Wednesday, September 28, 2016

20160928 - The 20160808 MolDX Q&A for Liquid Biopsy Testing




Frequently Asked Questions: MolDX ctDNA AV Specifications, M00135 (M00136, V1)

1. Does this apply to tests that have gone through FDA clearance/approval or will MolDX accept FDA regulatory review as proof of satisfactory analytic performance?
Our expectation is that any test that has received FDA clearance/approval will be able to achieve 'satisfactory analytical performance' by our specifications.

2. Does this apply to individual marker testing on ctDNA or panels?
We expect this to apply to any ctDNA testing.
3. If testing is only performed as part of a panel, must you have a CV done on each target reported or just ones that are most common? What about multiple variants or targets within a gene? Example BRAF V600E and K or EGFR d19, L858R and t790m? 
First, remember that these specifications only address AV, not CV or CU, which will be addressed separately and may be dependent on the clinical context (e.g., at initial diagnosis versus recurrence). That said, CV for such tests, whether testing for one gene or many, usually relates to variants within a given gene having diagnostic, prognostic or predictive significance, and are usually established by reference to the literature. While we expect this to be true for ctDNA tests as well, we would point out that the 'clinical validity' of these alterations (e.g., drug responses in patients with the variants) was established for tissue-based testing with FDA-approved companion diagnostics, and (to our knowledge) has not formally been established for tissue-based LDTs, let alone by ctDNA-based LDTs, detecting the same alterations.
4. The article uses 'qualitative' and many of the companies are performing 'quantitative' testing - does that exclude a 'quantitative' testing lab from this requirement? 'with ctDNA or quantitative ctDNA assays intended for drug response/disease monitoring or minimal residual disease (MRD) applications.' What if a laboratory characterizes assays as quantitative or semi-quantitative assays?
As even CAP and NYSDOH indicate in their checklists, the requirements for analytical validation for any quantitative or semi-quantitative test are different (usually more extensive) than those for a qualitative test, as in most cases are the clinical applications. Therefore, if a lab is reporting quantitative measurements for its ctDNA-based test(s), or has indications requiring quantitative testing (e.g., MRD and TDM), then different AV, CV, and CU requirements will almost certainly apply, which at the moment will be addressed on a case-by-case basis.
Does this mean digital will not fall under this criteria?
As indicated in the answer to #2 above, these criteria apply to all ctDNA-based testing regardless of methodology, so this would include ddPCR.

5. Are liquid biopsies being considered differently than LDT testing that is currently routinely performed on tissue and reimbursed under the current clinical lab fee schedule?  Examples are tests that are reimbursed as LDTs for 81235 (EGFR) or 81275 (KRAS) etc.
Yes, because the analytical performance characteristics and appropriate clinical use for LBx-based tests are different than tissue biopsy (TBx)-based tests, as indicated in the recent FDA approval for Roche’s cobas EGFR Mutation Test v2.
Is the intention to increase the evidence for all molecular testing including validation on tissue testing? Currently, there is no standard that makes tissue based reference labs perform more than a CLIA level validation when non NGS.
No, MolDX has previously indicated (M00127, M00130) that any TBx-based tests that are not tissue-only 'hotpsot' panels, just like all liquid biopsy (LBx)-based tests, should be billed under 81479 (NOT using Tier 1 or 2 codes, or GSP codes 81445, 81450, and 81455).

6. 'ttDNA' is used throughout the document and we conclude that is Tissue DNA but want to confirm.
Yes, tumor tissue DNA (ttDNA) to distinguish it from circulating tumor DNA (ctDNA)

7. Section I
General Laboratory Requirements
B. New York State Department of Health (NYSDOH) final test approval      

Q: Does this apply to laboratories that are not servicing the state of NY? Or do you support this as a standardization tool that will be used as a requirement for future tests? If you gain FDA clearance, thus not needing NY approval (please confirm) – what, if any, validation is required?
Based on feedback we have received since the publication of these specifications, we have decided to remove the requirement for final test approval from NYSDOH, and the document will be updated to reflect this.  For FDA-cleared or approved tests, please see our response to question #1 above.
8. Section III
Analytical Requirements
E. The technology platform and/or sequencing chemistry used for the orthogonal reference method should be different than that for the ctDNA assay being validated. For example, if the ctDNA assay uses Illumina sequencing, preferred orthogonal reference methods include digital PCR or Ion Torrent sequencing. Alternatively, reference sequence information provided by a vendor is acceptable.

Q: Please provide more clarity around 'orthogonal reference' Many laboratories are performing Quant PCR and confirming with Sanger Sequencing for example does this state that a change would need to take place and perform by Quant PCR and confirm with NGS for example (completely different method)?
'Orthogonal reference method' is merely a term of art for the 'other' method against which the LBx-based test is being compared. It can be any method the test developer chooses, and can even be multiple methods for different types of alterations (e.g., Sanger sequencing for indels, qPCR for fusions, etc.). The test developer just needs to indicate what exactly the 'orthogonal reference method' is.  Our guidance is simply pointing out a best practice that is common in the industry that the technology platform and/or sequencing chemistry used for the orthogonal reference method should if at all possible be different than that for the assay being validated. And again, we remind test developers that the 'reference' sequence may be one provided by a vendor (e.g., NIST, Coriell, Horizon, etc.)
9. G.4 The laboratory will report the sample-level (not variant level) PPA and NPA, a sub-analysis of Tables 1, 2 (if performed) and 3, for the following 13 gene-variant groups…
Q: Is sample-level the number of samples?

Yes, in other words, does the sample contain any of the specific gene-variant groups listed? For example, does the sample have any ALK SVs, or any BRAF V600E/K SNVs, etc.?

Last Updated: 08/08/2016