Ladin	Imhoff

Medical device manufacturers interested in marketing their products in the United States are required to secure FDA marketing clearance. The process for securing such a clearance is generally based on the device’s risk level: Class I (lowest risk) devices are exempt from applying for approval, Class II devices typically follow a regulatory path known as the 510(k) process, and Class III devices are generally required to submit a premarket approval (PMA) application. The 510(k) process requires the sponsor to demonstrate that the new (candidate) device is substantially equivalent to a predicate device that was either in interstate commerce before May 28, 1976, or had been cleared for marketing using the 510(k) process. Medical products that could not be determined as substantially equivalent to any predicate device had to be automatically classified as Class III and follow the PMA regulatory path, even when their risk level would justify a Class II classification. Class III default classification clearly increased the regulatory and testing burden necessary to secure marketing clearance for such devices.

Table I. (click to enlarge) Evaluation of automatic Class III designation (de novo) process.

Section 207 of the FDA Modernization Act of 1997 (FDAMA) created a new regulatory path that was intended to address the regulatory limitations of the 510(k) process for low-risk devices using new technologies. The process, titled Evaluation of Automatic Class III Designation (or ‘De Novo’ application), is intended to offer manufacturers of a new technology a path for device clearance under the 510(k) process even when no predicate device exists.¹ Table I lists the main steps involved in this process.

FDA has mandated three elements in this process that have clear timeframes:

• The sponsor’s request for evaluation of automatic Class III designation needs to be received by FDA within 30 days of issuing the not substantially equivalent (NSE) letter.
• FDA mandated itself to complete its review and announce its decision regarding the sponsor’s request within 60 days of receiving the request.
• FDA mandated itself to publish its finding in the Federal Register within 30 days following the issue of the classification order.

This article evaluates the actual timeframes associated with the different steps of the review process over the 12 years that the process has been in effect. It also evaluates whether timelines associated with the de novo process are in line with FDA’s stated goals.

Methodology

FDA has developed an electronic database that allows public access to certain elements of cleared regulatory submissions.²  A total of 54 records of devices cleared by the de novo process during the period of August 20, 1999–September 11, 2009 were identified and analyzed. 

The actual dates used for the determination of the review times were based on the information included in the clearance letters published by FDA (NSE letter issued by the agency and submission of the de novo application by the sponsor) as well as the information included in the 510(k) database for each submission (submission of the original 510(k) application).  The review periods were calculated as follows:

• 510(k) review time—the amount of time elapsed from the original 510(k) submission date (from the 510(k) database) to the date of the NSE determination (from the clearance letter).
• De novo review time—the amount of time elapsed from the submission of the de novo application (from the clearance letter) to the date stamp of the clearance letter.

Results

The objective of this analysis was to assess whether timelines associated with the de novo process are in line with FDA’s stated goals. Out of the 54 applications cleared during 1999–2009, 50 of them had all dates necessary to establish the corresponding individual review periods. Fifty-one applications had sufficient information to establish at least one of the review times.

Figure 1. (click to enlarge) Review times of the 510(k) phase (Step II in Table I).

Figure 1 shows the distribution of the 510(k) review time (Step II in Table I) as a function of the clearance date for each of the applications analyzed (1a), grouped in three-year intervals (1b). The average review time since the beginning of 2007 is 279 days, up from an average duration of 196 days prior to 2007 (the median review times for before 2007 and from 2007 on are 161 days and 268 days, respectively). Figure 1 shows a gradual increase of the median initial review time. The differences do not reach statistical significance (Mann-Whitney U test, p=0.14, 1998–2000 versus 2007–2009).

Figure 2. (click to enlarge) Review times of the de novo phase (Step IV in Table I).

Figure 2 describes the distribution of the de novo phase (Step IV in Table I as a function of the clearance date (2a—raw data; 2b—boxplot)). An examination of the data reveals the recent significant increase in the review periods over the last four years. Until the end of 2006, the review of all but two applications (95% of applications) was completed within 100 days, with an average review time of 62 days (median: 51). However, starting in 2007, only 4 out of 13 (31%) applications were reviewed in less than 100 days, and the average jumped to more than 241 days (median: 217). This difference is statistically significant compared with each previous three-year interval (Kruskal-Wallis test, p<0.01).

Figure 3. (click to enlarge) Duration of total review times of de novo products (Step II + Step IV in Table I).

The increase in the duration of the review period of the de novo phase had a dramatic effect on the duration of the overall review period (i.e., the sum total of both review periods), as demonstrated in Figure 3 (3a—raw data; 3b—boxplot). Whereas the average overall review time was 245 days (median: 220) until the end of 2006, medical devices cleared since the beginning of 2007 experienced an average overall review time of 482 days (about 1.3 years; median: 439 days). This difference is also statistically significant compared with each previous three-year interval (Kruskal-Wallis test, p<0.01).

Table II. (click to enlarge) Review times for diagnostic and therapeutic devices (*nonsignificant (ns), p=0.968; **p<0.01; ***ns, p=0.058; Mann-Whitney U test, two-sided).

In trying to better understand the pattern of review times, we have analyzed the data based on the nature of the device, i.e., whether the device was diagnostic or therapeutic. Of the 54 successful submissions, there were 38 diagnostic devices and 16 therapeutic devices. Table II provides a summary of the median and average review times of both diagnostic and therapeutic devices, detailed for both the 510(k) component and the de novo component. The median and average review times of the 510(k) phase are similar for both device types. However, comparing the review times of the de novo phase identified a large difference in both the median and average review times: the median review time of therapeutic devices (84 days) is longer than that of diagnostic devices (50 days) by more than 50%, while the average review time (205 days) is more than three times longer than that of diagnostic devices (65 days) (statistically significant, p<0.01). These differences carry over to the total review times, which are about 50% longer for therapeutic devices than they are for diagnostic devices.

Table III. (click to enlarge) Distribution of review times for different review panels.

The review times of the individual review panels are listed in Table III. Fourteen review panels have processed de novo submissions, with five panels clearing 60% of all devices (32/54). Three panels—immunology, microbiology, and clinical chemistry—cleared 22 immunological and genetic tests, representing the majority of diagnostic devices (22/38) and about 40% of all de novo submissions (22/54). Submissions cleared in less than a month included only diagnostic products, such as a test for West Nile Virus (cleared in 9 days), a Triage B-type Natriuretic Peptide test (cleared in 11 days), and a test for the Influenza A/H5 (Asian lineage) virus (cleared in 13 days).

Discussion and Conclusions

The de novo regulatory route was developed by FDA to offer a path to market for medical devices that present a lower level of risk than products required to follow the PMA process, while not qualifying for the traditional 510(k) process due to the lack of a predicate (i.e., similar) technology. A two-phase process, requiring first a complete 510(k) review, evaluating the technical, preclinical, and clinical information, followed by a well-structured review of the risk level of the new device, was developed by the agency with a commitment to complete the second review phase within 60 days. A review of all de novo submissions cleared by the end of 2009 and published on FDA’s Web site by April 2010 identified 54 such devices (the database lists 55, but one device was found to be substantially equivalent to a predicate, even though it is listed in the de novo group). The 510(k), de novo, and total review times were analyzed for all submissions. 

The study of review times revealed that even though 14 different review panels were able to process de novo submissions, most of the panels processed 1–3 submissions each, making it difficult to identify patterns. Three review panels (immunology, microbiology, and clinical chemistry) processed a total of 22 submissions, all involving diagnostic blood tests. Diagnostic products generally had shorter review times compared with therapeutic devices. It appears that tests that have a significant impact on public health were cleared within days (e.g., West Nile and influenza virus tests were cleared within 9 and 13 days, respectively).

The review process seemed to function well during the first nine years of the program’s existence. FDA was generally able to follow its self-imposed 60-day time limit for the classification of devices that embarked on the de novo regulatory route, with a resulting overall average review time of eight months. However, starting in 2007, review times were extended, leading to an average review time of eight months for the de novo phase alone and a total average review time of 16 months.

It appears that since 2007, the agency has not been able to meet its self-established deadlines for classification and clearance of devices that are placed in the de novo group of submissions. This has led to a troubling extension of the regulatory review process, far beyond FDA’s commitments under the Medical Device User Fee Amendments of 2007 (the reauthorization of the original MDUFA) to review 90% of all PMA-related panel-track submissions within 295 days and 98% of 510(k) submissions within 150 days.³ The review times of de novo products during the last four years have constantly increased. They are now almost twice as long as FDA’s promised review times of panel-track PMA submissions and more than three times longer than the committed review time of 510(k) submissions. Manufacturers should keep these figures in mind when evaluating the proper approval or clearance route for their medical device.

FDA has recently released the preliminary report and recommendation of the 510(k) Working Group, convened in September 2009 to conduct a comprehensive assessment of the 510(k) program. The group found the de novo regulatory pathway “…as currently implemented, is inefficient and has not been utilized optimally across the center.” It recommended that “…CDRH should reform its implementation of the de novo classification process to provide a practical, risk-based option that affords an appropriate level of review and regulatory control for eligible devices.” This recommendation was adopted by CDRH director Jeffrey Shuren, who added that: “As currently implemented, the de novo classification process tends to be associated with lengthy review timeframes and nontransparent data requirements, making it an impractical path to market for many device developers.” The data and analysis provided in this study strongly support these findings and the recommendation.

References

1.    New Section 513(f)(2)—Evaluation of Automatic Class III Designation, Guidance for Industry and CDRH Staff (Rockville, MD: FDA, Center for Devices and Radiological Health, 1998).
2.    510(k) Premarket Notification Database, available from Internet: www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm.
3.    MDUFA Performance Goals and Procedures (2007); available from Internet: www.fda.gov/downloads/MedicalDevices/DeviceRegulationandGuidance/Overview/MedicalDeviceUserFeeandModernizationActMDUFMA/UCM109102.pdf.