Lessons from MTB 1; Questions of "actionability"

Hello all.  I have been involved in running and supporting tumor boards for many years.  Over the last 2 I have been supporting Molecular Tumor Boards (MTB) for a regional cancer center.  This has been a rewarding experience, and I would like to share what we have learned in the community setting with the broader community.

I will start a (hopefully) monthly series reviewing interesting MTB cases, and what lessons we can derive from them.  I will also use them to cover controversial topics in the precision medicine space. I will try provide updates to cases when appropriate, so we can take a longitudinal look at the implications of the use of precision medicine.


Case 1:  Paraspinal Mass of no hope


Patient 1 is a 53 year-old woman who presents with a paraspinal mass, as well as numerous other lesions at multiple sites. 

Initial presentation: 1-year history of pelvic pain; a pelvic CT scan showed a 2x4 cm pelvic mass.  A left psoas muscle biopsy was taken and showed a poorly differentiated malignant neoplasm.  An extensive immunohistochemical workup yielded nothing conclusive about the tumor origin.  She was initially treated with radiation therapy and the lesion was thought to be localized.  She was also given one cycle of carbo/taxol.  the Diagnosis given was "poorly differentiated Carcinoma, NOS".  This lesion was thus considered a "CUP", or "carcinoma of unknown primary."


The patient returned 5 months later with uncontrolled back pain.  An MRI showed bone metastases and progression of the original lesion.  Biopsies at multiple locations suggest these lesions are of the same malignant process (based on histologic morphology), including a paraspinal mass taken that was subsequently submitted for next-generation sequencing (NGS) testing by a prominent third-party laboratory.

The patient was put on a course of Folfirinox. This continued for two months, but the patient failed to respond to therapy.  7 months after presentation, the patient's clinical condition was poor and it was decided to send the original specimen for NGS testing.

Clinical Questions:

What is the primary tumor?  What are available therapies can we give to this patient not responding to standard care?

Molecular Findings:

Relevant to this case:


The above are the "significant" findings in the case by the molecular lab.  Many other mutations were identified in several genes, these were labeled "Variants of Uncertain Significance (VUS)" in the resultant report. 

Significance of Molecular Findings:

Several cancer-related pathways are activated in this tumor, including Wnt, mTOR, and the cell cycle.  However, for brevity, I will focus here on the elephant in  the room- a mutation in a Mismatch Repair gene (MMR) MSH6 that results in loss of function.  This likely is responsible for Microsatellite instability (MSI), and subsequently a high tumor mutation burden, which is also seen in this case.  A detailed description of this pathway, an how it results in MSI, won't be covered here (maybe I'll leave if for a future case).  I would like this first MTB to discuss a controversial topic in the field.

Expected use of precision medicine: treatment-

The clinical significance of this finding is a suggested therapy that follows the current trend in the practice of oncology:  treating tumors with immune modulating drugs.  At an extremely high level, the basic mechanism at play here is that tumors with lots of mutations (like those due to deficient MMR processes) will have novel proteins created as a result of these mutations, and some of these proteins will be expressed at the cell surface, where they can act as antigens and be recognized by the immune system.  The tumor cells are then targeted for destruction by activated cytotoxic T-cells.  However, tumors develop resistance to this mechanism by expressing PD-L1 ligands on their cell surface, which binds to the PD-1 receptor and informs the immune cell NOT to kill it.  Immune therapies like pembrolizumab (Keytruda) and nivolumab (Opdivo) are antibodies that bind to the PD-1 receptor and prevent the PD-L1 ligand from binding, meaning that the tumors cells cannot bypass this process and cytotoxic T-cells can still kill the tumor.  This is a particularly good strategy in light of the numerous activated cancer pathways in this particular tumor.  However, there is no guarantee that the numerous mutations present create novel antigens for the immune system, and thus no guarantee that this strategy will be successful. Additionally, mutations in PTEN and JAK1, seen in this patient, may be relevant to resistance mechanisms to immunotherapies (see recent relevant papers for PTEN and JAK1).

Other relevant information derived form this tumor biology:

This tumor shows several mutations that are significantly associated with a colorectal primary adenocarcinoma.  APC nonsense mutations are seen frequently and are likely a key driving event in colorectal cancers.  These mutations are frequently paired with TP53 mutations, also seen here.  MMR mutations are often seen in colorectal cancers to a degree that at many institutions MSI or MMR testing is reflex in colorectal cases.  Based on this, it is likely that the tumors seen in this case are metastatic colorectal cancer.  It possible that this patient may have responded better to therapy if they were treated as a metastatic colon carcinoma patient rather than a CUP patient.

It should also be noted that MMR mutations can be hereditary, and given the patient's age it may be appropriate here to get a substantial family history or consult a genetic counselor.  Many labs can identify with a high level of confidence if the variant detected is of germline or somatic origin as well (the lab in this case unfortunately does not provide such information). Follow-up with this specific lab and getting variant allele frequencies suggested that this particular variant was of somatic origin.

The information gathered from NGS testing proved too late for this particular patient; she was in poor clinical condition when Keyrtuda was initiated.  Within a month she was placed into hospice care.


What is the definition of "clinically actionable"?  when is NGS testing warranted?  In this space, with the question of clinical utility of NGS testing, there is no agreement.  Many define it rigidly; specifically when a biomarker linked to clinical response from a specific approved therapy is expected in the results.  This is at least in part because of a need to eliminate "unnecessary" medical testing, particularly in light of paying for medical services. However, it is also likely because this is simply how clinicians think about laboratory tests in general.  I take a different view.  My opinion (and hopefully that of others as well) is that a "test" should be considered when it has the potential to change or influence the management or outcome of a patient.  These two views are not as similar as they would appear. 

The problem with NGS is that it is not really a test, at least, not as we define most other tests.  Other tests are ordered for a defined, specific purpose, and they are usually have a binary result.  They are used to confirm specific clinical hypotheses, or can be used in a combination to create a differential diagnosis by the clinician.  A hemoglobin A1c test, for example,  tells you one specific thing- if the A1c level is within or above an expected reference range of normal.  You choose to use that test for a very specific reason- that is to confirm or monitor diabetes. You would not really use this in other settings, and doing so would be characterized as "unneccessary". Other non-NGS genetic tests are the same- to confirm a mutation within a specific gene that can diagnose disease or guide treatment.

NGS is different, particularly for larger gene panels.  It is not a test to confirm specific clinical hypotheses as much as it is a hypothesis-generating tool to understand the genetic mechanisms of disease. It is not a single test, but an almost un-quantifiable number of concurrent tests (for each nucleotide sequenced in each gene tested, each having the ability to alter protein structure and function).   It is because of this, I believe, that payers have had difficulty accepting NGS above Sanger-sequencing-based tests, which clearly test one specific thing for one specific purpose.  This follows the idea of "one biomarker for one disease" mantra for companion diagnostics, an idea that is rapidly becoming obsolete.  In the case illustrated above, NGS testing was ordered not to answer a specific clinical question, but to gain insight into the patho-mechanisms of disease, to them try to understand the tumor and to find a treatment.  Interpretation of NGS assays can be similarly complex- where the result, particularly in large assays, is not binomial but requires professional judgement; it requires us to practice the "art" of medicine.  ...I am getting a bit off topic here, but there will be more on this subject in future cases.

NGS testing can be used to:

  1. Help establish a correct diagnosis (as in this case)
  2. Identify biomarkers associated with possible sensitivity to targeted drugs
  3. Identify biomarkers associated with prognosis
  4. Identify aberrant pathways that can be targeted by existing therapies (also seen here)
  5. Identify biomarkers associated with adverse events to specific therapies
  6. Identify biomarkers associated with resistance to specific therapies
  7. Identify putative mechanisms of disease that are driving oncogenesis
  8. understand the genetic complexity of the tumor; and when any treatment may be futile based on existing mechanisms
  9. Capture molecular data for this tumor for research purposes to better understand cancer mechanisms and responses to therapy in a registry.

I believe for all the points above (with possible exception of the last), NGS testing will ultimately be deemed "medically necessary" for all cancers as part of a diagnostic work-up.  However, it may be some time before physicians, payers, and regulators agree.