Private Payer and Medicare Coverage for Circulating Tumor DNA Testing: A Historical Analysis of Coverage Policies from 2015 – 2019

1.0. Background

In recent years, there has been a rapid rise in the use of blood-based circulating tumor DNA (“ctDNA”) sequencing tests for cancer, with the first commercially available test offered by Guardant Health in 2014. So-called “liquid biopsy” tests are used to select targeted therapy and monitor non-responding or progressive tumors. Although the majority of somatic (tumor) testing is performed on resected biopsy specimens, ctDNA tests can be used at the time of a cancer diagnosis to identify genomic alterations that may be effectively treated with targeted therapies. Further, they are frequently used after a patient progresses on targeted therapy to determine the mechanism of therapeutic resistance and following definitive therapy to detect recurrence. [1–9] Blood-based assays are particularly valuable in situations where testing needs to be performed rapidly and when tissue-based samples are unavailable.

The objective of this study is to review coverage policies for ctDNA-based testing panels in cancer (“ctDNA”) to examine trends from 2015 to 2019, with 2015 being the first instance of an available payer coverage policy. Payer coverage for genomic medicine continues to be an important determinate of access to testing including whether patients are tested and how they are tested. These types of coverage decisions are a key element in care as testing and the subsequent use of targeted therapy ultimately influence patients’ clinical outcomes.[10–13] This study is the first to analyze current coverage of private and Medicare policies for ctDNA-based panel tests and provide historical analysis.

The growth and adoption of ctDNA-based testing for cancer is expected to increase as well as the number of genes included in the test (e.g. Guardant360 – 54 genes in 2014, now includes 73 genes 2020).[14] In addition, similar “liquid biopsy” techniques are also being applied beyond targeting cancer treatments to early detection of cancer, which will further increase the scope of these tests as their use expands to healthy populations. In addition to providing tumor information, many of these assays may identify non-tumor germline cell-free DNA (cfDNA) (e.g. BRCA1/2 mutations) but reporting of such findings are inconsistent across labs.[15] Thus, our study addresses broader important clinical and policy issues that are emerging. Our systematic assessment of coverage policies provides greater insight into the development of coverage policies and underscores how payers make coverage decisions over time.

2.0. Methods

We analyzed private, commercially available payer coverage policies, Medicare (National Coverage Determination (NCD)) and Medicare Administrative Contractor coverage policies (Local Coverage Determination (LCD)) for ctDNA-based panel tests for cancer from 2015 through July 1, 2019.

2.1. Data Sources and Payer Cohorts

We used the Canary Insights Database (See Supplement) as the source of private policies and the Centers for Medicare and Medicaid Medicare (CMS) coverage database for Medicare LCDs and NCDs (https://www.cms.gov/medicare-coverage-database/search/advanced-search.aspx). The Canary Insights Database (http://canaryinsights.com/) is a medical policy library containing over 40,000 medical policies from over 200 commercial payers and links to public payer policies with over 100,000 members.

The private payer cohort was determined by including all payers that had a positive or negative coverage policy for ctDNA in the Canary Insights Database. The Medicare payer cohort was determined by reviewing all local and national Medicare policies on the CMS website with a positive coverage policy for ctDNA (https://www.cms.gov/medicare-coverage-database/search/advanced-search.aspx). Medicare does not issue negative coverage policies.

2.2. Types of Coverage Policies

2.2.1. Private

These are coverage policies issued by private companies for commercially available plans. They do not cover Medicare Advantage or payers that do not make their policies publicly available (e.g., Kaiser Permanente).

2.2.2. Medicare

These are coverage policies issued directly from CMS in the form of NCDs and by MACs (Medicare Administrative Contractors) in the form of LCDs. NCDs and LCDs are first issued as drafts before they are finalized. Draft policies are available for a public comment period and revision prior to the issuance of a Final NCD or LCD. There are seven MACs (in 12 jurisdictions) in the US that issue relevant coverage policies.

2.3. Search Strategy and Policy Selection

Policy titles and text were individually screened to determine if they met inclusion criteria. We included policies that specifically addressed ctDNA-based panels as a clinical diagnostic test and excluded policies that addressed single-gene ctDNA testing (e.g. EGFR, ALK, BRAF) that did not include a provision for ctDNA-based panel testing (See supplement).

2.3.1. Private Payers

We searched the Canary Database with the terms “liquid biopsy” and “circulating tumor DNA.” We confirmed that these terms captured all relevant policies from 2015 through July 1, 2019 (see supplement).

2.3.2. Medicare

We searched the CMS database for NCDs and LCDs with the terms “liquid biopsy” or “circulating tumor DNA” using the “all states” and “entire document” function. We confirmed that these terms captured all relevant policies from 2018 (the date of the first positive policy) through July 1, 2019.

2.4. Variables and Data Abstraction Policies

We chose variables that described the structure of the policy (e.g. how the policy was written, whether the policy was developed by a third party). For positive coverage policies, we chose variables that described how the coverage was worded, which cancers were included, if and how monitoring was included, and what specific ctDNA tests were covered. The specific variables that were coded for private and Medicare policies are in the Supplement Material.

We coded variables for Private and Medicare Policies for 2017 – 2019. We also coded if a policy existed for 2015–2016 and if it provided coverage (none provided coverage). Data were independently coded by two individuals and discrepancies resolved by discussion. We describe trends but we did not statistically assess differences. Trends examined include the changes to policies (e.g. non-coverage to coverage) and detailed conditions of coverage (e.g. test, genes, types of cancer)

3.0. Results

As of mid-2019, a total of 65 private payers and four MACs had published policies about ctDNA. We found a shift in private payers from no coverage in 2016 to 38% coverage in 2019, and a shift in Medicare policies from no LCDs for the use of ctDNA-based panel tests in cancer indications in 2017 to eight final LCDs, three draft LCDs, and one future effective final LCD (2/3/2020) in 2019. From 2017 to 2019, policies have increased in scope regarding the number of cancers included (from NSCLC to 12 solid/hematological cancers) and from a single gene (EGFR) to 73 genes (e.g. Guardant360).

Specifically, we found that 38% (28/73) of private payer coverage policies provide coverage of ctDNA-based panel testing for some clinical indications as of 7/1/2019 (Table 1). Positive coverage polices are from 28 payers, of which 22 are Blues Plans and negative coverage policies are from 45 payers, of which 25 are Blues Plans. Of the 200 payers in Canary Insights Database, 33% of private payers have explicit positive or negative coverage policies.

Table 1:

Coverage of ctDNA by Type of Payer, Year, and Cancer Type

	Coverage Policies
Payer	2015	2016	2017	Type of Cancers Covered (N)	2018	Type of Cancers Covered (N)b	2019	Type of Cancers Covered (N)b
Private	0/6 Positive (0%)	0/28 Positive (0%)	1/42 Positive (3%)	1 lung cancer only	13/66 Positive (20%)	11 lung cancer only; 2 Solid and hematologic malignancies	28/73 Positive (38%)	24 lung cancer only; 4 Solid and hematologic malignancies
Medicarea	0 Positive	0 Positive	0 Positive	N/A	4 Positive	4 lung cancer (Guardant360 only)	12 Positive	8 lung cancer - 4 Guardant360 - 4 InVisionFirst-Lung; 4 solid tumorsc - Guardant360 only

NOTES:

^aMedicare coverage is provided by LCDs issued from 4 of 7 Medicare Authorized Contractors. Medicare does not issue negative coverage policies so we did not calculate a percentage of policies with positive coverage.

^bBCBS-affiliated policies typically have two separate policy documents – one for lung and one for all other cancers. These policies are counted individually.

^cSolid Tumors (12 types): Non-small Cell Lung Cancer (NSCLC), Colorectal, Breast, Endometrial, Gastric and Gastroesophageal, Gastrointestinal Stromal Tumor, Melanoma, Ovarian, Pancreatic, Prostate, Thyroid, Chordoma

Of the positive policies, 87% (24/28) were for NSCLC only (Table 2) and nearly half (47%, 13/28) were written for testing EGFR Gene Analysis only (Table 3). Of those, 79% (22/28) named coverage only for specific tests (Cobas, Guardant360, or OncoBEAM) (Tables 2 and 3). Of those policies providing coverage, 11% (3/28) specifically state coverage for monitoring (e.g. no prior testing and progression, or progression on Tyrosine Kinase Inhibitors)(Table 2). Descriptions of all brand-named tests are in the Supplement.

Table 2:

Characteristics of 2019 Positive Coverage Policies by Payer

Category	Private Payer Policy Characteristics	No. (%) Policies (N=28)	Medicare Policy Characteristics	No. Policies (N=12)
Cancers Included	NSCLC - not defined - Stage III or IV - Stage IIIB or IV - Metastatic or Recurrent - Adv. Stage III, IV, or Metastatica	24 (86%) TOTAL - 11 (39%) - 8 (29%) - 3 (11%) - 1 (4%) - 1 (4%)	NSCLC (Stage IIIB or IV)	8
	Solid and hematologic malignancies	4 (14%)	Solid Tumorsb	4
ctDNA Test Covered	Single gene EGFR analysis using -Cobas EGFR Mutation Test v2, -Guardant360, or -OncoBEAM	12 (43%)	Comprehensive genomic analysis using InVisionFirst-Lung	4
	Comprehensive gene analysis using Guardant360	10 (36%)	Comprehensive gene analysis using Guardant360	8
	Multigene panel not defined	6 (21%)
Monitoring Covered	Monitoring not addressed	25 (89%)
	No prior testing and progression or progression on TKIs	2 (7%)	Not previously tested, or not responding to EGFR TKIs	8
	At progression on EGFR TKI inhibitors	1 (4%)	New primary cancer or different primary cancer and untreated or not responding	4

^aIncludes Adenocarcinoma, Large Cell, Squamous Cell, and NSCLC not otherwise specified

^bSolid Tumors (12 types): Non-small Cell Lung Cancer (NSCLC), Colorectal, Breast, Endometrial, Gastric and Gastroesophageal, Gastrointestinal Stromal Tumor, Melanoma, Ovarian, Pancreatic, Prostate, Thyroid, Chordoma

Table 3:

Coverage Clinical Scenarios by Private Payer and Medicare

Clinical Scenario	Cancer Type	Gene(s) Included	Test(s) Included
	Private Payer Policy Characteristics			N (%)
1	Solid and hematologic malignancies	Genes not defined	Multigene panel not defined	4 (14%)
2	NSCLC	EGFR gene analysis	Cobas EGFR Mutation Test v2, Guardant360, or OncoBEAM Only	12 (43%)
3	NSCLC	Genes not defined	Guardant360 Only	8 (29%)
4	NSCLC	EGFR, ROS, ALK, PDL1 gene analysis	Guardant360 Only	2 (7%)
5	NSCLC	EGFR gene analysis	Multigene panel not defined	1 (4%)
6	NSCLC	Genes not defined	Multigene panel not defined	1 (4%)
	Medicare Policy Characteristics			N
7	Solid Tumorsa	Genes not defined	Guardant360b	4
8	NSCLC	EGFR, ROS, ALK, BRAF	Guardant360 Only	4
9	NSCLC	EGFR, ROS, ALK, BRAF	InvisionFirst Test Only	4

^aSolid Tumors (12 types): Non-small Cell Lung Cancer (NSCLC), Colorectal, Breast, Endometrial, Gastric and Gastroesophageal, Gastrointestinal Stromal Tumor, Melanoma, Ovarian, Pancreatic, Prostate, Thyroid, Chordoma

^bOther liquid biopsies will be covered for the same indications if they display similar performance in their intended used applications to Guardant360®

We found wide variation in how payers structured their policies, that is, whether the policy focused on the technology generally or focused on a specific cancer or test (Table 4). Private payers used nine different approaches. The most common structure used by private payers was to organize their policy around “liquid biopsy” tests, including both ctDNA and Circulating Tumor Cell tests (36%). Conversely, Medicare used two different approaches. The most common structure used by Medicare payers was to organize their policy based on a specific named test (100%) (e.g. Guardant360 or InVisionFirst) and then narrowly on ctDNA NCSLC tests (67%).

Table 4:

Policy Structure Types Issued by Private Payers and Medicare

Policy Structure (focus of policy)	Private Payer No. (%) Policies (N=73)	MedicareaNo. (%) Policies (N=12)
Liquid Biopsy Overall	26 (36%)
3rd Party Policyb	12 (16%)
ctDNA NCSLC	11 (15%)	8 (67%)
Tumor Markers Overall	10 (14%)
Genetic Testing Overall	6 (8%)
ctDNA Solid Tumors	3 (4%)	4 (33%)
Test Specific – Guardant360	1 (1%)
Expanded Molecular Testing Panel Overall	1 (1%)
Other (e.g. non-covered services, Medical Technologies Database)	3 (4%)

^aMedicare does not issue negative coverage policies so we did not calculate a percentage of policies with positive coverage

^b3rd party policies are those that are managed by a laboratory benefit management company (e.g. Evicore or AIM).

There are no NCD policies explicitly providing coverage for ctDNA-based panel tests. Although the NCD on sequencing for advanced cancer[1] includes ctDNA-based panel test coverage if the test is FDA approved, there are currently no FDA approved ctDNA-based panel tests.

Although there is no national coverage policy, MACs can provide coverage to ctDNA-based tests with the issuance of an LCD. We found 12 draft or final LCDs (67% final, 33% draft LCDs) covering ctDNA-based panel tests (Table 1) from four MACs (out of seven). Four MACs issued final LCD policies in 2018 for the Guardant360 ctDNA-based panel test in NSCLC. The same four MACs issued final LCD policies in 2019 for the InvisionFirst ctDNA-based panel test in NSCLC (Table 2). Between March and May 2019, the same four MACs issued draft LCD policies (Table 1) that will cover the use of Guardant360 in ctDNA-based panel testing in 12 solid tumors, in addition to the already approved coverage policy for NSCLC (Table 2), when finalized. On December 19, 2019, one final LCD was issued by Palmetto GBA that finalized the LCD effective February 3, 2020, with a key addition that “other liquid biopsies will be covered for the same indications if they display similar performance in their intended used applications to Guardant360”

The first instance of a positive coverage policy was by Blue Cross Blue Shield Massachusetts in September 2017 for Guardant360 ctDNA-based panel test in NSCLC including the use of monitoring (data not shown). In 2018, there was a sharp increase in coverage for both private payers and Medicare, especially after the NCD on sequencing for advanced cancer[1] was issued in March.

4.0. Discussion

The overall trend in both private payer and Medicare coverage for ctDNA-based panel testing is an increasing number of coverage policies, an increasing number of positive policies, and an increasing scope of coverage. We found the majority of positive coverage policies are very specifically written with defined clinical scenarios, although we also found that Medicare coverage policies are evolving to pan-cancer uses, signifying a significant shift in coverage frameworks.

The majority of private payer positive coverage policies are still very specifically written with defined clinical scenarios. As might have been expected given that the only FDA approved ctDNA test is a single gene test for NSCLC, we found that the majority (87%) of policies were on NSCLC and that nearly half (47%) were for EGFR gene analysis. Further, of those policies on NSCLC, the majority (79%) only covered specifically named tests. One of the most interesting findings was in the case of EGFR gene analysis in which 43% (12/28) payers stated that multigene panel tests (Guardant360 and OncoBeam) would be a covered benefit for EGFR gene analysis only. Given that tests such as Guardant360 and OncoBeam are panel tests that evaluate and multiple genes, the “limited” coverage decision may actually result in testing that is far more comprehensive than intended.

We found that Blue Cross Blue Shield Association (BCBSA) plans were heavily represented in our sample but that such plans varied in their coverage policies. Over half (64%) of the private payer policies were BCBSA plans, comprising 56% of the covered lives for all policies included. Of these, 22 provided coverage, representing 76% of all positive covered lives and 25 did not provide coverage. Although the national BCBSA organization issued a technology assessment for “Circulating Tumor DNA and Circulating Tumor Cells for Cancer Management”, BCBSA plans operate independently and each plan determines its own policies.[10, 16]

One complexity of private payer coverage is that payers structure their policies in many ways ranging from policies focusing on very specific clinical conditions, genes and brand-named tests to broad policies that cover many conditions and generic tests. This wide variation reflects the fact that private payers vary greatly in how they develop and structure their coverage policies.[17, 18] Such variation increases the challenges in assessing and tracking coverage policies and may make it difficult for patients and providers to understand relevant policies.

An important finding is that Medicare coverage policies are evolving rapidly. While there is no NCD policy explicitly providing coverage for ctDNA-based panel tests (other than for FDA-approved tests), coverage is provided through LCDs. The policy framework for Medicare LCDs is evolving from coverage of specific cancers to policies providing coverage of pan-cancer scenarios, which marks a significant change from earlier LCDs. Early LCDs were written in a very specific manner in which each LCD addressed a specific brand-named test in a particular cancer. For example, four MACs issued final LCD policies in 2018 for the Guardant360 ctDNA-based panel test in NSCLC and the same four MACs issued final LCD policies in 2019 for the InvisionFirst ctDNA-based panel test in NSCLC. However, in early 2019, the same four MACs issued draft LCD policies that would provide pan-cancer coverage in 12 solid tumors. One MAC has since issued a final LCD (effective 2/3/2020). This coverage is still brand-name specific as it covers Guardant360, but the move towards pan-cancer coverage is significant and includes a provision that “other liquid biopsies will be covered for the same indications.” We found in earlier studies that coverage of pan-cancer tests conflict with many of the concepts of payer coverage frameworks including that sequencing tests are considered a “bundle” of individual gene tests, which is counter to the one-marker-one-drug evaluation of medical necessity.[18, 19]

We did not find any previous studies of payer coverage policies for ctDNA, but our results are consistent with other studies showing that coverage for sequencing tests is increasing in some clinical scenarios but that many tests are not covered. Our study evaluating payer coverage policies for exome sequencing in pediatric patients with neurodevelopmental disorders found that 50% of private payers changed from no coverage to positive coverage between 2015 – 2017[10] and our recent study of payer decision making found that private payers are increasingly willing to cover such tests.[13] However, our recent study of NGTS coverage policies found that half of private payers are not providing coverage for such tests as of mid-2019.[16]

4.1. Implications and Future Research

Our results suggest that there is a conundrum between what the test actually evaluates, what the payer is willing to cover, the information that the clinician receives, and can use to guide clinical decisions. As noted above, several payers provide coverage for multigene tests but only allow the analysis of a single gene from those tests. For example, a clinician can order a particular test that interrogates 73 genes (e.g. Guardant360), but the payer will cover that test but only for analysis of the single EGFR gene (43% of payers provide coverage in this manner), which leads to a report that may have results that are not covered by the payer but may still be used to inform the clinician’s ability to properly manage the patient’s cancer (e.g. reports are identical independent of payer coverage policy). Additionally, limited policies, or having no policy, may burden clinicians with prior authorization requirements and appeals and thus delay or preventing access to testing. These issues will require further study. Another conundrum is that payer policies may not reflect what is found in reviews and guidelines. For example, the American Society of Clinical Oncology and College of American Pathologists published a review in 2018 that found no evidence of clinical utility and little evidence of clinical validity of ctDNA assays in early-stage cancer, treatment monitoring, or residual disease detection.[7] The National Comprehensive Cancer Network Guideline indicates that ctDNA testing can be considered when the patient is medically unfit for invasive tissue sampling or if following pathologic confirmation of an NSCLC diagnosis there is insufficient material for molecular analysis.[20] Additionally, studies that support clinical utility will likely result in greater consistency among coverage policies.

Another conundrum is that the technology used by some labs to detect ctDNA tests interrogates all cell-free DNA (cfDNA). The cfDNA contains germline DNA, which can provide insight into both related and non-related cancer susceptibility (e.g. BRCA1/2 mutations in breast cancer vs. BRCA1/2 mutations in colon cancer). In addition to the ctDNA analysis, some laboratories provide the clinician with the results of cfDNA germline analysis for select genes. This fact presents a challenge given that these results may be unexpected and can confront both providers and patients with ethical and clinical dilemmas.

While there has been a plethora of coverage for the use of ctDNA testing in NSCLC from both the private payers and Medicare, the coverage is primarily for treatment selection with only 11% of payers with positive coverage including any monitoring indications. We expect that coverage for additional clinical indications such as monitoring of minimal residual disease after definitive therapy or detecting recurrence may eventually be added, albeit payers may be reluctant to adopt these indications as this would expand use to move extensive serial testing over time and thus may greatly increase costs.

4.2. Limitations

We did not include Medicaid policies and the Canary Insights Database does not include all policies from all private payers in the US as it excludes self-insured and small payers. However, the Database includes more than 200 payers and our analysis reflects coverage policies for 75% of the US population including the 172 million private payer covered lives and the 75 million Medicare enrollees in 2019. Second, we were limited by the amount of information provided in the coverage policies by each payer, which were highly variable in their detail and clarity. We could not examine the actual decision-making process undertaken by individual payers. Third, published payer coverage policies do not necessarily reflect actual coverage or reimbursement for all “covered” tests as plan purchasers can elect to exclude coverage for certain tests when purchasing plans for their employees. This is particularly true for self-insured groups, where the insurer acts as a third-party administrator. Last, Medicare only issues positive coverage policies, which limits the comparison both in policy content and in timing, as the first Medicare policy was issued in 2018 while we included private payer policies from 2015–2019.

5.0. Conclusions

In sum, we found that coverage of ctDNA-panel testing for cancer indications increased from 2015 to 2019. Medicare policies that may be finalized in 2019 would provide a significant expansion of coverage to pan-cancer tests. Future research should continue to track coverage over time and evaluate the impact of coverage changes on test utilization. Given that genomic medicine is rapidly changing, payers and policymakers will need to continue to evolve policies to keep pace with emerging science and standards in clinical care.