Response to Comments: MolDX: Pharmacogenomics Testing

(Note: this comment was submitted to multiple MolDX contractors and has been edited to be representative of the comments submitted to all MACs)

On behalf of the Association for Molecular Pathology (AMP) and the College of American Pathologists (CAP), we thank you for the opportunity to review and comment on CGS Administrators, LLC’s proposed coverage policy for MolDX: Pharmacogenomics Testing (DL38394).

AMP is an international medical and professional association representing approximately 2,500 physicians, doctoral scientists, and medical technologists who perform or are involved with laboratory testing based on knowledge derived from molecular biology, genetics, and genomics. Membership includes professionals from academic medicine, hospital-based and private clinical laboratories, the government and the in vitro diagnostics industry.

As the world’s largest organization of board-certified pathologists and leading provider of laboratory accreditation and proficiency testing programs, the CAP serves patients, pathologists, and the public by fostering and advocating excellence in the practice of pathology and laboratory medicine worldwide.

We are submitting joint comments because both of our organizations share the same perspective regarding this draft local coverage determination (LCD). We appreciate the effort that has gone into the development of this proposed LCD, and we offer the following recommendations for CGS’s consideration.

AMP and CAP commend [the Contractor] for recognizing the importance of providing coverage to Medicare beneficiaries for pharmacogenomic testing. Additionally, both groups greatly appreciate that CGS directly referred to both the Clinical Pharmacogenetic Implementation Consortium (CPIC) guidelines and the AMP minimum allele testing guidelines for pharmacogenomics as valid and informative resources on pharmacogenomic testing. AMP and CAP applaud [The Contractor] for covering a number of diverse genes in one policy and encourage [The Contractor] to continue utilizing pharmacogenomic-related clinical practice guidelines, such as those created by CPIC and AMP, in addition to the pharmacogenomic information included in FDA-labeling¹, to help craft future pharmacogenomic policies for additional genes.

1 https://www.fda.gov/drugs/science-and-research-drugs/table-pharmacogenomic-biomarkers-drug-labeling

1. Specific Coverage Information

This coverage policy is limited to CYP2D6, CYP2C19, CYP2C9, HLA-B*15:02, and HLA-A*31:01. We appreciate your inclusion of all of these genes in this draft LCD; however there are currently more pharmacogenes with high clinical validity supported by the CPIC guidelines², such as TPMT, NUDT15, DPYD, UGT1A1, CYP3A5, HLA-B*57:01, HLA-B*58:01, CACNA1S, CFTR, RYR1, SLCO1B1, G6PD, CYP2B6. CPIC has evaluated the evidence for other pharmacogenes that have not been incorporated into this draft policy and we recommend that these genes also be considered for inclusion. Beyond CPIC, three additional sources that should be used for determining pharmacogenes with clinical validity for future policies are the FDA-approved prescribing information¹, the Dutch Pharmacogenetics Working Group³, and PharmGKB⁴.

Additionally, we agree that the policy should remain unlimited by provider type.

2. Clinical Indications

[The Contractor] provides the clinical indications of single gene testing for CYP2D6, CYP2C19, CYP2C9, HLA-B*15:02, and HLA-A*31:01. AMP and CAP recommend the following revisions for the coverage of CYP2D6, CYP2C9, and CYP2C19.

CYP2D6

The draft LCD states that the following null alleles must be included for CYP2D6: *3, *4, *5, *6, *7, *10, *17, and *41. Additionally, the following duplications must be tested: *1xN and *2xN.

AMP and CAP recommend the following revisions:

• CYP2D6 *17 and *41 are decreased function alleles, not null alleles. These should be removed from the list as null alleles5 and placed in a separate category of “decreased function” alleles.
• CYP2D6 *7 is an allele that is only found in 0.1 percent of Caucasians⁶. Therefore, this should be removed from the list.
• CYP2D6 *29, a decreased function allele found in 6.5 percent of African Americans⁶, is missing and should be included.

Additionally, AMP will be publishing minimum testing requirements for CYP2D6. Upon release of those recommendations, we ask that CGS update the coverage requirements as needed based on those guidelines. AMP plans to provide this updated guidance to CGS as soon as it is available.

CYP2C9

We appreciate CGS’s recognition of AMP and CAP’s joint recommendation for variants *2, *3, *5, and *6 to be included as part of the CYP2C9 test. However, the proposed CYP2C9 requirement does not include CYP2C9 *8 and *11. These alleles are recommended by AMP together with CAP as part of a set of minimum alleles needed for clinical testing of CYP2C97. CYP2C9 *8 and *11 are present in 6.7 percent and 1.4 percent of African- Americans, respectively⁷. Additionally, CYP2C9 *9, a decreased function allele found in 2 percent of Caucasians, should be included as well. For reference, please refer to “CYP2C9_frequency_table”⁷.

2 https://cpicpgx.org/guidelines/

3 https://www.pharmgkb.org/page/dpwg

4 https://www.pharmgkb.org/

5 https://www.pharmvar.org/gene/CYP2D6

6 https://api.pharmgkb.org/v1/download/file/attachment/CYP2D6_frequencies. xlsx, downloaded 10/2/2019

Additionally, the draft LCD considers single gene testing for CYP2C9 reasonable and necessary when “the patient has a diagnosis for which a provider is considering treatment with an antidepressant, anxiolytic, mood stabilizer, or Mayzent, and the patient is open to treatment with such a medication.” We appreciate the broader coverage of medication classes when multiple medications are impacted by variation in a specific pharmacogene (i.e., CYP2D6 and antidepressant therapy). We propose that “antidepressant, anxiolytic, mood stabilizer” be changed to “nonsteroidal anti-inflammatory drug or phenytoin” to more accurately reflect the data supporting the use for testing of CYP2C9 in the CPIC guidelines².

CYP2C19

We suggest that the description for CYP2C19 be changed to “provider is considering treatment with an antidepressant, anxiolytic, mood stabilizer, clopidogrel, or antifungal” to more accurately reflect the data supporting the use for testing of CYP2C19 in the CPIC guidelines⁸.

3. Technical Requirements

Pharmacogenomic testing provides the greatest benefit to patients when the healthcare provider is able to easily determine when an actionable prescribing change and/or treatment is indicated by the patient’s genotype. We agree with the requirements included in the draft LCD and support requiring the completion of an interpretation in order for the test to be considered reasonable and necessary.

4. Noncovered Indications

Pharmacogenomic testing is not covered if a treating clinician is not considering treatment with a medication that has an actionable drug-gene interaction. Indications for pharmacogenetic testing continue to evolve as evidence is generated. We recommend that currently noncovered indications for pharmacogenetic testing be re-considered on an annual basis.

5. Special Documentation Requirements

AMP and CAP are concerned that it would be difficult for many laboratories, including commercial reference laboratories, to complete the proposed documentation requirements for pharmacogenomic testing. In most instances, commercial reference laboratories do not have access to the patient’s medical record, which would make obtaining the proper documentation for coverage exceedingly difficult, if not impossible. Additionally, it is unclear how a laboratory would document that the ordering physician has “the necessary experience/training to both diagnose the condition being treated and also to prescribe medication.”

In regard to the ordering physician, the draft LCD states that the patient’s medical record “must reflect the specific drug-gene interaction(s) of concern. General classes of medications or drug-gene interactions do not meet this requirement.” This may be difficult because some doctors may not be knowledgeable of the exact gene that must be documented. This means that if a physician, other than a pharmacogenomic specialist, orders a test it might not be covered.

7 Pratt et al., 2019, J Mol Diag, 21(5): 746. https://jmd.amjpathol.org/article/S1525-1578(18)30594-4/fulltext

8 https://cpicpgx.org/content/guideline/publication/voriconazole/2016/27981572.pdf

We agree that physicians should document medical necessity for any testing in the patient’s medical record, as part of good medical care, however, tying the documentation to coverage is difficult for the commercial reference laboratory. We recommend that [The Contractor] remove this requirement as it is unworkable with how pharmacogenomic tests are ordered. Additionally, we fear that physicians would be unlikely to order pharmacogenomic tests based on this level of justification required, thereby potentially restricting patients’ access to appropriate testing.

6. Subject Matter Panel and Contractor Advisory Committee (CAC) Meeting on June 26, 2019

We agree with the panel that combinatorial pharmacogenomics tests with a proprietary algorithm that are not available for public review should require independent evidence to establish validity and utility. We believe that proprietary algorithms with gene content that do not have associated CPIC or other society guidelines may make it more difficult to assess their clinical validity and potential utility. Additionally, we appreciate that you view CPIC as a valuable source of pharmacogenomic evidence.

Pharmacogenomic Testing in Psychiatric Disease

Gene-drug interactions

AMP and CAP encourage the use of CPIC’s gene-drug practice guidelines, specifically on gene-drug interactions9 and recommends the use of drug-gene pairs that have high level of evidence (e.g., CPIC’s Level A or B). We would note that the CYP2D6 and methylphenidate pairing is currently assigned a CPIC level of “B/C”.

7. CPT Coding

We suggest that the following CPT Codes be included:

Gene CPT code

TPMT 81335

SLCO1B1 81328

VKORC1 81355

CYP3A5 81231

CYP2B6 81479 (Not otherwise specified)

IFNL3 81283

DPYD 81232

G6PD 81247

UGT1A1 81350

9 https://cpicpgx.org/genes-drugs/

HLA Class II Typing 81382

NUDT15 81306

HLA Class I Typing 81381

In the case that [The Contractor] agrees with us to expand coverage to the additional pharmacogene RYR1 (for more information, please refer to our comments on “Specific Coverage Information” above), we recommend that additional CPT codes added to this policy include, but not be limited to, the following list:

Gene CPT Code

RYR1, targeted sequence of exons 81406

RYR1, full sequence 81408

8. ICD-10 Coding

We request that any ICD-10 code relevant to psychiatric or neurologic disease (such as depression, anxiety, mood disorders, neurologic, etc) or pain be added to this list. Specifically, we request that additional ICD-10 codes added to this policy include, but not be limited to the following list:

B02.22 Postherpetic trigeminal neuralgia

B20 Human immunodeficiency virus [HIV] disease

D57 Sickle-cell disorders

F31.0 Bipolar disorder, current episode hypomanic

F31.10 Bipolar disorder, current episode manic without psychotic features, unspecified

F31.11 Bipolar disorder, current episode manic without psychotic features, mild

F31.12 Bipolar disorder, current episode manic without psychotic features, moderate

F31.30 Bipolar disorder, current episode depressed, mild or moderate severity, unspecified

F31.31 Bipolar disorder, current episode depressed, mild

F31.32 Bipolar disorder, current episode depressed, moderate

F31.60 Bipolar disorder, current episode mixed, unspecified

F31.61 Bipolar disorder, current episode mixed, mild

F31.62 Bipolar disorder, current episode mixed, moderate

F31.70 Bipolar disorder, currently in remission, most recent episode unspecified

F31.72 Bipolar disorder, in full remission, most recent episode hypomanic

F31.74 Bipolar disorder, in full remission, most recent episode manic

F31.76 Bipolar disorder, in full remission, most recent episode depressed

F31.78 Bipolar disorder, in full remission, most recent episode mixed

F31.89 Other bipolar disorder

F45.41 Pain disorders exclusively related to psychological factors

G43-G43.D1 Migraines

G44-G44.89 Headache syndromes

G50.0 Trigeminal neuralgia

G50.1 Atypical face pain

G52.1 Disorders of glossopharyngeal nerve

G54.6 Phantom limb syndrome with pain

G56.4-G56.42 Causalgia, upper limb

G57.7-G57.72 Causalgia, lower limb

G89 Pain, not elsewhere classified

G89.0 Central pain syndrome

G89.2 Chronic pain, not elsewhere classified

G89.21 Chronic pain due to trauma

G89.22 Chronis post-thoracotomy pain

G89.29 Other chronic pain

G89.3 Neoplasm related pain (acute) (chronic)

G89.4 Chronic pain syndrome

G90.5-G90.59 Reflex sympathetic dystrophy

H57.1-H57.13 Localized pain, eye pain

H92.0-H92.09 Localized pain, ear pain

K08.8 Localized pain, tooth pain

K14.6 Localized pain, tongue pain

M25.51-M25.519 Localized pain, shoulder pain

M25.5-M25.579 Localized pain, joint pain

M54-M54.9 Localized pain, spine pain

M54.5 Localized pain, lumbar region pain

M54.9 Localized pain, back pain

M79.10-M79.18 Myalgia

N23 Localized pain, renal colic

N64.4 Localized pain, breast pain

M79.6-M79.676 Localized pain, limb pain

N94.810 Vulvar vestibulitis

N94.81-N94.819 Vulvodynia

R07.0 Localized pain, throat pain

R07.1-R07.9 Localized pain, chest pain

R10-R10.9 Localized pain, abdomen pain

R10.2 Localized pain, pelvic and perineal pain

R30.9 Localized pain, painful urination

R51 Localized pain, headache

R52 Generalized pain NOS

T82.84-T82.848S, T83.84-T83.84XS, T84.84-T84.84XS, T85.84-T85.84XS Pain from prosthetic devices, implants, and grafts

T88.7 Unspecified adverse effect of drug or medicament

Z17.0 Estrogen receptor positive status [ER+]

Z21 Asymptomatic human immunodeficiency virus [HIV] infection status

Z94.0 Kidney transplant status

Z94.1 Heart transplant status

Z94.2 Lung transplant status

Z94.3 Heart and lungs transplant status

Z94.4 Liver transplant status

Z94.81 Bone marrow transplant status

Z94.82 Intestine transplant status

Z94.83 Pancreas transplant status

Z94.84 Stem cells transplant status

In the case that [The Contractor] agrees with us to expand coverage to the additional pharmacogenes RYR1 and CACNA1S (for more information, please refer to our comments on “Specific Coverage Information” above), we suggest that ICD-10 codes for any procedure that would involve volatile anesthetic agents (surgeries) be added. Additionally, we request that specific ICD-10 codes for malignant hyperthermia be added to this policy that include, but are not limited to the following:

T88.3 Malignant hyperthermia due to anesthesia

T88.3XXA Malignant hyperthermia due to anesthesia, initial encounter

T88.3XXD Malignant hyperthermia due to anesthesia, subsequent encounter

T88.3XXS Malignant hyperthermia due to anesthesia, sequela

Thank you again for the opportunity to review and comment on this proposed policy. We are happy to be of assistance in providing additional clinical or other information to assist you with this draft LCD.

Association for Molecular Pathology

College of American Pathologists

Thank you, members of AMP and CAP, for your insightful comments. We will address your questions in order (we have added numbers to each question to facilitate this process).

1. Other CPIC-supported genes not mentioned in the draft policy: Based on a literature review of a subset of well-utilized biomarkers, studies, and comments from the CAC members, the draft policy provided coverage for Pharmacogenomics (PGx) in the appropriate clinical setting that have demonstrated clinical actionable use. This was defined (in the definition section) as PGx markers that are required based on FDA labels for safe drug use and based on the level categorization A or B in the CPIC guidelines. This comment is insightful in that coverage set within the draft policy submitted for comment was limited to only a subsection of either FDA safety markers and/or CPIC guidelines and thus was not internally consistent with the scope of the intent of the policy. Based on our own criteria for actionability (utility), other markers that have met these criteria should be covered in the policy; furthermore, any changes to either FDA label information or CPIC guidelines would render this policy out of date and not compliant with the very metrics used to determine clinical actionability stated in the policy. To remedy this and to clarify the coverage intended in the policy, detailed sections of coverage criteria that mirrored existing CPIC guideline information for a subset of PGx markers have been entirely excised from the policy. This has the net effect of keeping the same scope of coverage and allows the scope to remain relevant as CPIC guidelines and FDA safety information evolve. Coverage for PGx tests is provided in the same clinical context (a drug is being considered for use that has a known drug-gene interaction) and the interaction is clinically actionable based on the FDA or CPIC level A or B guidelines. Additional information about the guidelines has also been added as reference in the policy. PharmGKB is also referenced in the policy as a resource, however, it is not clear that the annotations made are sufficient for a coverage determination.

2. Thank you for these clarifications about specific alleles. The LCD is focused on coverage policy based on a review of evidence and lays the groundwork for establishing coverage based on evidentiary and expert review. Your comment highlights that specific minimum testing criteria that would be evaluated as part of a technical assessment are likely to evolve over time- details such as the specific alleles required for testing is better suited in other formats such as TA forms or coverage Articles that can be more readily modified to keep up with changing science. As such, the minimum criteria are removed from the policy, but the coverage criteria require a technical assessment and states that acceptance criteria are set by experts and professional guidelines. 

3. Thank you for this comment supporting that findings must be understood by the clinician to be clinically meaningful. We have added additional language to clarify exactly what is required. 

4. We hope that the changes made in reference to the comments above will render it less likely that this policy will need to be continuously revisited. However, should the evidence require that we reconsider portions of this policy, we look forward to it. 

5. Documentation criteria. Thanks for the comments regarding the burden the documentary requirements will create. Upon review, we agree that the criteria set are not feasible as written. Specifically, it is unnecessary for the ordering physician to either A) have a documented understanding of the specific gene-drug interaction as a precondition to order the test; or B) have a documented plan for the result of the test to demonstrate that the test was medically necessary. To be reasonable and necessary, the drug that has been selected by the clinician must, aside from being medically indicated clinical reasons, have an established and clinically actionable gene-drug interaction. The clinician does not need to have a geneticist-level understanding to make use of a relevant test- only to know that a test is necessary for a drug in use or under consideration. Further, it is not necessary for the clinician to have a premediated documentation of what course of action they would take as a result of the test. Either the drug may be used safely, or the drug should not be used because it will be contraindicated, or the dosing may require adjustment. We take it as a matter of standard care that the ordering clinician will act accordingly based pm the test results, which as discussed in the policy, must include an interpretive genotype-phenotype explanation. For these reasons, the documentation requirements in the draft policy submitted for comment from the clinician are largely removed. However, the lab or provider that is performing the service must maintain proper documentation to perform the service in question and is responsible for ensuring that the service rendered is reasonable and necessary. As such, the lab must know the drug that is being considered and the indication for its use to ensure the service they will render is reasonable and necessary; this would further be necessary in rendering the proper annotation or interpretation of the result. Regarding the comment about documentation of the ordering physician’s ability to diagnose the stated disease and administer medication- this should be interpreted to mean that the ordering provider should be the qualified person who will treat the patient with the medications in question. There is no specific mention of a documentary requirement.  

6. Thank you for adding additional support for the CAC committee comments and draft policy statements.  

7. (and 8) Thank you for specific CPT and ICD-10 information. We will review these as we finalize the associated Billing and Coding Article for this policy. It should be noted, however, that transplant services are bundled with part A services and as such not typically billable or included with this policy.

We are physicians using pharmacogenetic testing to care for our patients, and we believe it’s imperative to share information with you regarding the Clinical Pharmacogenetics Implementation Consortium (CPIC^®) guidelines (see attachment). CPIC guidelines form the cornerstone of efforts to optimize medication therapy when a patient’s pharmacogenetic test results can be made available. 

“As an emergency physician and toxicologist, I utilize pharmacogenetic testing and CPIC guidelines to help choose medications that are likely to be effective while maximizing safety. For instance, I use CYP2C19 test results to determine which patients should be prescribed the antiplatelet medication clopidogrel in the peri-cardiac catheterization period. In my toxicology clinic, I order HLA genotyping in patients that have developed, or are at risk of developing drug induced Stevens-Johnson syndrome due to carbamazepine or oxcarbazepine. In addition to my clinical practice, I serve as the Chair of the Drug-Gene Prioritization group in the Pharmacogenomic Implementation Committee of Colorado. Using the CPIC guidelines, we choose which drug-gene pairs to implement across the UCHealth system, which cares for half the patients in Colorado. Thus, CPIC guidelines help me care for my individual patients and impact an enormous population across Colorado.”  

“As an anesthesiologist, I find CPIC guidelines essential in minimizing postoperative nausea and vomiting, avoiding prolonged muscle paralysis following succinylcholine administration, minimizing narcotic use and abuse in the perioperative period and reducing the rare but life-threatening incidence of malignant hyperthermia.”

“As a psychiatrist, I have encouraged patients to become enrolled in our Pharmacogenetic Research Clinic where genotyping of drug metabolizing enzymes (e.g., CYP2D6 and CYP2C19) is offered at no cost to patients. I recommend genetic testing particularly each time a patient has failed at least one previous antidepressant medication or has developed intolerable to side effects. Similarly, when patients have not responded on several antipsychotics, or are being treated with multiple medications (polypharmacy) without improving significantly, I suggest pharmacogenetic testing to these patients based on the CPIC guidelines. I find that pharmacogenetic testing with CPIC guidelines helps optimizing therapy in these patients. I have frequently utilized the guidelines developed by CPIC to determine which tests to implement in our clinical practice.”  

[CPIC] is an NIH-funded organization with a membership of more than 300 clinicians, scientists, laboratorians, and others pharmacogenetic experts with the purpose of facilitating the use of pharmacogenetic test results for patient care. One barrier to implementing pharmacogenetic testing in the clinic is the difficulty of translating genetic laboratory test results into actionable prescribing decisions for affected drugs. CPIC’s goal is to address this barrier by creating freely available, peer-reviewed, evidence-based, and updatable gene/drug clinical practice guidelines.  

CPIC uses a rigorous and systematic system to grade levels of evidence, and only gene/drug groupings with strong evidence for actionable prescribing are selected for guideline development. [Guidelines] help clinicians understand how to use available genetic test results to guide prescribing and do not focus on whether to order genetic tests. There are currently 23 such guidelines, including medications relevant to many areas of medical practice. The guidelines are indexed in PubMed as clinical guidelines and supported by many other organizations including the American Society of Health Systems Pharmacists (ASHP), the American Society for Clinical Pharmacology and Therapeutics (ASCPT), and the College of American Pathologists (CAP). Some examples of health systems and other institutions that rely on CPIC guidelines for patient care include the Mayo Clinic, Cleveland Clinic, University of Florida Health, Indiana University, St. Jude’s Children’s Research Hospital, Brigham and Women’s Hospital, Boston Children’s Hospital, Northshore University Health System and many [more].  

Numerous studies have shown that for certain gene/drug groupings, the use of actionable pharmacogenetic information for drug therapy can improve patient outcomes via increased efficacy, reduced adverse events, and improved safety. Enhanced adherence to drug therapy may be an additional benefit. Reduced hospitalizations and other measures of healthcare utilization have been reported, and multiple studies show that pharmacogenetic testing for certain gene-drug groupings can be cost effective or even cost saving. Based on this evidence, recently some payers have proposed expanded coverage for some pharmacogenetic tests.  

Despite recent payer decisions, clinicians still agree that reimbursement is a challenge. Costs for genetic testing are steadily decreasing and thus access to genetic testing is becoming more widespread in the US and worldwide. We hope that you will embrace the CPIC guidelines when you consider policy regarding pharmacogenetic testing. If you have any questions, please contact us.  

Additional drugs and current guidelines:

23 guidelines; 20 genes and 46 drugs

• TPMT, NUDT15

– MP, TG, azathioprine

• CYP2D6

– Codeine, tramadol, hydrocodone, oxycodone, TCAs, tamoxifen, SSRIs, ondansetron, tropisetron, atomoxetine

• CYP2C19

– TCAs, clopidogrel, voriconazole, SSRIs, PPIs (in progress)

• VKORC1

– Warfarin

• CYP2C9

– Warfarin, phenytoin, NSAIDs (in progress)

• CYP4F2

– Warfarin

• HLA-B

- Allopurinol, CBZ, Oxcarbazepine, abacavir, phenytoin

• HLA-A

– CBZ

• CFTR

- Ivacaftor

• DPYD

– 5FU, capecitabine, tegafur

• G6PD

– Rasburicase

• UGT1A1

– Atazanavir

• SLCO1B1

– Simvastatin

• IFNL3 (IL28B)

– Interferon

• CYP3A5

– Tacrolimus

• CYP2B6

– Efavirenz

• RYR1, CACNA1S

– Inhaled anesthetics

• mtRNR1 (in progress)

– aminoglycosides

Thank you for your comments.

The American Psychiatric Association, the medical specialty society representing over 38,500 psychiatric physicians and their patients, is pleased to have the opportunity to comment on the proposed Local Coverage Determination (LCD) for Pharmacogenetics Testing. We appreciate the willingness of Noridian to cover pharmacogenetic testing when it has value for a patient’s treatment. Although the LCD is relevant to medications of multiple classes for multiple disorders, we will focus our remarks on indications for pharmacogenetic testing in the treatment of psychiatric disorders.  

In general, we view several indications as appropriate for pharmacogenetic testing. With some medications, pharmacogenetic testing prior to treatment initiation is important to identify whether a patient is at heightened risk of developing a serious complication. In this context, knowledge of the patient's genetic status can contribute to a decision to avoid use of a specific medication when several possibilities are under consideration. For example, as noted in the LCD, it is important to be able to test for HLA-B*15:02 and HLA-A*31:01 prior to initiating treatment with carbamazepine, oxcarbazepine and phenytoin to detect whether a patient may be at risk for Stevens-Johnson syndrome (SJS)/Toxic epidermal necrolysis (TEN). Although these medications are most often used for their anticonvulsant properties, we would advocate for pharmacogenetic testing to be permitted when these medications are used to treat other diagnoses, including bipolar disorder.  

With other medications, such as those metabolized through cytochrome P450 enzymes, pharmacogenetic testing may be less relevant to initial medication selection but may be important for optimizing medication doses to limit toxicity or enhance outcomes based on principles of pharmacokinetics and known metabolic pathways. In these contexts, pharmacogenetic testing may be indicated once a medication is selected for use or may be more relevant when doses are being adjusted after a patient is already taking a medication. For example, with pimozide, product labelling recommends CYP2D6 genotyping to identify poor metabolizer status before exceeding a dose of 4 mg of pimozide daily in adults or 0.05 mg/kg/day in children. In a patient with partial medication response and no major side effects at the upper end of the typical dosing range, identifying an ultra-rapid metabolizer status via pharmacogenetics would suggest a reason to use higher doses in an effort to improve outcomes.  

We believe that each of these circumstances is alluded to in the LCD but that they should be delineated more explicitly in the sections of the document on Coverage Indications, Limitations, and/or Medical Necessity and Specific Coverage Information. Although we recognize that most indications for pharmacogenetic testing will be related to treatment safety, we would advocate for coverage determinations to also consider enhanced efficacy as a legitimate reason for pharmacogenetic testing in the limited circumstances described above. 

1. We concur with the text of the LCD in noting that:

   Pharmacogenomics testing is considered reasonable and necessary in limited circumstances as described in this Local Coverage Determination (LCD) as an adjunctive personalized medicine decision-making tool once a treating clinician has narrowed treatment possibilities to a small group of specific medications based on other considerations including the patient’s diagnosis, the patient’s other medical conditions, other medications, professional judgement, clinical science and basic science pertinent to the drug, and the patient’s preferences and values. Pharmacogenomics testing is not considered reasonable and necessary merely on the basis of a patient having a particular diagnosis.  

   However, the LCD also comments that:

   … if the record does reflect that the treating clinician has already considered non-genetic factors to make a preliminary prescribing decision, pharmacogenomics testing is not considered reasonable and necessary.  

   We would suggest that non-genetic factors will be important in narrowing the choice of possible treatments and even in selecting a preferred treatment, but that this would not eliminate a subsequent indication for pharmacogenetic testing. For example, if carbamazepine were determined to be the optimal medication for a patient on non-genetic grounds, it would still be important to test for the patient's HLA-B*15:02 and HLA-A*31:01 status before initiating treatment. Similarly, if pimozide were preferred, treatment could be initiated based on non-genetic factors, but pharmacogenetic testing would still be indicated at higher medication doses as recommended in the product labelling. Consequently, throughout the LCD, text should be modified to make clear that pharmacogenetic testing may be indicated during treatment with a specific medication and not simply when a medication is being selected for use.  

   In the discussion of clinical Indications for testing of CYP2D6, CYP2C19, and CYP2C9, we would suggest modifying the general format of the text, with added text denoted by italics:  

   The patient has a diagnosis for which a provider is considering treatment with <<list of medication classes>>, and the patient is open to treatment with such a medication. Alternatively, the patient may be receiving treatment with one of these medications. The patient’s record must reflect this.  

   There must be a specific actionable use (where “actionable use” is defined above) for the result of a <<specified>> genotype in at least one medication that the provider and patient are considering or that the patient is already receiving.

    

   In the discussion of Special Documentation Requirements, we suggest adding a statement to address testing that occurs during treatment with a specific medication rather than prior to treatment initiation. For example, "If pharmacogenetic testing is being ordered to address safety or efficacy of a medication that the patient is already receiving, the record must describe the relevant drug-gene interaction and how the results of testing will influence clinical decision-making." For medications that have readily available serum levels and an evidence base for therapeutic and toxic ranges of serum levels (e.g., imipramine, desipramine, amitriptyline, nortriptyline, clozapine), data on CYP450 enzyme metabolizer status will be less informative than measuring serum levels directly.

2. In terms of combinatorial pharmacogenomics testing, we concur with the expert consensus described in the LCD that independent evidence will be needed to establish the validity and utility of these approaches. Results from well-designed clinical trials with appropriate controls and adequate sample sizes will be particularly important when proprietary algorithms are used that cannot be subjected to independent review. A recent review of combinatorial pharmacogenomic approaches to antidepressant selection (Zeier Z, Carpenter LL, Kalin NH, Rodriguez CI, McDonald WM, Widge AS, Nemeroff CB. Clinical implementation of pharmacogenetic decision support tools for antidepressant drug prescribing. Am J Psychiatry. 2018 Sep 1;175(9):873-886) concluded that "there is insufficient evidence to support widespread use of combinatorial pharmacogenetic decision support tools at this point in time" and noted that "a high level of evidence has been achieved only for the cytochrome P450 genotype data." Even less information is known about possible benefits of combinatorial pharmacogenetics approaches for use of antidepressant medications in conditions other than depression or for use of other psychotropic medications.

3. In addition to these general comments on the indications and coverage of pharmacogenetic testing, we have a number of suggestions on specific aspects of the LCD text:

   In the section on definitions, the definition of actionable use specifies that "selection, avoidance, or dose change must be based on the FDA-label for the drug, an FDA warning or safety concern, or a CPIC level A or B gene-drug interaction." However, the CPIC prioritization flowchart (https://cpicpgx.org/prioritization/#flowchart) shows that genes may have other rationales for testing (e.g., actionable in other professional societies guidelines; recommended by an external group such as FDA; PharmGKB annotation levels 1A, 1B, 2A or 2B) prior to a full determination by the CPIC. We would suggest that these other levels of supporting evidence should also be considered as potentially actionable, particularly with respect to pharmacogenetic aspects of drug metabolism.

4. In the section on general coverage information, the LCD notes that "A multi-gene panel is not considered reasonable and necessary if only a single gene on the panel is considered reasonable and necessary." This restriction does not seem to consider the possibility that, under some circumstances, a multi-gene panel may be less costly either to the payer or to the patient than single gene testing. 

   We have concerns about the restriction that "Genotyping a specific gene is reasonable and necessary only once per lifetime per patient, unless repeat testing is for variants with an actionable use that have not previously been tested in that gene." A clinician who is providing care for a patient may not be aware that a patient has undergone prior genetic testing and, even if aware, the clinician may not have timely access to such results. A clinician who is acting in good faith to assure the safe prescribing of medications to a patient should not be penalized for repeat testing nor should the patient be penalized by denial of coverage.

5. In the discussion of the CYP 2D6 testing, we would suggest changing the older term "neuroleptic" to the more common term "antipsychotic." In terms of specific drugs for which CYP2D6 pharmacogenetic testing may be indicated, we recommend including iloperidone for which the package insert states "The dose of iloperidone should be reduced in patients who are poor metabolizers of CYP2D6." Similarly, a potential need for dose reduction in CYP2D6 poor metabolizers is noted in the product labelling for clozapine. For both clozapine and iloperidone, the CPIC site notes an actionable pharmacogenetic interaction. Additionally, based on our review of the literature and package inserts as part of the new APA Practice Guideline for the Treatment of Schizophrenia (anticipated December 2019 publication), there are also other antipsychotic medications that are metabolized through CYP2D6 and might also have levels affected. These include chlorpromazine, fluphenazine, haloperidol, perphenazine, and thioridazine. This guideline also reviews the use of tetrabenazine, deutetrabenazine and valbenazine in the treatment of tardive dyskinesia, each of which is primarily metabolized through CYP2D6. Product labeling notes that, in CYP2D6 poor metabolizers, doses of tetrabenazine should not exceed 50 mg per day and doses of deutetrabenazine should not exceed 36 mg per day. CYP2D6 poor metabolizers may also be at an increased risk for concentration-dependent adverse reactions, such as QT prolongation, with valbenazine treatment. Thus, addition of these medications to the LCD would be consistent with our guideline as well as with the evidence that these medications are major substrates for CYP2D6.

   Among the antidepressant medications, duloxetine is listed on the CPIC website as having actionable information provided by CYP2D6 pharmacogenetic testing and fluoxetine is listed as having informative information. In the CPIC guideline on selective serotonin reuptake inhibitors (Hicks JK, Bishop JR, Sangkuhl K, Müller DJ, Ji Y, Leckband SG, Leeder JS, Graham RL, Chiulli DL, LLerena A, Skaar TC, Scott SA, Stingl JC, Klein TE, Caudle KE, Gaedigk A; Clinical Pharmacogenetics Implementation Consortium. Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline for CYP2D6 and CYP2C19 genotypes and dosing of selective serotonin reuptake inhibitors. Clin Pharmacol Ther. 2015 Aug;98(2):127-34), fluoxetine is described as having complex metabolism involving CYP2C9 as well as CYP2D6 and no clear data on whether CYP2D6 phenotypes influence the relative levels and effects of fluoxetine and its active metabolite, norfluoxetine. Nevertheless, we would advocate for inclusion of fluoxetine in the LCD for CYP2D6 pharmacogenetic testing because of instructions in the product labelling that CYP2D6 poor metabolizer status may predispose to increased fluoxetine exposure and contribute to QT prolongation. Knowledge of CYP2D6 poor metabolizer status would also be valuable given the prolonged half-lives of fluoxetine and norfluoxetine and relevance of CYP2D6-related drug-drug interactions with many other psychotropic medications.

   In contrast, citalopram and sertraline are listed in the LCD as having a gene-drug interaction for CYP2D6 as well as for CYP2C19, however, the CPIC website does not identify either drug as actionable at CYP2D6 (https://cpicpgx.org/genes-drugs/). In addition, other drug information databases (e.g., Lexicomp?, IBM Micromedex?) do not describe major effects of CYP2D6 in citalopram or sertraline metabolism, although both may function as weak CYP2D6 inhibitors and use CYP2D6 as a minor metabolic pathway.

   For CYP2C19, diazepam is listed on the CPIC website as having actionable pharmacogenetic information. Knowledge of metabolizer status may be particularly useful with diazepam given its long half-life and multiple active metabolites.

   In conclusion, with current evidence, the APA does not view diagnosis, per se, as providing a rationale for pharmacogenetic testing but neither do we think that diagnosis should be used as a reason for excluding coverage of pharmacogenetic testing if it would otherwise be appropriate for a specific medication. For example, making a diagnosis of a depressive disorder should not be a sufficient rationale for ordering pharmacogenetic testing. Instead, pharmacogenetic testing may be useful if a specific medication is considered or used, but only if pharmacogenetic information would aid in making clinical decisions about the use or dosing of that medication. Testing would also be relevant if that same medication were used for a different disorder (e.g. an anxiety disorder). By the same token, testing should be covered for individuals with any diagnosis (including, but not limited to, schizophrenia, bipolar disorder, depressive disorders, anxiety disorders or neurological disorders), if the specific medication is justifiably being used to treat a given condition and if pharmacogenetic testing provides clinically important information about the use of that medication.

   Again, we appreciate the willingness of Noridian to cover pharmacogenetic testing when it has value for a patient’s treatment. As noted in the LCD, we expect that evidence on pharmacogenetics will continue to evolve and we would value opportunities to provide input on future versions of the LCD.

We would like to sincerely thank the APA for their insightful comments and will address each as numbered below.

1. Thank you for the general support for this policy. Regarding drugs that are already in use that may benefit from PGx testing, we have modified the text to allow this. We hope the additional modifications made to this policy based on other stakeholder comments have added additional clarity regarding coverage.
2. Thank you for the support on the current policy for combinatorial testing. As evidence evolves in this field we will monitor and adjust this policy to follow accepted evidence.  
3. While we consider other sources of information such as PharmGKB annotations that review the existing literature, we do not at this time believe that the process used for annotation of variants can be used as a basis for determining policy or is sufficient to demonstrate clinical utility. The CPIC review these same annotations and place considerably more thought and discussion to control the process prior to creating a guideline. 
4. Thank you for these comments. The general coverage criteria describing multi-gene testing has been modified to cover additional possible scenarios. It should be noted that cost on its own is not an acceptable rationale for coverage. Regarding repeat testing, this has been removed from the policy but only because it is redundant with an already existing MolDX policy regarding repeat germline testing. Please review that policy located in the MCD to understand the scope and exceptions to repeat germline testing.
5. Thank you for clarifying gene and drug details. These sections have been removed from the coverage section of the policy. We have incorporated some of your comments in our review of evidence section.

Blank appreciates the opportunity to comment on MolDX: Pharmacogenomics Testing (DL38294). We recognize the significant effort of the MolDX team in seeking expert opinions and developing this draft policy. 

In the second paragraph under “Coverage Indications, Limitations, and/or Medical Necessity”, we suggest removing any mention of combinatorial pharmacogenomics tests, as shown below. 

“This is a limited coverage policy for CYP2D6, CYP2C19, CYP2C9, HLA-B*15:02, HLA-A*31:01, multi-gene panels, and combinatorial pharmacogenomics tests. These tests are generally covered (with a few exceptions) as described in further detail below to increase safety in the use of specific medications by avoiding potentially harmful medications or doses.”

Rationale: The content of DL38294 focuses on interactions between one gene and specific drugs. The Definitions, General Coverage Information, Specific Coverage Information, Noncovered Indications, and Special Documentation Requirements sections of the LCD do not include any information about combinatorial pharmacogenomics tests. Furthermore, as described in detail in our submitted comments on MolDX: Combinatorial Pharmacogenomics Limited Coverage (DL35633), the coverage framework for combinatorial pharmacogenomic tests should be based on the published-evidence supporting their clinical utility and is fundamentally different than that for single-gene tests. Therefore, we believe removing mention of combinatorial pharmacogenomic tests as suggested above will allow both LCDs to serve independently and without conflict. 

We appreciate your thoughtful review of our suggested change.

Thank you for your comments. We considered both the inclusion and exclusion of combinatorial PGx testing in this policy. However, the intent of this policy is NOT to focus on single gene or multi-gene panels but to act as a foundational policy for PGx testing. Combinatorial PGx testing analyzes the same markers and is used for entirely overlapping indications. Furthermore, the draft policy did discuss provider and expert feedback on the utility and necessity of combinatorial testing. We received multiple comments from experts and societies on this policy that explicitly discussed combinatorial PGx testing, including from AMP, CAP, and APA. Their additional thoughts were added to this policy, which also lead us to add additional evidentiary review that has similarly been incorporated into this policy. In the end, we felt there was little reason to maintain two separate PGx-related policies.

I was thrilled to see the MolDX: Pharmacogenomics Testing (DL38294) proposed LCD and how it embraces CPIC guidelines and expanded coverage, which will be so helpful to patients. Many thanks again for including me as a panelist. 

I have some revisions to suggest, and because they range from factual corrections and typos to suggestions for important additional drugs to include, I am running them by you to help me decide which edits would be best for me to submit officially and which you would like to handle just by email. I am attaching a scan of the pages with edits in red to make them easy to glance through.

Page 5

• “Antiemetic” should not be hyphenated.
• For CYP2D6, the ADHD medication atomoxetine has level A CPIC evidence and a CPIC guideline, and I recommend adding it to the LCD (Brown JP et al. 2019)
• For CYP2C19, voriconazole is an important additional medication I suggest including (an anti-fungal, particularly important for leukemia patients). Voriconazole/CYP2C19 has CPIC level A evidence for clinical actionability, and there is a CPIC guideline (Moriyama B et al. 2014).
• For CYP2C9, I recommend adding mention of phenytoin. It is a medication with CPIC level A evidence and a CPIC guideline (Caudle KE et al. 2014 – CPIC guidelines for CYP2C9 and HLA-B genotypes and phenytoin dosing).

I suggest deleting “antidepressant, anxiolytic, mood stabilizer”. I am not aware of any antidepressants or anxiolytics with strong evidence for CYP2C9 actionability. Phenytoin can act as a mood stabilizer in addition to its anticonvulsant role, but it is the only mood stabilizer with CYP2C9 strong evidence.

Page 6

• CYP2C19 needs the gap deleted after CYP
• CYP2D6 This section has some factual errors. *10, *17 and *41 are decreased function alleles, not null alleles. The term “no function” is used in the CYP2D6 literature rather than “null” (Caudle KE et al. 2017. Standardizing terms for clinical pharmacogenetic test results: consensus terms from the Clinical Pharmacogenetics Implementation Consortium (CPIC)).

Decreased function alleles important to test are *9, *10, *17, *29 and *41. The current LCD draft is missing *9 and *29. *9 is present in ~2% of Caucasians, and *29 is present in ~7% of African Americans. (refer to CYP2D6 allele frequency tables in CYP2D6 CPIC guidelines at www.cpicpgx.org/guidelines).

*7 should be deleted from the list of alleles that must be included, since it is rare (present in ~0.07% of Caucasians). 

The sentence about duplications is incomplete, since *1 and *2 are not the only alleles that can have duplications. I recommended instead to state that testing must include assessment for gene duplication. 

In summary, I recommend that this section should read as follows: 

• The following no function alleles must be included: *3, *4, *5, *6 
• The following decreased function alleles must be included: *9, *10, *17, *29, *41  
• Testing must include assessment for gene duplication

Page 7

Under the subheading “Background”

• It is misleading to mention just genetic sequencing technology. The majority of labs use TaqMan technology (real time PCR), which is much cheaper and faster than sequencing (NGS). I would change the first sentence to state “genetic sequencing and other technologies for variant detection”. In the next paragraph, I suggest revising to state “With the advance of sequencing and real time PCR technologies”.

Page 8

• Typos to correct in “Implementation” and “Consortium”.
• After “described by Dr. Annette Taylor as follows” add “and as referenced at https://cpicpgx.org:”. Most of the wording is from the website.
• Delete “see below” (near the bottom of the page).

Page 9

• Add “and voriconazole” in the text.
• Add CYP2C19 Voriconazole to the table.

Page 10

• Add CYP2D6 tropisetron to the table. This is an antiemetic in the same CPIC guideline as ondansetron.

As a general statement, since MolDx is embracing CPIC guidelines, I would suggest including in the LCD all drug/gene pairs that have CPIC guidelines (see attached list of CPIC guidelines). If not included in this LCD, could another MolDX LCD be developed to include them?

Let me know which edits you will take care of without submission.

I would be happy to assist with your review of submitted statements and revisions of the LCD, if that would be helpful. Let me know – and feel free to contact me anytime with any questions. I am delighted to work with you as a PGx SME.

Thank you for the comments and edits. As discussed in response to previous comments, the coverage section discussing specific gene-drug interactions has been entirely removed. This is, as you suggest, because we simplify and clarify this policy to include all FDA and CPIC Level A and B gene-drug interactions.

I have submitted documents before on the lack of evidence in support of pharmacogenetic testing for psychiatric medications. Companies, prominently including Myriad, keep publishing the results of incorrectly designed and interpreted studies comparing outcomes in patients on antidepressants whop receive their tests to patients receiving treatment as usual (TAU). While patients who change medications do somewhat better, that would have happened with reference to any reasonable change ion treatment, especially treatment changes guided by free standard protocols rather than expensive tests. Furthermore, there is no evidence that the genes used in the combinatorial tests determine much about drug choice, and the studies repeatedly show no significant effect on their primary measures. All favorable evidence arises from ‘cherry-picked’ data, a bad analytic procedure. In essence, as has been well documented, the tests appear to have modest value because the patients enrolled are on the wrong medications to begin with, including medications not recommended for standard treatment, and the outcome from using genetic tests is exactly what would be expected from simply going to a standard treatment regimen. 

At this time, all experts with no financial relationship to the genetic test industry, and the American Psychiatric Association, recommend that genetic testing not be used in the treatment of depression. The FDA has determined that genetic tests should not be used to recommend specific drug choices, and they further stated that such use could lead to mistreatment. I believe you have all of the previous documents on those positions. In addition, I have attached a new review that appeared online at the request of Harvard Health Publishing, from Harvard Medical School, addressing the same points.

It would be a large waste of taxpayer money to pay for invalid and possibly misleading tests.

Thank you for your consideration of these points. 

- Harvard Health Blog - https://www.health.harvard.edu/blog - Gene testing to guide antidepressant treatment: Has its time arrived?

Posted By Blank On October 9, 2019 @ 10:30 am In Anxiety and Depression,Health,Mental Health,Tests and procedures | Comments Disabled Depressive disorders are among the most common conditions that disrupt lives. 

Fortunately, medications, psychotherapies, and lifestyle changes are usually successful in treating depression and related disorders, even if symptoms are not entirely eliminated. Sometimes people don’t gain sufficient relief from treatment, or must try several medications before finding one that works well. In an age of exciting advances, including brain imaging and genetic testing, many doctors and patients reasonably hope that new technologies will offer answers. And in fact, for antidepressant choice, several companies sell genetic testing as a means to guide treatment. But do these tests work? 

Genes, gene testing, and depression

Genes determine some of our risk for depression and some of our response to treatment.

However, no single gene or small number of genes determines much of either in the general population. And the few genes used in the current commercial test panels do not appear to be the key genes determining risk or response. Some of the genes tested are related to drug metabolism. These genes can affect drug levels in the blood, but generally don’t predict clinical response. Other factors, including age, diet, hormonal state, gut bacteria, and any other concurrently taken drugs, are far more important in determining how a person metabolizes a drug and responds to treatment.

Most people with depression improve with careful evaluation of all of these factors, appropriate antidepressant choice and dosing according to expert guidelines, as well as follow-up care to monitor treatment response and address any side effects. Currently, there is no scientific evidence that gene tests are needed or would be helpful as part of those assessments. 

What do the studies tell us?

A dozen studies focusing on patients with depressive disorders have reported outcomes from using commercially available gene test panels to guide antidepressant choice. Most studies were completely unblinded — that is, doctors and patients knew a special test was given. Even with that bias, the use of gene results showed no evidence of effectiveness. A few studies were partially blinded, but doctors and patients still knew some patients got a special test. In these studies, too, the tests failed to show value on their key measures of efficacy. Notably, many patients had not responded well before entering a study because they were receiving inappropriate treatments. They improved when switched to more standard treatments. However, the same changes would have been made without guidance from the test if the treating clinicians had simply followed good practice, rather than getting an unproven and expensive genetic test. And our ongoing review of newer studies on these tests suggests similar flaws and no further evidence favoring their use. 

What do experts say about using gene tests to determine treatment for depression?

Against this background, experts with no financial interest in genetic testing have repeatedly recommended that genetic tests should not be used in choosing treatments for depression (see here and here). The American Psychiatric Association convened a task force that reviewed the Harvard Health Blog Gene testing to guide antidepressant treatment... https://www.health.harvard.edu/blog/gene-testing-to-guide-antidep...

1 of 2 10/9/19, 10:48 AM evidence and agreed: the tests should not be ordered. 

Recently, the FDA advised that the tests had no proven value and should not be used. Then they went two steps further, stating that use of the tests could lead to inappropriate treatment choices that might harm patients. Additionally, the FDA sent a warning letter to one company and has contacted others selling the tests, advising them that they cannot legally make specific recommendations to clinicians or patients based on their test results.

Why are gene tests so appealing?

Genetic testing is appealing, both to vulnerable patients and time-constrained doctors. And it is vigorously marketed to both parties by the companies that sell it: through news reports, websites, television, and magazines, and to doctors in their offices. There are few restraints that hold that marketing to the facts, yet the facts are clear in evidence summarized by numerous experts and agencies. Currently available genetic test panels have no proven value for choosing antidepressant treatment, and their use risks providing inappropriate care. So, while gene testing can be very useful for some other conditions, notably some cancer treatments, that success does not yet apply in treating depression. Perhaps this will change with more research, but appropriate tests are years away.

If gene tests aren’t effective, what is?

In the meantime, there are many good and effective actions to take if treatment is not working well. You and your doctor can review your symptoms and diagnosis review side effects of other medications you take to see if this is part of the problem double-check that you are taking the medications correctly consider other factors that might affect your response to treatment, such as alcohol, marijuana use, or other substances change the doses or types of medications based on guideline recommendations, or seek a consultation with an expert. 

When a medication change is needed, the clinician treating you should follow available guidelines (such as these) or help you obtain a consult from a mental health professional who is more knowledgeable about psychiatric medications. Psychiatry consultations are available at most hospitals and clinics; some hospitals offer these consults by phone or through their websites.

Related Information: Understanding Depression

Article printed from Harvard Health Blog: https://www.health.harvard.edu/blog

URL to article: https://www.health.harvard.edu/blog/gene-testing-to-guideantidepressant-treatment-has-its-time-arrived-2019100917964

Copyright © 2017 Harvard Health Publishing Blog. All rights reserved.

Thank you for these comments. Regarding combinatorial testing, we believe these thoughts are in alignment with the final policy presented.

I am a Family Practitioner in Pennsylvania. I have been in Family Practice for over 30 years. I have been certified in Psychopharmacology by the American Society of Clinical Psychopharmacology (ASCP). I am also one of the co- editors of ASCP’s Core Curriculum for Teaching Psychopharmacology to Primary Care Residents and Medical Students.  

Primary care physicians are the first to see and treat psychiatric patients in 70% of the population. I believe that Primary Care Physicians should also be included in the coverage of Psychiatric Genetic Testing as are Psychiatrists. In other words, I do not believe that the proposed policy by Medicare should exclude Primary Care Physicians.  

If further elaboration is needed regarding this view please feel free to contact me.

Thank you for your comments. The final draft of this policy does not preclude any specific provider type from ordering PGx tests. The only requirement is that the ordering clinician also be the one capable of diagnosing the problem and treating the patient.

Please find attached comments on the proposed LCD DL38294 from the Clinical Pharmacogenetics Implementation Consortium (www.cpicpgx.org ). (These comments were submitted via a markup of the LCD PDF)

Thanks for the opportunity to comment.

1. Pharmacogenomics testing is considered reasonable and necessary in limited circumstances as described in this Local Coverage Determination (LCD) as an adjunctive personalized medicine decision-making tool once a treating clinician has narrowed treatment possibilities to a small group of specific medications based on other considerations including Consider making this less restrictive. Many antidepressants (including seven tricyclics and five selective serotonin reuptake inhibitors) are impacted by either CYP2D6, CYP2C19, or both. It is less expensive and more time-effective to test for at least these two genes "up front" than to wait until such time as patients may need to switch to a different antidepressant (either because of lack of efficacy or adverse effects) along with the patient’s diagnosis, the patient’s other medical conditions, other medications, professional judgment, clinical science and basic science pertinent to the drug, and the patient’s preferences and values.

2. Pharmacogenomics testing is not considered reasonable and necessary merely on the basis of a patient having a particular diagnosis. Additionally, if the record does reflect that the treating clinician has already considered non-genetic factors to make a preliminary prescribing decision, pharmacogenomics testing is not considered reasonable and necessary. Rather such testing may be considered reasonable and necessary if a particular treatment is being considered for the patient’s diagnosis, and there is a significant gene-drug interaction of concern. A preliminary prescribing decision based on non-genetic factors should not preclude one from also ordering PGx to ensure the decision is safe and effective.

3. This LCD does not address (provides neither coverage nor non-coverage criteria) pharmacogenomic testing for anticoagulation dosing, which is addressed by National Coverage Determination (NCD) 90.1. The primary focus of this LCD is pharmacogenomics in psychiatric and neurologic conditions, though coverage of pharmacogenomic testing is addressed for other indications as well. Also add, “For example, when the treating clinician decides to use atomoxetine for treatment ADHD due to a comorbid tic disorder (a non-genetic factor), testing for presence of CYP2D6 poor metabolism still guides titrating the treatment dose in a safe manner.”Gene – the term “gene” in this document will be used as a term to encapsulate all of the following: gene, pseudogene, and genetic locus.

4. A multi-gene panel is considered reasonable and necessary if any two single genes on that panel would be considered reasonable and necessary. A multi-gene panel is not considered reasonable and necessary if only a single gene on the panel is considered reasonable and necessary, unless the multigene panel includes more actionable variants per gene that the single gene tests do.

   Genotyping a specific gene is reasonable and necessary only once per lifetime per patient, unless repeat testing is for variants with an actionable use that have not previously been tested in that gene, or there is reason to suspect the integrity of the prior test (e.g. sample mix-up).

   This is a limited coverage policy for CYP2D6, CYP2C19, CYP2C9, HLA-B*15:02, and HLA-A*31:01.

5. Clinical Indications

   Single gene testing for CYP2D6 is considered reasonable and necessary when the following conditions are met: Please note that CYP2D6 is also the subject of actionable CPIC guidelines for atomoxetine, codeine, ondansetron, tramadol, and tropisetron.

   The patient has a diagnosis for which a provider is considering treatment with an antidepressant, anxiolytic, mood stabilizer (including affective diseases associated with neurodegenerative conditions), a medication treating nociception or pain, an anti-emetic, stimulant or a neuroleptic please use the term “anti-psychotic” instead of neuroleptic, and the patient is open to treatment with such a medication. The patient’s record must reflect this.

   Single gene testing for CYP2C19 is considered reasonable and necessary when the following conditions are met: Please note that CYP2C19 is also the subject of actionable CPIC guidelines for clopidogrel, voriconazole, and (coming soon) proton pump inhibitors such as omeprazole.

   Single gene testing for CYP2C9 is considered reasonable and necessary when the following conditions are met:

   The patient has a diagnosis for which a provider is considering treatment with an antidepressant, anxiolytic, mood stabilizer nonsteroidal anti-inflammatory drug or phenytoin, or Mayzent, and the patient is open to treatment with such a medication. The patient’s record must reflect this. We are unaware of data supporting use of CYP2C9 for antidepressants, anxiolytics, or mood stabilizers, but CYP2C9 is subject to CPIC guidelines on warfarin (covered under another LCD), phenytoin, and NSAIDs.

   Single gene testing for HLA-B*15:02 is considered reasonable and necessary when the following conditions are met:

   The patient has a diagnosis for which a provider is considering treatment with carbamazepine, oxcarbazepine, fosphenytoin or phenytoin (testing included FDA label) and the patient is open to treatment with such a medication. The patient’s record must reflect this.

   CYP2C9

   The following alleles must be included at a minimum: *2, *3, *5, *6, *8, *11 AMP has set minimum guidelines for CYP2C9 testing: https://jmd.amjpathol.org/article/S1525-1578(18)30594-4/fulltext

   With the advance of sequencing technology, there is little question that such testing is now technically feasible, but for a test to be reasonable and necessary there must be sufficient evidence that it provides incremental information that changes physician management recommendations in a way that improves patient outcomes.

   The Clinical Pharmacogenetic Implementation Consoritum Implementation Consortium (CPIC) was described by Dr. Annette Taylor as follows:

   At present, CPIC has a number of guidelines for the dosing of drugs based on gene-drug interactions.¹³ The gene-drug interactions have been assigned a level A or B by CPIC are given in the table below. This list does not include all drugs, but is limited to drugs that are or can be used to treat psychiatric or neurologic conditions in addition to clopidogrel. Not clear why clopidogrel is singled out; there are many other drugs/genes that are CPIC level A or B, including those mentioned on page 5 that are CYP2D6, CYP2C19, or CYP2C9 substrates.

   The following gene-drug interactions have a level A rating by the Clinical Pharmacogenetic Implementation Consoritum Pharmacogenetics Implementation Consortium (CPIC):

   HLA testing

   Two specific HLA alleles are recommended for testing.¹⁵ They are HLA-B*15:02 and HLA-A*31:01. Both of these have level A or B gene-drug recommendations from CPIC as noted above. The same HLA test being performed to assess 15:02 and 31:01 could detect the HLA-B 57:01 variant (needed for abacavir prescribing) and HLAB 58:01 variant (needed for allopurinol).

6. While there are some large studies in pharmacogenomics, we are not aware of large high quality studies that used a clear evidence-based prescribing approach in the control arm. As a number of prominent psychiatrists with expertise in the biological underpinnings of mental health have pointed out, it is not clear that pharmacogenomics testing is a better tool for drug selection than using a standardized evidence-based protocol which does not rely on genetics, or a consultation from a knowledgeable provider. We are unaware of any studies that used a clear standardized evidence-based protocol in the control arm. Moreover, psychiatric conditions have significant complexity involving many genes and factors outside of pharmacogenomics. As such, while the evidence does suggest that pharmacogenomics testing can be used to refine the selection of a medication or dose, there is not sufficient evidence to suggest that pharmacogenomics is reasonable and necessary for the initial narrowing or selection of potential medications to treat a patient. But it would be reasonable to allow clinicians to use genotype in the initial selection or narrowing of agents if any TCA or SSRI was being contemplated, and not force patients to first fail on an antidepressant before allowing for testing.

   Since the medical necessity for testing a gene in pharmacogenomics can only come from the ability of that test to inform a management decision based on a gene-drug interaction, a test for a specific gene or allele is not reasonable and necessary unless and until a clinician is considering using a drug that has an interaction with a specific gene or allele. Once this initial decision has been made it may reasonable and necessary to test for that allele or that gene. For example, a clinician using a treatment approach based on the STAR*D study, may consider use of citalopram as a first line treatment. Citalopram is considered to have an actionable CYP2C19 gene-drug interaction. We are unaware of an actionable drug-gene interaction for CYP2D6 for citalopram. Alternatively, if the clinician were considering the use of fluvoxamine, this has an actionable drug-gene interaction for CYP2D6 but not for CYP2C19. If a clinician is not considering using a medication with an interaction with a gene being tested, or if the patient who is being tested is unable or unwilling to use a medication interacting with the gene or allele being tested, then there is no benefit to the patient to run a test of that gene or allele. It is true that only CYP2C19 is relevant to Step 1 of the STAR*D algorithm, but both CYP2C19 and CYP2D6 are relevant to medications in Step 2 (i.e., sertraline, venlafaxine). CYP2D6 is relevant to medications recommended in Steps 3-5 (i.e., nortrptyline, venlafaxine). Testing for both genes upfront will better inform use of the STAR*D algorithm.

   Following up on blank’s earlier comment, CYP2C19 and/or CYP2D6 involve the clearance of several drugs used in psychiatry. Thus, one argument for conducting testing for both genes simultaneously would be to inform the selection of an alternative medication should treatment with the first agent not produce the desired response

ICD-10 Codes that Support Medical Necessity Group 1 Paragraph:

The following diagnosis codes are covered for Group 1 CPT^®/HCPCS codes if the treatment for the diagnosis involves the treating clinician considering treatment with a medication that makes the service billed from Group 1 reasonable and necessary. The ICD-10 code for ADHD, F90.0, which can establish the need for use of atomoxetine should be added.

We suggest the following ICD-10 codes be added:

ICD-10 Codes that Support Medical Necessity, Group 1 Paragraph: page 21

- Late onset Alzheimer's disease with behavioral disturbance (HCC) [G30.1, F02.81]

- Early onset Alzheimer's disease with behavioral disturbance (HCC) [G30.0, F02.81]

- Lewy body dementia with behavioral disturbance (HCC) [G31.83, F02.81]

- Vascular dementia with behavior disturbance (HCC) [F01.51]

- Dementia associated with alcoholism with behavioral disturbance (HCC) [F10.27]

- Dementia due to Parkinson's disease with behavioral disturbance (HCC) [G20, F02.81]

- Other frontotemporal dementia with behavioral disturbance (HCC) [G31.09, F02.81]

- Dementia associated with other underlying disease with behavioral disturbance (HCC) [F02.81]

Reasoning /Rationale on including above ICD-10 codes

• Patients with above mentioned diagnosis often suffers from MDE (Major Depressive Episode) and other symptoms of BPSD (Behavioral and Psychological manifestation of Dementia) warranting use of medication linked to major drug gene interaction.
• Very often these patients are empirically tried on medications mentioned on page – 9,10,11 (medications with major drug gene interaction)
• In addition, these patients are older and are on multiple other medications (not listed) which could alter enzyme activity (phenoconversion)
• Because of cumulative co-morbid conditions and age related pharmacokinetic changes these patients are at high risk for side-effects or drug failure.
• Above mentioned diagnosis narrows down the list to increase specificity of diagnosis covered and all of them have behavioral manifestation attached to it.

This could further help in designing studies around above dx and collect evidence.

Thank you very much for the detailed feedback and comments. We will address your comments in order.

1. Although the wording of this statement has been altered (see above comments) to allow for testing when appropriate for drugs that are already in use, there is no time requirement as a part of the coverage criteria that would result in forcing initiation of unsafe treatments before submitting for testing. Instead, the requirement is that the clinician has an intent to treat and is considering the use of a qualifying drug. If more than one gene is relevant to this process, a panel can be used. The resultant data can be used for a host of other subsequently considered medications.

2. A preliminary treatment decision does not preclude the use of PGx testing in the final version of the policy. 

3. This policy is no longer primarily focused on any indicated use of PGx testing. 

4. Consistent with other MolDX policies, panels are required when single gene testing is not sufficient to safely administer the drug in question or where multiple drugs are being considered that have different genes requiring testing. The density of markers per gene on a panel is not a reason to perform a multi-gene panel if only a single gene is needed for testing from a clinical need perspective. If a given single gene test does not include all the needed information for that gene for the given indication, then that test would not have sufficient utility for coverage. If a multi-gene panel is used for convenience when only a single gene is necessary, that service performed can be billed only for what was reasonable and necessary- here a single gene test- according to this policy. 

5. Thank you for these detailed clarifications. As noted in response to earlier comments, we have removed the specific gene-drug information from the coverage section. However, we incorporated some of your edits into the section describing common genes tested. 

6. If the clinician is considering BOTH sertraline and venlafaxine for therapy, then it would be reasonable to test for a gene panel that includes AT LEAST CYP2D6 and CYP2C19 based on the current language of this policy. 

7. Thank you for the additional list if ICD-10 codes.

I fully support the use of pharmacogenomic testing for psychiatric patients who fail to respond to initial treatments. There are no guidelines for what to do after a number of treatments fail, and the current practice is based on anecdote, training and personal experience without data to guide this. Pharmacogenomic testing can provide data to guide next choices. In addition to pharmacokinetic genes, and often supplemented with plasma drug levels, the pharmacodynamic genes offered can create useful hypotheses for prior treatment failure, and show where the balance of the evidence lies in selecting the next option. No pharmacogenomic test will tell what drug to use all the time with assurance of efficacy, nor what drug to avoid with assurance that it will not work or always cause side effects. Pharmacogenomics helps supply information so that the selection of new treatment options is informed by the balance of the evidence which informs whether a given option has a weight towards working or not, or towards side effects or not. Numerous studies show that pharmacogenomic informed treatment decisions cause clinicians to think more rationally, and with neurobiologic empowerment, and should be available in clinical practice today in my opinion.

Thank you for your thoughtful comments. We will continue to assess evidence regarding pharmacodynamic interactions. And utility in selecting new therapies post treatment failure.