Does your study require FDA approval?

Recently became aware of a pathology group in a large community hospital who submitted a retrospective study to look at whole slide imaging versus conventional light microscopy for purposes of primary H&E diagnosis.

The goals of the study included essentially a self-validation of the pathologists in the group participating in the study to look at concordance between whole slide images and light microscope.

The group and their hospital partner have recently invested in a whole slide scanner and the group desired to look at how they compared with themselves with “referred” or “gold-standard” diagnoses with virtual compared with conventional microscopy.

An honest broker in the study would insure the cases would be randomized so that the primary pathologist on the case would not be presented the same case, albeit more than 2 years after the original sign out and cases where a “ground truth” diagnosis could not be established by the panel of referee pathologists would be eliminated from cases selected to be viewed in this intraobserver concordance study.

The study design, number of cases, case selection and diagnosis adjudication all seemed to be complete and thorough and address the major pain points in a study of this sort.  A general community setting with a large subset of representative cases in a busy surgical practice with many pathologists comparing their diagnoses by the two different means, if they were able to show a high degree or superior degree of concordance with themselves and the “gold standard diagnosis” would provide another useful piece of data to show that whole slide imaging is non-inferior to light microscopy for primary H&E surgical pathology diagnoses.

The institutional IRB replied with similar thoughts in terms of study design, approach, number of cases & wash-out time as being appropriate but requested that the investigators seek a waiver from the FDA for their study as these devices have yet to be approved by the FDA for primary diagnoses.

The alternative, short of being granted a waiver from the FDA to perform the study would be to consent each and every patient prior to biopsy or surgery for this minimal to no risk, retrospective intraobserver, peer-vetted study that slides produced from their biopsy or surgical procedures may be viewed on a device that has not been approved for clinical use for primary diagnosis by the FDA.

This is a new one on me.  I thought I had heard it all or at least most of the responses by an institutional IRB when trying to assess risk versus benefit and patient safety/quality as a principal investigator for a number of clinical trials involving surgical pathology case review, including for validation of telepathology to meet the standards of care in a community hospital setting.

Isn’t this the goal of doing the study?  To show efficacy and potential for a self-validated laboratory developed test for the hospital and laboratory?

If the device were approved would the pathology group be exempt from validating the technology for themselves like any new test, assay or platform developed, implemented or installed in a laboratory?

Most pathologists and laboratories I know usually don’t run an instrument or new test without extensive testing, validation and training to insure its safe and effective use within the laboratory.  Beyond that, continual quality assurance, quality control and subsequent proficiency testing insure safe, accurate and precise performance of the analyte, test, device and/or operator/personnel involved in the test result.  In this case, an anatomic pathology diagnosis.

Does anyone else have any experience with this and a request of this nature.  To my knowledge, most if not all of the studies performed to date looking at concordance across two platforms, such as this one does did not mention the principal investigator requesting FDA waiver or the study necessitating it to grant the words “Following institutional IRB approval, a series of cases, …” in the Methods and Materials of most studies I have seen.

Would be interested by comment or e-mail if you have any experience or thoughts with this.