for medical advances for this disease where they are urgently needed.

Why It is Important to Improve Diversity in Clinical Trials Used as the Basis for FDA Approval

Even if a drug or device is more frequently used in one subgroup, whether males or a large minority population, or individuals of a particular age,  it should be possible to conduct trials for treatments of common diseases that include sufficiently large subgroups in order to determine its safety and effectiveness for most if not all users. If the FDA was not willing to approve a drug or device for the subgroups that were not adequately studied, the companies would have the incentive they need to include adequate representation of those subgroups in their clinical trials.

In addition to requiring companies to include demographic subgroups in adequate numbers in Phase III clinical trials, the FDA should broaden their requirements to include adequate demographic subgroup representation in early phase trials and post-market studies, in order to obtain a broader picture of how safe and effective these products are for subgroups.

Devices

Diversity in clinical trials was even more limited for devices than for drugs and biologics.  The report concluded that age and sex were usually described in the PMA studies, but in fact, women were less than one-third of the patients in 21% of the device trials. Almost one-third of the PMA applications did not report any information about race or ethnicity. Subgroup analysis was even less likely. Twelve percent of the PMA applications did not include analysis by sex, 30% did not include analysis by age, and 73% did not include analysis by race or ethnicity.

Women

Female subjects were often underrepresented in the drug and device trials even when the diseases were prevalent among women. For example, Figure 1-2 shows that the two COPD medication trials had only 20% and 30% women, and yet a 2010 report indicates that 2006 was the sixth consecutive year in which more women (63,006) than men (57,970) died of COPD.[5] Only 10 out of 33 medical device studies had greater than 40% female representation, as shown in Figure 2-2. The PMA for an endovascular occlusion device had only 18% female representation, but this under-representation of women was justified by stating that patients are often selected for such devices if they have larger coronary size and less diffuse nature of coronary disease – and hence typically male. However, this product was not approved only for men; it was approved for all adults.

Racial and Ethnic Minorities

Racial and ethnic subgroups were also not adequately represented in drug or device trials. Hispanics composed 17% of the U.S. population in 2012.[6] However, 17 out of 23 (74%) of device trials were less than 10% Hispanic, as shown in Figure 2-3. Therefore, the majority of device trials were not representative of the Hispanic population.  African Americans make up approximately 13% of the U.S. population,[7] but 87% of the CDER/CBER trials included fewer than 13% African Americans. To give an example with important public health implications, approximately 13% of African Americans have diabetes,[8] but African Americans represented only 2% of participants for type 2 diabetes clinical trials. Such inadequate representation is not useful for determining whether a medical product is safe or effective for African Americans, and is particularly disturbing given the prevalence of diabetes in that population.

The 2011 census indicates that Asians are approximately 5% of the U.S. population, but most (54%) of the trials cited had less than 5% Asian participants. For example, Table 1-3 indicates that both Hepatitis C trials had only 2% Asian composition, whereas there is significantly higher prevalence of Hepatitis C in this population, and a higher rate of liver cancer and a differential response to antiviral therapy in this population.[9]

Demographic data was often collected inconsistently; sometimes race and ethnicity were collected together, and sometimes they were collected as two separate categories. For CDER/CBER, ethnicity was not analyzed for the report since some applications reported race and ethnicity as one item. For medical devices, 23 out of 33 trials (70%) included separate ethnic data. Hispanic ethnicity was sometimes separated into subcategories, and sometimes it was kept as a single ethnic category.  Such inconsistent data collection leads to results that are very difficult to interpret and use.

Age Groups

Age group data were also reported inconsistently, reducing the usefulness of the data. In 8 of the 31 drugs approved (26%), age was reported as a median with range or with a cut-off of 60 instead of 65, rather than actual numbers. These inconsistencies, due to lack of specifics in the guidance, resulted in omission of this 26% of studies from this analysis. For CBER and medical devices, age data were presented as ranges without medians, and were spread over such large age windows that the information was not useful for comparing age groups (Figures 1-4 and 2-1). Although all the device studies included patients up to age 75, for example, the report does not specify how many patients in each study were in any specific age group, such as over 65. A specific example of where a lack of elderly representation is problematic is ALCL. Although it is arguably less common in elderly (20-50% of all cases), the subtype which occurs in elderly is biologically distinct, thus requiring a different treatment, and has poorer clinical outcomes.[10], [11] Lack of children is also a problem in clinical trials; despite 2007 PREA legislation, a 2012 study showed that 96% of all intensive care pediatric patients, and 100% of those ages 13-17, receive off-label medications that have not been tested in those age groups.[12]

Labels

Even the limited information gained from demographic subgroup analyses in these trials was rarely presented in the medical products’ labeling. In subgroup analysis of safety based on sex shown in figure 1-8, only 5 of 30 drugs had data information in the label (17%). In most of these 30 drugs (57%), subgroup analyses were not included on the label, but only in public review material.  In another 5 out of the 30 drugs (17%), the label did not include any subgroup analysis by sex, despite it being mentioned in the public review. For subgroup analysis of efficacy by sex as shown in Figure 1-8, only 6 out of 30 (20%) had efficacy subgroup analysis on the label. Although inadequate, this is not surprising, because there is no requirement for that information to be included on the label. For medical devices, 63% had a statement in device labeling about sex subgroup analyses, 57% about age analyses and 16% about race/ethnicity analyses. The report then glossed over these unimpressive statistics, stating that, “This demonstrates that FDA publicly communicated information on subgroup analyses for sex and age for more than 50% of the PMA applications approved in 2011.”

RECOMMENDATIONS:

  • FDA should issue regulations to require that all applications for devices, drugs, and biologics provide data on safety and effectiveness by sex, age, race and ethnicity.
  • FDA should finalize the draft guidance for sex-specific analysis that the agency proposed in 2011 and issue similar guidance for racial and ethnic minorities and the aged.
  • CBER should require biologics sponsors to report summary subgroup data, just as the CDER requires.
  • In its regulations, guidance documents, and decisions, the FDA should make it clear that the agency will not approve medical products for all populations if the product was not tested on major demographic subgroups with meaningful subgroup analysis.
  • FDA should require that ethnicity and race information be recorded and reported separately, as they have already described in their draft guidance Collection of Race and Ethnicity Data in Clinical Trials in 2005. Unless this happens, data will continue to be collected in an inconsistent manner, making it useless for analysis, as this report demonstrated.
  • Postmarket study requirements should never be a substitute for demographic subgroup analyses in pre-market studies, but post-market studies should be required to provide additional information about long-term safety and efficacy for subgroups.
  • Labels should include subgroup-specific analyses in language that is understandable by health professionals and patients. If not enough information on subgroups was collected to analyze and draw conclusions about potential differences, FDA should be required to state that on the label, and approval should not be assumed for major subgroups that were not analyzed. If lack of subgroup data and analysis is not explicitly stated, physicians and patients will erroneously assume that those groups have been adequately tested. Similarly, as for pediatrics and geriatrics, sex, race and ethnicity should be mandatory, standardized, easy-to-understand sections on the label, so that patients and doctors can quickly find this information or be aware where information does not exist.
  • FDA should issue a public report every 2 years that evaluates compliance with these recommendations.

In conclusion, if the FDA took the firm stance of not approving medical products for the general population unless they have been adequately tested on subgroups instead of simply recommending it, all medical products on the market would have information on safety and efficacy for most potential users.

  1. Arch Dermatol. 2010;146(7):741-746]
  2. Cancer Causes Control 1997;8(2): 246-252.
  3. J Am Acad Dermatol 2004;51(6): 1031-1032.], [end Arch Dermatol 2006;142(6): 704-708.
  4. J Surg Oncol 2001;78(1): 10-16.
  5. American Lung Association, State of Lung Disease in Diverse Communities 2010 report.
  6. U.S. Census Bureau. Profile America: Facts for Features. Available at http://www.census.gov/newsroom/releases/archives/facts_for_features_special_editions/cb13-ff19.html. Accessed October 31, 3013.
  7. United States Census Bureau, State and County QuickFacts. Available at http://quickfacts.census.gov/qfd/states/00000.html , 2012. Accessed October 23, 2013.
  8. U.S. Department of Health and Human Services, Office of Minority Health. Diabetes and African Americans. Available at http://minorityhealth.hhs.gov/templates/content.aspx?ID=3017. Accessed October 23, 2013.
  9. Rev Gastroenterol Disord. 2003;3(3):125-34.
  10. J Pathol. 2003 May; 200(1):4-15.
  11. Am J Clin Pathol 2007; 127:707-722.], [end Mol Cancer Ther July 2011 10; 1127.
  12. AAP 2012 study