Some of the biggest drug makers in the world have a problem with the way the US Food and Drug Administration defines different types of biomarkers in its draft guidance on pharmacogenomics data submissions, according to documents released by the agency recently.
Companies like Pfizer, Johnson & Johnson, GlaxoSmithKline, and Roche — and even the National Institutes of Health — took issue with the biomarker framework as well as the portion of the draft describing voluntary genomic data submissions, or VGDSs, which the FDA claims will help it keep tabs on new pharmacogenomic technologies and applications.
The companies’ remarks, which appear on the FDA’s website, were submmitted during the 90-day public-comment period required before a draft guidance can become a formal guidance document. The comment period for the draft, whose docket number is 2003d-0497, expired on Feb. 2. The FDA said a formal version of the guidance will appear any day.
Though their letters critiqued several components of the draft, the most common and detailed issues were with VGDSs and the way in which the FDA defines and classifies biomarkers.
Pfizer, the world’s largest drug maker by revenue and R&D spending, took a particularly strong stance against VGDSs. In a 6-page letter dated Feb. 3, Melissa Tassinari, senior director of worldwide regulatory policy and intelligence at the drug maker, said the VGDS portion of the draft “is not workable in its current format” and that the entire VGDS process ought to “be removed from the proposed guidance at this time. …
“[It] is more appropriate for progress in the advancement of understanding to be through the use of reports generated in the scientific literature, than it is to provide what may be very preliminary data,” Tassinari wrote.
She went on: “We remain concerned as to how and when voluntary [sic] submitted data become suitable for regulatory decision-making and in turn become a submission requirement. The pathway and procedures that would be followed are not clearly defined in the current draft guidance. It is particularly important that the constitution and remit of the Interdisciplinary Pharmacogenomics Review Board [IPRB] and the nature of its interaction with the Review Divisions be clarified to ensure that voluntary submissions would not influence regulatory decision-making on INDs or NDAs,” she concluded. Read Pfizer’s letter here.
For its part, Johnson & Johnson suggested in its letter that the IPRB be populated with individuals who “will not be directly involved in regulatory decision making for a compound. Clarification is required on how the FDA will use the data from VGDS as well as on the responsibility and constituency of the IPRG.”
Like J&J, Roche was more lenient than Pfizer on the issue of VDGSs. In its 4-page letter, dated Feb. 2, the Swiss drug and diagnostic giant said it “welcomes [the] concept” of VGDSs. Roche’s only suggestion, written in bold-face type, is for the FDA to “further elaborate” on the term “regulatory decision making” so that sponsors can better understand how data submitted as a VGDS “can and cannot be applied to the regulatory review process. … “
The letter, written by regulatory group director Joanna Waugh, managed to plug the company’s AmpliChip product, which Roche Diagnostics was forced to scuttle late last year following an FDA decision [see ]. “The proposed requirements for GLPs under 21 CFR Part 58 and their application to non-clinical studies are currently unrealistic with regard to technologies such as the Affymetrix platform at this time,” Waugh wrote. Then, in bold-face, she added: “Roche proposes that additional flexibility must be considered for pharmacogenomic analytical techniques with respect to 21 CFR Part 58.” Read Roche’s letter here.
GlaxoSmithKline, writing in an unsigned letter dated Feb. 24, was also relatively conciliatory. The British drug maker, which is believed by its peers to be the biggest spender on pharmacogenomics technologies, did not invest a lot of space in its letter on VGDSs, per se. The company did openly wonder why certain data — specifically, proteomics data — were left out of the draft: “It is unclear why proteomics is excluded from the guidance,” the company wrote. “It would be helpful for the science to have clarity on handling these exploratory biomarker data, and indeed other ‘omics’ such as metabolomics, and thus permit sponsors to submit such data for review and discussion under VGDS.” (The FDA’s draft guidance, which can be seen here, reads: “Pharmacogenomics … does not refer to data resulting from proteomic or metabolomic techniques. This document is not meant to provide guidance on pharmacoproteomics . … “) Read GlaxoSmithKline’s letter here.
In what is perhaps the most detailed comment letter sent to the FDA, J&J said there is an “urgent need for transparency” on the cases in which VGDSs are found to have an impact on the regulatory process. “VGDS presents some concerns to the industry, including: cost and effort, inappropriate replication of data analysis, how the data will be used, how the VGDS may be regulated, IRB delays (e.g. if informed consent might not have been obtained for VGDS), and the regulatory implications,” J&J wrote in its Jan. 7 letter.
The company goes on: “It will be imperative for the FDA to clarify its position on intellectual property (both inventorship and ownership) if, for example, its staff recognizes novel patterns of efficacy or safety related to PGx markers in VGDSs from multiple companies. From an intellectual property perspective, it is important for pharmaceutical companies to understand the ownership of the data voluntarily submitted and of any inventions deriving from those data in whole or part. …
“If FDA staff, through meta-analysis of data from multiple companies, discover that a biomarker is important for safety or efficacy, who owns that invention and the right to commercialize it? Would the FDA scientist be included as an inventor? In addition, would VGDS data be subject to public review through the Freedom of Information Act? It would be reassuring to include a statement that the FDA would not seek to patent any inventions conceived or reduced to practice from the use of the submitted data,” J&J wrote.
J&J’s unsigned 15-page letter also stressed that the FDA ought to include guidelines describing DNA and RNA collection. “Sample collection is the proximal portion of the PGx process and an important one. If DNA/RNA is not collected, the current Draft Guidance is ineffective,” the company wrote. “We would like to propose that the FDA consider including a section in the Guidance for PGx Data Submission on ‘DNA/RNA sample collection,’ or alternatively, a separate guidance on sample collection.”
J&J also took issue with the way the FDA stratifies and defines different biomarker types. The company said that “several areas” of the draft guidance “require reevaluation and clarification.”
“The following questions were recurrently raised during our review: For a given biomarker, what impact would a change in biomarker status have on data submission requirements? What are the boundaries among ‘Known Valid Biomarkers,’ ‘Probable Valid Biomarkers’ and ‘Non-Valid Biomarkers?’ Who is responsible for the validation of biomarkers? There is a clear need for improved definitions of boundaries between the categories of data,” the company said.
J&J goes on to “encourage the FDA” to monitor the definitions of biomarker. “What if the FDA learns, through multiple submissions, that test results qualify for Known Valid Biomarker status; how will the FDA communicate changes to biomarker status to sponsors?” J&J wrote. “Will there be new requirements imposed on the sponsors for PGx testing? Will the guidance be updated whenever a change in status has occurred or will the information be communicated using a different mechanism? A description of this information and the procedures for communicating it back to industry would be valuable,” the company said.
In its letter, GlaxoSmithKline wrote that “the value of implementing a ‘validity gradient’ is questionable and will likely result in confusion and inconsistent application by various sponsors. GSK suggests that the primary point of emphasis for the submission decision algorithm be whether the data are used for regulatory decision making. … “
The company specifies: “There is a need to better define the terms ‘valid biomarker,’ ‘known valid biomarker’ and ‘probable valid biomarker’ to aid transparency and predictability for sponsors. In addition, the purpose of the ‘biomarker’ (e.g. diagnostic, predictive, prognostic) should also be specified to ensure common understanding with regard to the utility of the information.
“It would be particularly helpful to Industry if representative real examples from FDA were included for what are considered to be known valid and probable valid biomarkers (including for the drug metabolizing cytochrome P450 enzymes), both for the IND phase [and] for the unapproved or approved NDA/BLA phase of the regulatory review of such data. It would be beneficial to industry if FDA elaborated on what constitutes an ‘established scientific framework or body of evidence’ for a valid biomarker with specific examples.” [Italics are Glaxo’s.]
For its part, Pfizer agreed that the definitions of biomarkers “require further clarification.” Tassinari wrote that “it is not clear how the criteria ‘widely accepted in the scientific community’ are to be achieved. Similarly, what criteria are required to achieve the status of ‘known valid?’
“It is important to note that validation is often specific to the use of a test for a particular purpose,” Tassinari continued. “For example, we would not consider the genotyping of CYP2D6 to be a valid biomarker for identifying intermediate metabolizers, though it may be valid for the identification of poor metabolizers.”
Tassinari went on to suggest that the FDA creates a “regularly updated ‘living’ list to be created and published” on its website.