Letter to CMS regarding the Proposed Decision Memo for Next Generation Sequencing (NGS) for Medicare Beneficiaries with Advanced Cancer (CAG-00450N)
To Whom It May Concern:
I am a molecular pathologist specializing in massively parallel sequencing (MPS, commonly referred to as next-generation sequencing or NGS). I am certified by the American Board of Pathology in Molecular Pathology, Anatomic Pathology, and Clinical Pathology. As a physician, my primary concern is for my patients, and it is my sincere belief that the Center for Medicare & Medicaid Services (CMS) shares my priorities. Nevertheless, as a physician I believe it is my responsibility to convey that I have significant concerns about the Proposed Decision Memo for Next Generation Sequencing (NGS) for Medicare Beneficiaries with Advanced Cancer (CAG-00450N).
I work at a major academic health center, with one hundred percent of my effort dedicated to patient care or training others in caring for their patients. I do not engage in research, other than efforts to improve the clinical testing we offer to our patients. Using four different MPS assays, we sequence approximately 4,000 specimens annually in our laboratory, a volume of testing that is steadily increasing and shows no sign of abating. For a significant and increasing proportion of our oncology patients, the identification of clinically relevant genomic changes by in-house MPS is already standard-of-care.
It is important to understand that MPS is a technique, not an assay. Like any other technique (e.g. PCR, MRI, microscopy), MPS can be employed in any number of ways, with the details optimized for specific clinical needs. For example, an assay can be designed for rapid results at the expense of sensitivity. Alternatively, an assay can be designed to include fewer genomic targets, lowering the cost per patient. At our institution, by offering a reflex assay for lower quality and quantity specimens, we have been able to reduce our rate of rejected specimens to roughly 3%, far lower than any commercial entity. We routinely make such considerations when developing and implementing assays in our laboratory. It would be inappropriate to restrict microscopy for use at specific magnifications, or to optimize an MRI machine for only metastatic cancer. Similarly, it is inappropriate to restrict MPS assays to the design decisions of remote commercial entities and institutions. Patient care is personalized and optimized by permitting individual institutions to offer the clinically validated assays that best serve their specific patient populations.
Personalizing and optimizing the results of an MPS assay for a given patient includes the interpretation and reporting of the data. It is natural to assume that the interpretation and output of MPS assays are analogous to single-mutation assays, simply with more findings. I believed as much before my formal training. This is a misapprehension – MPS is in many ways a sea-change from other forms of laboratory testing. As its name implies, MPS generates massive amounts of data for every patient – exponentially more than most other laboratory techniques. Genomic diagnostic medicine is multi-dimensional and multi-parametric, and as such has much in common with other interpretive disciplines such as anatomic pathology or radiology. The output of the report focuses on the identified mutations, but in the appropriate clinical context, I can routinely address the following clinically-relevant concerns:
- Is mutation X really there?
- Was a mutation in gene Y missed?
- Could this mutation be native to the patient, rather than the cancer?
- Is the pattern of mutations significant?
- Could there be a misdiagnosis or multiple disease processes?
- Is the disease heterogeneous?
- Is there evidence the patient is responding to treatment?
- Is there any evidence of contamination?
- Is additional testing warranted?
- Is additional material available for testing?
- Do the results conflict with other laboratory findings, and if so, how can we resolve the discordance?
I can address these concerns because I have advanced training, access to the sequencing data and the patient’s medical record, and a detailed understanding of the performance characteristics of the assays we use. I care about the answers to the questions because these are my patients. Restricting reimbursement to entities with FDA-approved assays performed at a select few laboratories will dissociate analysis of patient data from appreciation and understanding of the individual patient. For optimal patient care, the performance, interpretation, and reporting of sequencing assays should be managed by a trained medical professional who is part of the patient’s health care team.
One of the dramatic advantages of MPS is the flexibility it offers to assay design. The field of clinical genomics evolves constantly, with near-constant discovery of new clinically relevant genes and mutations , and continual development of novel targeted therapies (as with larotrectinib, an inhibitor directed at NTRK fusion genes). New laboratory and computing tools also become available, permitting improved sensitivity, specificity, and detection of new, clinically relevant data (as with mutational tumor burden). Requiring FDA-approval locks MPS testing into what we currently know and are currently able to interrogate. In order to respond to dynamic changes in genomic medicine, laboratory professionals must be free to innovate appropriately.
As it happens, I recently authored a paper using clinical data to demonstrate some of the challenges imposed by FDA-approved assays. In “Genomic heterogeneity of ALK fusion breakpoints in non-small-cell lung cancer” (PMID: 29327716; DOI: 10.1038/modpathol.2017.181) we demonstrate that a laboratory-developed MPS-based oncology assay outperforms two FDA-approved companion diagnostics (FISH for ALKrearrangement and IHC for ALK protein expression). The MPS panel offers more clinically relevant data, with greater specificity and potentially improved clinical outcomes. Moreover, we show that the MPS data require a significant level of expertise to properly integrate into testing algorithms.
In summary, it is my opinion that the proposal by CMS is excessively restrictive, interfering with the delivery of personalized, high-quality care to my patients. It prevents me from offering a complete interpretation of my patients’ results; it prevents me from designing and delivering the assays that will best serve my patients’ needs; it prevents me from integrating new data and new testing as the field evolves. For the sake of my patients, I strongly urge CMS to reconsider this proposal.
Jason N. Rosenbaum, MD
Assistant Professor, University of Pennsylvania Perelman School of Medicine
Attending Pathologist, Center for Personalized Diagnostics