PRIVILEGED AND CONFIDENTIAL — ATTORNEY WORK PRODUCT
CALDERWOOD & HARKNESS LLP Healthcare & Life Sciences Practice Group
MEMORANDUM
TO: Miriam Calloway, Partner, Healthcare & Life Sciences
FROM: Sean Morrissey, Senior Associate, Healthcare & Life Sciences
DATE: April 28, 2024
RE: Review of Larkfield Health Economics, LLC HEOR Analysis for Incorporation into OncoPrecise GXP-47 MolDx Technical Assessment Dossier
MATTER NO.: 1043-00002
CLIENT: Genome Dx Therapeutics Inc. ("GDT")
STATUS: DRAFT — For Partner Review
I. PURPOSE AND SCOPE OF REVIEW
This memorandum summarizes Calderwood & Harkness LLP's review of the health economics and outcomes research ("HEOR") analysis prepared by Larkfield Health Economics, LLC ("Larkfield") for potential incorporation into the MolDx Technical Assessment Dossier (Version 1.0) being prepared for submission to Ridgeline Administrative Services, LLC ("Ridgeline" or "RAS"), the contractor administering the MolDx program on behalf of the Centers for Medicare & Medicaid Services ("CMS"). The Larkfield deliverables were prepared under the direction of Dr. Nathan Ostroff, Principal Consultant, Larkfield Health Economics, LLC, 7910 Woodmont Avenue, Suite 610, Bethesda, MD 20814. The intended recipient of the MolDx dossier is the Ridgeline MolDx program office, headquartered at 1600 Hampton Street, Suite 700, Columbia, SC 29201, with the technical assessment to be directed by Dr. Catherine Embry, PhD, MolDx Technical Assessment Director.
Larkfield was retained directly by GDT — not by Calderwood & Harkness — to prepare the clinical utility and cost-effectiveness analyses that form a central component of the MolDx dossier. Calderwood & Harkness's role in connection with the Larkfield HEOR analysis is limited to reviewing the work product for the following purposes:
(1) Alignment with MolDx technical assessment requirements and evidentiary expectations;
(2) Regulatory compliance considerations under applicable CMS policies, including local coverage determination ("LCD") requirements and National Coverage Determination ("NCD") precedent;
(3) Consistency with the clinical evidence base as currently available, including the status of the ongoing CLARITY-LUNG clinical validation trial; and
(4) General suitability for incorporation into the MolDx Technical Assessment Dossier to be submitted in connection with GDT's LCD request.
The Larkfield HEOR analysis comprises two principal deliverables: (a) a Clinical Utility Evidence Report ("Clinical Utility Report") and (b) a Cost-Effectiveness Analysis ("CEA"), both relating to OncoPrecise GXP-47, a 47-gene RNA expression panel performed on formalin-fixed paraffin-embedded ("FFPE") non-small cell lung cancer ("NSCLC") tumor tissue. The test produces a composite Therapy Response Index ("TRI") score intended to predict response to immune checkpoint inhibitor ("ICI") therapy in patients with Stage IIIB/IV NSCLC.
The MolDx Technical Assessment Dossier is intended for filing with Ridgeline as part of a local coverage determination request to Cascade Fiscal Intermediaries, LLC ("CFI" or "Cascade"), the Medicare Administrative Contractor ("MAC") for Jurisdiction E (California). CFI's principal offices are located at 2800 SW Moody Avenue, Suite 500, Portland, OR 97201. The CFI Medical Director for molecular diagnostics coverage is Dr. William Hargrove, MD, FACP, and the assigned LCD Analyst is Sandra Phu.
It is important to note at the outset that the LCD filing and MolDx dossier submission have not yet occurred. The target filing date is currently August 2024, subject to reassessment following review of interim results from the CLARITY-LUNG clinical validation trial, which are discussed in detail in Section IV below. A separate Revised LCD/MolDx Strategy Memorandum dated April 22, 2024, prepared by this team, recommends deferral of the LCD filing until final CLARITY-LUNG data is available. The revised projection for final data readout is Q2 2025. The observations and recommendations set forth in this memorandum should be read in conjunction with and in the context of the filing deferral recommendation.
II. BACKGROUND ON ONCOPRECISE GXP-47 AND CLINICAL CONTEXT
A. Test Description
OncoPrecise GXP-47 is a laboratory-developed test ("LDT") performed exclusively at GDT's single CLIA-certified, CAP-accredited reference laboratory located at 4720 Torrey Pines Road, Suite 300, San Diego, CA 92130. The laboratory holds CLIA Certificate No. 05D2187432 and CAP Accreditation No. 9074821. The laboratory's National Provider Identifier ("NPI") is 1902847365. GDT does not currently operate any satellite or secondary laboratory facilities, and the test is not available through any third-party reference laboratory.
The OncoPrecise GXP-47 test is a 47-gene expression panel that utilizes RNA extracted from FFPE NSCLC tumor tissue specimens. The assay employs a proprietary algorithm to analyze the expression levels of the 47 target genes and generates a composite Therapy Response Index ("TRI") score. The TRI score is a continuous numeric value ranging from 0 to 100, which is further stratified into three clinically actionable categories for reporting purposes:
• High Likelihood of Response (TRI ≥ 70): Patients in this category are predicted to derive clinically meaningful benefit from first-line ICI therapy, with the TRI score suggesting that the tumor microenvironment is favorably disposed toward immune-mediated anti-tumor response.
• Intermediate (TRI 40–69): Patients in this category fall within an intermediate range where the predicted ICI response is uncertain, and additional clinical factors and biomarkers (including PD-L1 expression and tumor mutational burden) should be weighed by the treating oncologist in determining treatment selection.
• Low Likelihood of Response (TRI < 40): Patients in this category are predicted to have a low probability of deriving meaningful benefit from first-line ICI monotherapy or combination therapy, and the TRI score suggests that platinum-based cytotoxic chemotherapy may represent a more appropriate initial treatment approach.
The intended clinical use of OncoPrecise GXP-47 is to guide oncologist decision-making regarding whether to initiate first-line ICI therapy — including but not limited to pembrolizumab (Keytruda®, Quillon Group), nivolumab (Opdivo®, Halcyon Industries), and atezolizumab (Tecentriq®, Pemberton Partners LLC) — versus platinum-based chemotherapy in adult patients with Stage IIIB or Stage IV NSCLC. The test is intended to be used in conjunction with, not as a replacement for, existing biomarkers including PD-L1 expression (as assessed by immunohistochemistry) and, where clinically indicated, tumor mutational burden.
B. Analytical and Clinical Validation Status
The analytical validation of OncoPrecise GXP-47 was completed on December 14, 2022. The results of the analytical validation study were published in the Journal of Molecular Oncology Diagnostics, Volume 18, Issue 4, pages 412–429, December 2022. The publication describes the assay's performance characteristics including accuracy, precision (within-run and between-run reproducibility), analytical sensitivity, analytical specificity, and specimen stability parameters. The analytical validation data are generally considered to be complete and adequate for purposes of the MolDx dossier, as assessed in our prior review memoranda.
The clinical validation of OncoPrecise GXP-47 is currently being evaluated through the CLARITY-LUNG trial, registered on ClinicalTrials.gov under Identifier NCT05891247. CLARITY-LUNG is designed as a prospective, multi-site, blinded clinical validation study being conducted across 14 clinical sites in the United States. The trial's Principal Investigator is Dr. Priya Sundaram, who serves as GDT's Vice President of Clinical Development. The Data Safety Monitoring Board ("DSMB") for the trial is chaired by Dr. Evelyn Matsuda, MD, PhD, of the Carrington Cancer Institute in Houston, Texas.
The CLARITY-LUNG trial was designed with a target enrollment of 1,200 evaluable patients. As of the date of this review, approximately 847 patients have been enrolled, representing approximately 70.58% of the enrollment target. The trial's primary endpoints are the positive predictive value ("PPV") and negative predictive value ("NPV") of the TRI score for 12-month progression-free survival ("PFS") on ICI therapy, with PPV assessed in the High Likelihood of Response cohort (TRI ≥ 70) and NPV assessed in the Low Likelihood of Response cohort (TRI < 40).
A pre-planned interim analysis was conducted at the 50% enrollment milestone (n=600 evaluable patients). The results of this interim analysis were released on March 29, 2024, and disclosed the following:
• Positive Predictive Value (PPV): 0.68 (pre-specified performance threshold: PPV ≥ 0.75)
• Negative Predictive Value (NPV): 0.71 (pre-specified performance threshold: NPV ≥ 0.80)
Notably, neither the observed PPV nor the observed NPV met the trial's pre-specified performance thresholds at the interim analysis. The DSMB convened on April 5, 2024, to review the interim results and issued a recommendation to continue enrollment to full accrual. However, the DSMB characterized the interim results as "inconclusive and below pre-specified performance benchmarks." The DSMB did not recommend any protocol modifications at this time, but did note that the final analysis results will need to demonstrate substantial improvement over the interim values to meet the trial's success criteria.
The total budget for the CLARITY-LUNG trial is $28,500,000. Expenditures through May 2024 are approximately $19,100,000, representing 67.02% of the total trial budget. The final data readout, originally projected for Q4 2024, has been revised to Q2 2025 based on current enrollment velocity and the required follow-up period for the remaining enrolled patients.
C. Existing GDT Product Context
For context, GDT currently markets and sells one commercially available diagnostic product, OncoPrecise GXP-12, a 12-gene expression panel used in early-stage breast cancer prognosis. OncoPrecise GXP-12 is reimbursed by Medicare under existing LCD L38472, which was issued by CFI with an effective date of August 15, 2021. GDT reported fiscal year 2023 revenue of $6,200,000 from the GXP-12 product line. The existing reimbursement history and coverage relationship with CFI provides GDT with a degree of institutional familiarity with the MAC, but the GXP-47 product presents a substantially different clinical context and evidentiary profile that will require an independent coverage determination.
III. SUMMARY OF LARKFIELD HEOR DELIVERABLES
A. Clinical Utility Evidence Report
Larkfield's Clinical Utility Evidence Report presents a systematic review and synthesis of the evidence supporting the clinical utility of genomic-based ICI response prediction in NSCLC. The report is structured to address the clinical utility framework that MolDx reviewers apply when assessing molecular diagnostic tests for coverage, and draws upon both the published peer-reviewed literature and proprietary data generated by GDT.
The literature synthesis draws primarily from GDT's master bibliography of 43 peer-reviewed publications, which was compiled by this team and provided to Larkfield at the commencement of their engagement. The bibliography encompasses publications spanning analytical validation, clinical validation, clinical utility, health economics, and background literature on ICI therapy in advanced NSCLC. Larkfield supplemented this bibliography with additional publications identified through its own literature search, though as discussed below in Section IV.F, the Larkfield analysis references only 28 of the 43 publications in the master bibliography.
The Clinical Utility Evidence Report addresses three principal domains of clinical utility, consistent with the framework commonly applied by MolDx and other technology assessment bodies:
(1) Impact on Clinical Decision-Making. The report presents evidence and arguments supporting the proposition that OncoPrecise GXP-47 test results would change treatment selection in a measurable proportion of cases. Specifically, the report posits that oncologists who receive TRI score information alongside conventional biomarker results (PD-L1 expression, tumor mutational burden) would make different first-line treatment decisions than they would in the absence of TRI score data. The report cites survey-based and chart-review evidence from analogous testing contexts suggesting that genomic test results alter physician treatment decisions in approximately 25–40% of cases, though direct evidence for GXP-47 is limited.
(2) Impact on Patient Health Outcomes. The report presents evidence supporting the hypothesis that patients stratified by TRI score would experience differential PFS and overall survival ("OS") outcomes, with High Likelihood of Response patients (TRI ≥ 70) deriving greater PFS benefit from ICI therapy relative to Low Likelihood of Response patients (TRI < 40). The report relies primarily on retrospective analyses and the CLARITY-LUNG interim data for this domain, supplemented by mechanistic arguments drawn from the published literature on gene expression correlates of immune response in NSCLC.
(3) Impact on Healthcare Resource Utilization. The report argues that use of OncoPrecise GXP-47 would result in avoidance of ineffective ICI therapy in patients with low TRI scores, thereby reducing costs associated with ICI drug acquisition, infusion services, immune-related adverse event management, and downstream care for patients who progress on ICI therapy. The report quantifies potential cost savings using a modeling approach that is linked to but distinct from the formal cost-effectiveness analysis described below.
In addition to the three primary clinical utility domains, the Clinical Utility Evidence Report includes a narrative review of analogous genomic tests in other oncology settings that have received favorable MolDx technical assessments and LCD coverage. These analogues include multi-gene expression panels used in breast cancer (including OncoPrecise GXP-12 and competing assays), colorectal cancer, and prostate cancer. The report uses these analogues as precedent for the proposition that gene expression-based treatment selection panels can meet MolDx clinical utility standards.
B. Cost-Effectiveness Analysis
Larkfield prepared a formal decision-analytic cost-effectiveness model comparing two treatment strategies for patients with newly diagnosed Stage IIIB/IV NSCLC:
(1) TRI-Guided Strategy. Under this strategy, patients undergo OncoPrecise GXP-47 testing, and the TRI score is used to guide treatment assignment. Patients with TRI ≥ 70 (High Likelihood of Response) are assigned to first-line ICI therapy; patients with TRI < 40 (Low Likelihood of Response) are assigned to first-line platinum-based chemotherapy; and patients with TRI 40–69 (Intermediate) receive ICI therapy or chemotherapy based on PD-L1 expression and clinical judgment, consistent with current practice patterns.
(2) Standard-of-Care Strategy. Under this strategy, treatment assignment is based on PD-L1 expression testing (by immunohistochemistry) combined with clinical judgment, without the benefit of TRI score information. This reflects current standard-of-care practice for first-line treatment selection in advanced NSCLC.
The cost-effectiveness analysis employs a Markov cohort simulation model with three mutually exclusive health states: (i) progression-free disease, (ii) progressed disease, and (iii) death. Patients enter the model in the progression-free state and transition between states according to transition probabilities that vary by treatment arm and TRI score category.
The key model parameters are as follows:
• Model Time Horizon: 5 years (with a lifetime horizon modeled as a sensitivity analysis).
• Analytic Perspective: U.S. healthcare payer (Medicare).
• Primary Outcome Measure: Incremental cost-effectiveness ratio ("ICER") expressed as cost per quality-adjusted life year ("QALY") gained.
• Cycle Length: One month (with half-cycle correction applied).
• discount Rate: 3% per annum for both costs and health outcomes, consistent with standard CMS and Agency for Healthcare Research and Quality ("AHRQ") guidance.
• Test Cost Input: $3,850 per test, corresponding to GDT's target Medicare reimbursement rate.
• ICI Therapy Costs: Based on published wholesale acquisition cost ("WAC") data for pembrolizumab, nivolumab, and atezolizumab, weighted by current market share data.
• Chemotherapy Costs: Based on published cost data for standard platinum-based doublet chemotherapy regimens (carboplatin/pemetrexed, cisplatin/pemetrexed) in NSCLC.
• Adverse Event Management Costs: Based on published estimates of the costs of managing immune-related adverse events (irAEs) associated with ICI therapy and the costs of managing chemotherapy-related toxicities.
• Quality-of-Life Utilities: Health state utilities drawn from published literature on quality-of-life measurement in advanced NSCLC.
Larkfield's base case analysis yields an ICER of approximately $38,500 per QALY gained for the TRI-guided strategy relative to the standard-of-care strategy. Larkfield characterizes this result as "highly cost-effective" relative to commonly cited willingness-to-pay ("WTP") thresholds in the health economics literature, which generally range from $50,000 to $150,000 per QALY gained in the United States.
The analysis includes several forms of sensitivity analysis:
• One-Way (Deterministic) Sensitivity Analysis: Key model parameters are individually varied across plausible ranges to assess the impact on the ICER.
• Probabilistic Sensitivity Analysis (PSA): A Monte Carlo simulation (1,000 iterations) is used to characterize joint parameter uncertainty and generate a cost-effectiveness acceptability curve.
• Scenario Analyses: Alternative scenarios are modeled, including variations in test performance characteristics (PPV and NPV), test price, ICI drug costs, disease progression rates, and model time horizon.
IV. CRITICAL OBSERVATIONS AND AREAS REQUIRING REVISION
Our review of the Larkfield HEOR deliverables has identified several significant issues that must be addressed before the analysis can be incorporated into the MolDx Technical Assessment Dossier. These issues are discussed in detail below, together with our specific recommendations for revision.
A. Reliance on CLARITY-LUNG Interim Data — Evidentiary Sufficiency Concern
Issue. The Larkfield HEOR analysis relies substantially on the CLARITY-LUNG interim analysis results as key model inputs for the cost-effectiveness analysis. Specifically, the base case Markov model utilizes the interim PPV of 0.68 and interim NPV of 0.71 as the primary test performance parameters driving the model's simulation of treatment assignment accuracy. As discussed in Section II.B above, these interim values fall below the pre-specified performance thresholds established in the CLARITY-LUNG trial protocol: PPV ≥ 0.75 and NPV ≥ 0.80.
The reliance on these interim values presents a fundamental evidentiary challenge for the HEOR analysis and, by extension, for the MolDx dossier as a whole. The interim data represent a preliminary assessment based on 50% of the target enrollment (n=600 of 1,200), and the DSMB has characterized the results as "inconclusive and below pre-specified performance benchmarks." The use of these values as base case inputs in a cost-effectiveness model — without adequate contextualization of their preliminary nature — creates a risk that the model's conclusions will be viewed by MolDx reviewers as insufficiently supported.
Moreover, the ICER result of approximately $38,500 per QALY gained is sensitive to the assumed test performance characteristics. If the final PPV and NPV are materially lower than the interim values, the cost-effectiveness profile of OncoPrecise GXP-47 could deteriorate significantly, potentially exceeding commonly applied WTP thresholds and undermining the value proposition presented to MolDx.
Recommendation. Larkfield should clearly characterize the CLARITY-LUNG interim data as preliminary and should model sensitivity scenarios using both the observed interim values (PPV = 0.68, NPV = 0.71) and the pre-specified target values (PPV ≥ 0.75, NPV ≥ 0.80). The current draft presents the interim values as established performance characteristics rather than as interim estimates subject to revision upon completion of the trial; this framing must be corrected.
The analysis should include explicit and prominent discussion of the uncertainty surrounding the final clinical validation results and the potential impact on cost-effectiveness conclusions if final PPV and NPV differ materially from the interim values — in either direction. In particular, the analysis should model a downside scenario in which final values fall below the interim estimates (e.g., PPV = 0.60, NPV = 0.65) to demonstrate the robustness (or fragility) of the cost-effectiveness conclusions under less favorable assumptions.
The dossier should also transparently disclose that the final CLARITY-LUNG data readout has been revised from Q4 2024 to Q2 2025, and should note that the clinical validation evidence will be updated in a subsequent dossier revision upon availability of the final data.
Risk Assessment. Based on our experience with MolDx technical assessments and LCD processes, we assess a significant risk that Ridgeline's technical assessment team (Dr. Catherine Embry's group) and CFI (Dr. Hargrove's office) will closely scrutinize the clinical validation evidence base. MolDx reviewers have historically demonstrated a preference for completed clinical validation studies, and have expressed skepticism toward coverage submissions that rely primarily on interim or preliminary data. Filing a MolDx dossier with an HEOR analysis that is anchored to interim data failing to meet pre-specified performance thresholds creates a substantial risk of an unfavorable technical assessment, which could result in a non-coverage recommendation or a request for resubmission with additional evidence. This concern is consistent with the analysis and recommendation set forth in our Revised LCD/MolDx Strategy Memorandum dated April 22, 2024, which recommends deferral of the LCD filing until final CLARITY-LUNG data is available.
B. Test Cost Assumption and Crosswalk Alignment
Issue. The Larkfield cost-effectiveness model uses $3,850 as the test cost input in the base case analysis. This figure corresponds to GDT's target Medicare reimbursement rate for OncoPrecise GXP-47. However, this reimbursement rate has not yet been established by CMS through either the crosswalk or gapfill pricing methodology, and there is no assurance that the eventual rate will match GDT's target.
Our Reimbursement Rate Analysis Memorandum, delivered on May 14, 2024, together with the supporting pricing analysis workbook, presents a detailed crosswalk analysis based on three comparator tests. The weighted average of the crosswalk comparators is $3,622.33, calculated as follows:
GDT's target rate of $3,850 represents a 6.28% premium over the weighted average of the crosswalk comparators. While it is not uncommon for manufacturers to seek reimbursement rates above the comparator average — particularly where the test under review offers features or performance characteristics that differentiate it from comparators — the use of a target rate that exceeds the weighted average as the sole base case assumption in a cost-effectiveness analysis may raise credibility concerns among MolDx reviewers who are familiar with the crosswalk methodology and the comparator landscape.
Recommendation. Larkfield should run sensitivity analyses at the weighted average comparator rate ($3,622.33), the lowest comparator rate ($2,919 under CPT 81455), and the highest comparator rate ($4,206 under PLA 0239U), in addition to the GDT target rate of $3,850. This range of sensitivity analyses will demonstrate the robustness of the cost-effectiveness conclusions across the plausible range of reimbursement rates.
For the base case analysis, Larkfield should either use the weighted average comparator rate ($3,622.33) as the primary input — which would represent a conservative and defensible approach — or, if the base case retains the $3,850 target rate, the analysis should include a clear methodological note explaining that this rate is GDT's target rate for purposes of the crosswalk pricing application and that the rate is subject to final determination by CMS through either the crosswalk or gapfill pricing process. The note should briefly describe both pricing methodologies and acknowledge that the final reimbursement rate may differ from the assumed value.
In addition, the analysis should address the gapfill alternative. Under the gapfill process, MACs determine a test's payment rate based on range factors including resources required, charges for the test, and applicable fee schedule amounts for comparable tests. If CMS selects gapfill rather than crosswalk for GXP-47 pricing, the eventual rate could diverge more significantly from the crosswalk comparator average. The sensitivity analysis should therefore include a broader range of test cost assumptions to account for this possibility.
Rationale. MolDx reviewers at Ridgeline, and particularly the LCD analyst at CFI (Sandra Phu), will be aware that the reimbursement rate for OncoPrecise GXP-47 has not yet been established. Presenting a cost-effectiveness analysis that assumes a rate not yet determined — and one that is at a premium to the weighted average of comparators — may undermine the credibility of the economic analysis and invite questions about whether the cost-effectiveness conclusions are artificially favorable.
C. Clinical Utility Evidence — Insufficient Direct Evidence for OncoPrecise GXP-47
Issue. The Clinical Utility Evidence Report relies heavily on analogical evidence from genomic tests used in other oncology indications, including multi-gene expression panels in breast cancer and colorectal cancer. While such cross-indication analogies are permissible as supporting evidence within MolDx technical assessments, and indeed have been cited favorably in prior LCD determinations for novel molecular diagnostics, the MolDx program's technical assessment framework ultimately requires demonstration that the specific test under review — in this case, OncoPrecise GXP-47 — provides actionable clinical information that changes patient management and improves health outcomes.
Our review of the Clinical Utility Evidence Report reveals that the direct evidence supporting GXP-47-specific clinical utility is thin. The report identifies no completed prospective studies demonstrating that use of the TRI score changes oncologist treatment decisions or improves patient outcomes. The CLARITY-LUNG trial, once completed, may provide such evidence depending on its design and outcome, but the interim data currently available do not directly address the clinical utility question — they address clinical validity (i.e., the association between TRI score categories and patient outcomes) but not the distinct question of whether use of the test in clinical practice actually changes management.
This evidentiary gap is significant because MolDx reviewers have consistently distinguished between clinical validity and clinical utility, and have issued non-coverage determinations for tests with adequate clinical validity evidence but insufficient clinical utility evidence.
Recommendation. Larkfield should strengthen the direct evidence section by incorporating the following:
• Data from the CLARITY-LUNG trial protocol design demonstrating that TRI score results were intended to inform specific, pre-defined treatment decisions. While the trial is blinded (and therefore the TRI score results are not being used to direct treatment during the trial), the protocol's treatment algorithm provides evidence of how TRI score categories are expected to map to clinical management decisions in real-world practice.
• Any available retrospective sub-analyses or case series from GDT's laboratory experience with the GXP-47 assay, including data from pre-commercial or analytical validation specimens where treatment outcomes are known and can be correlated with TRI score retrospectively. We understand from discussions with Kevin Driscoll that GDT may have access to a retrospective dataset of approximately 200–300 NSCLC cases with known treatment and outcome data, though we have not independently reviewed this dataset.
• A clear and prominent distinction between clinical validity (the demonstrated association between TRI score categories and differential patient outcomes) and clinical utility (evidence that use of the test in clinical practice changes management decisions and leads to improved outcomes). This distinction should be framed consistently with the ACCE framework (Analytic validity, Clinical validity, Clinical utility, and associated Ethical, legal, and social implications), which is referenced by MolDx in its technical assessment methodology.
• A revised framing of the analogical evidence from other oncology indications, clearly characterizing these analogues as supporting precedent that establishes the plausibility and potential of gene expression-based treatment selection panels, rather than as primary evidence for GXP-47-specific clinical utility.
Cross-Reference. Our Clinical Evidence Gap Analysis Memorandum dated April 15, 2024, provides a detailed assessment of the gap between the currently available clinical utility evidence for OncoPrecise GXP-47 and the evidentiary standard that MolDx and CFI have historically required for LCD coverage of molecular diagnostic tests. Larkfield should review that memorandum and align the Clinical Utility Evidence Report's analysis and conclusions with the gap assessment and recommended evidence development strategy set forth therein.
D. Model Structure and Transparency Concerns
Issue. The Markov model employed in the cost-effectiveness analysis, while generally appropriate in its structural design for modeling treatment outcomes in advanced NSCLC, lacks certain transparency elements that Ridgeline reviewers will expect and that are consistent with best practices in health economic modeling as reflected in the Consolidated Health Economic Evaluation Reporting Standards ("CHEERS") checklist and the ISPOR (International Society for Pharmacoeconomics and Outcomes Research) modeling guidelines.
We have identified the following specific transparency deficiencies:
• Insufficiently Sourced Transition Probabilities. The transition probabilities between health states (progression-free disease → progressed disease, progression-free disease → death, and progressed disease → death) are not fully sourced to peer-reviewed publications in the current draft of the cost-effectiveness analysis. Several transition probabilities — particularly those for the TRI-guided treatment arm — are cited to "internal GDT data" without further specification of the dataset, methodology, or sample size underlying these estimates. This lack of sourcing will be problematic in a MolDx review, where transparency and reproducibility of model inputs are expected.
• Inadequate Adverse Event Modeling. The model does not appear to incorporate differential adverse event profiles associated with ICI therapy versus platinum-based chemotherapy in a manner consistent with recent published meta-analyses. Immune-related adverse events ("irAEs") associated with ICI therapy — including pneumonitis, colitis, hepatitis, endocrinopathies, and dermatologic toxicities — have been extensively characterized in published literature and carry distinct clinical and economic consequences relative to the hematologic and non-hematologic toxicities associated with cytotoxic chemotherapy. The current model appears to use a simplified adverse event cost estimate that does not capture these differential profiles with adequate granularity.
• Undocumented discount Rate Assumptions. The discount rate assumptions — 3% per annum for both costs and health outcomes — are appropriately selected and consistent with standard CMS and AHRQ guidance. However, these assumptions are not discussed or justified in the model documentation. MolDx reviewers will expect explicit documentation of discount rate selection and its consistency with applicable guidelines.
Recommendation. All model inputs should be transparently sourced, with primary reliance on published peer-reviewed evidence. Where proprietary GDT data is used as a model input, the data should be clearly identified as such, accompanied by a description of the underlying methodology (including study design, sample size, patient population, and data collection methods), and presented as supplemental to — rather than a substitute for — published evidence. Where published data is available for a given parameter, it should be used as the primary input, with GDT proprietary data presented in a sensitivity analysis.
The adverse event modeling should be updated to reflect the most current published safety data for ICI agents used in advanced NSCLC, including data from landmark randomized controlled trials (KEYNOTE-024, Tidewater Data Systems-042, CheckMate-227, and IMpower110) and published meta-analyses of ICI safety profiles. The model should incorporate both the incidence of grade 3–4 adverse events and the associated costs of management, differentiated between ICI-related irAEs and chemotherapy-related toxicities.
The model documentation should include a dedicated section addressing methodological choices, including discount rate selection, cycle length, half-cycle correction, and extrapolation methods (where applicable for the lifetime horizon sensitivity analysis), with references to the applicable guidelines and standards.
E. Revenue Projection Assumptions (for Internal Use Only — Not for Dossier)
Observation. During our review of the Larkfield HEOR deliverables, we noted that certain revenue projection assumptions and market sizing data are embedded within the cost-effectiveness analysis in a manner that suggests they may have been included for GDT's internal commercial planning purposes rather than for the MolDx dossier submission. Upon further investigation, we have confirmed that these revenue projections mirror — in some cases identically — the financial projections developed by Verdant Life Sciences Capital, LLC, GDT's investment bank and financial advisor. The Verdant engagement is led by Christine Lau, Managing Director.
Specifically, the following data points appear in the Larkfield cost-effectiveness analysis and correspond to the Verdant financial projections:
Recommendation. These revenue and volume projections should not be included in the MolDx Technical Assessment Dossier under any circumstances. The HEOR analysis prepared for submission to Ridgeline should address clinical cost-effectiveness exclusively from a payer and/or societal perspective, consistent with the purpose and conventions of health economic evaluations submitted in support of coverage determinations. Market-sizing data, revenue projections, volume forecasts, and commercial uptake assumptions are not relevant to the clinical cost-effectiveness question and, if included, could undermine the perceived scientific objectivity of the clinical utility and cost-effectiveness arguments.
Any revenue and market projection data should be removed from the dossier-facing deliverable and maintained separately for GDT's internal commercial planning and investor relations purposes. If Larkfield has prepared a separate market access deliverable for GDT that incorporates these projections, that deliverable should be clearly distinguished from and kept segregated from the MolDx dossier-facing HEOR analysis.
The presence of financial projection data within a clinical cost-effectiveness analysis also raises a broader concern about the objectivity and independence of the Larkfield analysis. We recommend that Larkfield include a conflict-of-interest disclosure statement in the final version of the HEOR deliverables, noting its engagement by GDT and affirming that the clinical and economic conclusions are based on independent scientific analysis.
F. Bibliography and Literature Coverage
Observation. The Larkfield HEOR analysis references 28 peer-reviewed publications. Our MolDx Dossier Bibliography — compiled by Sean Morrissey and Christopher Aiken, an associate on this team — contains 43 peer-reviewed publications organized by category: analytical validation, clinical validation, clinical utility, health economics, and background literature on ICI therapy in NSCLC. The bibliography includes GDT's analytical validation publication (Journal of Molecular Oncology Diagnostics, Vol. 18, Issue 4, pp. 412–429, December 2022).
A comparison of the Larkfield reference list against our master bibliography reveals that 15 publications in our bibliography are not cited in the Larkfield analysis. While not all 43 publications in the master bibliography would necessarily be cited in the HEOR deliverables — some relate to background or contextual topics that may not be directly relevant to the clinical utility or cost-effectiveness analyses — several of the omitted publications appear to be directly relevant to the HEOR analysis, particularly in the clinical utility and health economics categories. The omitted publications include recent systematic reviews and meta-analyses of ICI efficacy in NSCLC, published cost-effectiveness analyses of companion diagnostics and molecular-guided treatment selection, and clinical utility framework publications that provide methodological support for the analytical approach.
Recommendation. Larkfield should cross-reference our 43-publication master bibliography and incorporate additional relevant publications, particularly those in the clinical utility and health economics categories that are directly relevant to the analyses presented. The HEOR analysis should be consistent with and supported by the complete evidence base assembled for the dossier, as any inconsistency between the HEOR analysis references and the dossier's broader bibliography may invite questions from MolDx reviewers about the comprehensiveness of the literature review.
Conversely, any publications cited by Larkfield that are not already included in our master bibliography should be flagged and transmitted to this team for review and potential inclusion in the master bibliography. Consistency between the Larkfield reference list and the dossier bibliography is important for presenting a coherent and comprehensive evidence package.
V. REGULATORY COMPLIANCE CONSIDERATIONS
A. CMS/MolDx Evidentiary Standards
The MolDx program, administered by Ridgeline on behalf of participating MACs, requires that clinical utility and cost-effectiveness claims for molecular diagnostic tests be supported by evidence meeting specific evidentiary standards. These standards are articulated in the MolDx program's published technical assessment framework and in guidance documents issued by Ridgeline. Key elements of the evidentiary framework include:
• A preference for prospective clinical studies over retrospective analyses as the primary evidence base for clinical validity and clinical utility claims.
• A requirement that clinical utility be demonstrated through evidence showing that test results change clinical management decisions and that such changes lead to improved patient outcomes, as distinct from evidence showing only that the test has prognostic or predictive value (clinical validity).
• An expectation that cost-effectiveness analyses be based on transparent, reproducible models using published data inputs and accepted health economic methodology.
• An evidence hierarchy that places randomized controlled trial evidence at the highest tier, followed by prospective cohort studies, retrospective cohort studies, case-control studies, case series, and expert opinion/consensus.
The Larkfield analysis should explicitly map its evidence to the MolDx evidence hierarchy, identifying where the current evidence for OncoPrecise GXP-47 falls within that hierarchy for each domain of the analysis (analytical validity, clinical validity, clinical utility, and health economics). This mapping exercise will enable MolDx reviewers to efficiently assess the evidentiary basis of the claims and will demonstrate GDT's awareness of and adherence to the program's requirements.
Particular attention should be given to ensuring that clinical utility claims are clearly supported by evidence that the test results change clinical management in a measurable way — not merely that the test has prognostic or predictive value. This distinction is a recurring theme in MolDx technical assessments and has been the basis for several non-coverage determinations involving molecular diagnostic tests that demonstrated adequate clinical validity but insufficient clinical utility evidence.
B. Medicare Benefit Category and Reasonable and Necessary Standard
The HEOR analysis should directly address the statutory standard for Medicare coverage of diagnostic tests. Under the Social Security Act § 1862(a)(1)(A), Medicare may only cover items and services that are "reasonable and necessary for the diagnosis or treatment of illness or injury or to improve the functioning of a malformed body member." This "reasonable and necessary" standard is the foundational legal requirement for Medicare coverage and is the standard that CFI will apply in its LCD determination.
The cost-effectiveness analysis prepared by Larkfield should be framed within the context of this "reasonable and necessary" determination. The analysis should demonstrate not only that OncoPrecise GXP-47 is cost-effective in the pharmacoeconomic sense (i.e., that it generates health benefits at an acceptable cost relative to alternatives), but also that the test is reasonable and necessary for the diagnosis or treatment of NSCLC. This requires demonstrating that the test provides clinically meaningful, actionable information that is not otherwise available through existing diagnostic modalities (including PD-L1 immunohistochemistry and tumor mutational burden testing), and that the use of the test leads to treatment decisions that improve patient outcomes relative to decisions made without the test.
The cost-effectiveness analysis should not be presented as a standalone pharmacoeconomic evaluation divorced from the coverage context. Rather, it should be integrated with the clinical utility evidence to present a comprehensive case for coverage under the "reasonable and necessary" standard.
C. Compliance with Anti-Kickback Statute and Stark Law Considerations
While compliance with the federal Anti-Kickback Statute ("AKS") and the Physician Self-Referral Law ("Stark Law") is not a primary focus of the HEOR analysis, we flag the following considerations for GDT's and Larkfield's awareness:
Any arrangements between GDT and clinical sites participating in the CLARITY-LUNG trial — including site agreements, investigator payments, and specimen procurement arrangements — must be structured to comply with the federal AKS, 42 U.S.C. § 1320a-7b(b), and the Stark Law, 42 U.S.C. § 1395nn. Similarly, any arrangements between GDT and referring physicians or institutions related to the commercial utilization of OncoPrecise GXP-47 must comply with these statutes.
From the perspective of the HEOR analysis and MolDx dossier, the relevance of these compliance considerations is twofold. First, the HEOR analysis should not reference or rely upon any data generated under arrangements that could raise AKS or Stark Law compliance concerns, as the integrity of the underlying data is a prerequisite for the credibility of the analysis. Second, MolDx and MAC reviewers may inquire about the structure of clinical trial arrangements as part of their assessment of the independence and reliability of the clinical evidence, and GDT should be prepared to demonstrate that all such arrangements are compliant.
We note that a comprehensive AKS/Stark Law compliance review of the CLARITY-LUNG trial site agreements is being conducted separately by our colleague Patricia Engström in the firm's Healthcare Regulatory Practice Group and is not addressed further in this memorandum.
D. FDA Regulatory Status
OncoPrecise GXP-47 is currently marketed as a laboratory-developed test under the authority of the Clinical Laboratory Improvement Amendments of 1988 ("CLIA"), 42 U.S.C. § 263a, and implementing regulations at 42 C.F.R. Part 493. The test has not received clearance or approval from the U.S. Food and Drug Administration ("FDA") under any pathway, including 510(k), de novo classification, or premarket approval ("PMA").
The HEOR analysis should note this regulatory status clearly and should not make any claims or representations that could be construed as implying FDA clearance, approval, or authorization. This is particularly important given the evolving regulatory landscape for laboratory-developed tests. The FDA's proposed rule regarding LDT oversight, published in the Federal Register on October 3, 2023, could affect the regulatory status of OncoPrecise GXP-47 if finalized. While the proposed rule does not currently require FDA authorization for LDTs already on the market, its potential impact on future marketing and coverage of LDTs — including the perception of LDTs by MolDx and MAC reviewers — should be monitored.
For purposes of the MolDx dossier, GDT should be prepared to address questions from Ridgeline or CFI regarding the FDA regulatory status of the test and GDT's plans (if any) with respect to FDA submission. The HEOR analysis should characterize the test's regulatory status factually and neutrally, without editorial commentary on the appropriateness of LDT regulation.
VI. MARK-UP RECOMMENDATIONS — SPECIFIC SECTIONS
The following section-specific mark-up recommendations are provided for transmission to Larkfield through GDT (Kevin Driscoll). These recommendations supplement the broader observations in Section IV above with targeted, actionable editorial guidance.
A. Clinical Utility Section
We recommend the following revisions to the Clinical Utility Evidence Report:
1. Add a dedicated subsection on clinical actionability of TRI score categories. This subsection should specify how each TRI score category (High, Intermediate, Low) corresponds to a distinct clinical management decision. The subsection should describe the clinical decision pathway that would be followed for a patient in each TRI category, including the specific treatment options that would be recommended or avoided, the timing of treatment initiation, and the role of the TRI score in the context of other clinical and biomarker information available to the treating oncologist.
2. Revise the discussion of analogous tests. The current framing of analogical evidence from breast cancer, colorectal cancer, and other oncology indications should be revised to clearly characterize these as supporting precedent for the general proposition that gene expression-based treatment selection panels can demonstrate clinical utility, rather than as primary evidence for the clinical utility of OncoPrecise GXP-47 specifically. The revised framing should explicitly acknowledge the limitations of cross-indication analogies and should note the differences in disease biology, treatment landscape, and biomarker context between the analogue indications and advanced NSCLC.
3. Add a limitations subsection. A new subsection should be added to the Clinical Utility Evidence Report explicitly acknowledging the current status of the CLARITY-LUNG trial, the interim nature of the available clinical validation data, and the limitations that this status imposes on the strength of the clinical utility evidence. The subsection should note that the interim results have not met pre-specified performance benchmarks and that the final results, expected in Q2 2025, may materially affect the clinical utility conclusions.
4. Incorporate the CLARITY-LUNG trial design as evidence of clinical utility intent. The report should reference the CLARITY-LUNG trial protocol (NCT05891247) as evidence of GDT's commitment to generating robust prospective clinical utility data and should describe the trial's design features — including the pre-specified treatment decision algorithm, the multi-site design, and the blinded assessment — as structural elements supporting the eventual demonstration of clinical utility upon trial completion.
B. Cost-Effectiveness Section
We recommend the following revisions to the Cost-Effectiveness Analysis:
1. Revise test cost assumptions. Replace the single-point test cost assumption of $3,850 with a range reflecting crosswalk comparator variability ($2,919 to $4,206) in the sensitivity analysis, as detailed in Section IV.B above. If the base case retains the $3,850 target rate, include a methodological note explaining the basis for this assumption and its provisional nature.
2. Add a tornado diagram. Include a tornado diagram (one-way sensitivity analysis summary plot) showing the key drivers of ICER variability. This is a standard feature of published cost-effectiveness analyses and is expected by health economics reviewers. The tornado diagram should include, at minimum, the following parameters: test PPV, test NPV, test cost, ICI drug cost, chemotherapy cost, adverse event incidence rates, quality-of-life utilities, and discount rate.
3. Revise the base case test performance inputs. The base case should use the interim PPV (0.68) and NPV (0.71) as the primary inputs, clearly identified as interim values from an ongoing clinical validation trial. The pre-specified target values (PPV ≥ 0.75, NPV ≥ 0.80) should be modeled as a scenario analysis representing the "target performance" scenario, rather than being presented as the base case — which could be perceived by MolDx reviewers as presenting aspirational rather than demonstrated performance.
4. Source all model inputs. Ensure that all Markov model inputs — including transition probabilities, adverse event rates, costs, and utilities — are sourced to peer-reviewed publications or, where proprietary data is used, are clearly identified as such with accompanying methodology descriptions. Unsourced "internal GDT data" citations should be replaced with specific references to the underlying datasets and analytical methods.
5. Remove revenue and commercial data. Remove all market-size, revenue projection, commercial uptake, and volume forecast data from the cost-effectiveness analysis section, as discussed in Section IV.E above. These data are not relevant to the clinical cost-effectiveness question and should be maintained separately for GDT's internal use.
6. Add a budget impact analysis. Add a dedicated budget impact analysis ("BIA") from the Medicare payer perspective. MolDx reviewers frequently consider the budgetary implications of new molecular diagnostic tests alongside cost-effectiveness, and the inclusion of a BIA strengthens the economic evidence package. The BIA should estimate the incremental cost to Medicare of covering OncoPrecise GXP-47 over a defined time horizon (e.g., 1-year and 5-year), taking into account projected test utilization, the cost of the test, and any offsetting cost savings from avoided ineffective ICI therapy and reduced adverse event management. The BIA should use conservative volume assumptions distinct from the Verdant revenue projections.
VII. COORDINATION WITH OTHER WORKSTREAMS
The Larkfield HEOR analysis must be consistent with and fully integrated into the broader MolDx Technical Assessment Dossier (Version 1.0), which Calderwood & Harkness has been retained to prepare. The dossier framework was delivered to GDT on April 30, 2024, and the HEOR sections — specifically Sections 4 (Clinical Utility) and 5 (Health Economics) of the dossier — will incorporate Larkfield's deliverables following the completion of the revisions recommended herein.
Kevin Driscoll, GDT's Vice President of Market Access & Reimbursement, is the designated day-to-day point of contact coordinating among Larkfield, Calderwood & Harkness, and Carraway Regulatory Advisors, Inc. ("Carraway") on coding, coverage, and dossier alignment matters. Carraway is led by Janet Carraway, Principal, and is located at 233 South Wacker Drive, Suite 4100, Chicago, IL 60606. Carraway's role in the engagement includes advising on coding strategy (CPT/PLA code assignment), assisting with the coverage determination filing, and reviewing the dossier for regulatory and procedural compliance.
To ensure consistency across the various workstreams contributing to the MolDx dossier, we recommend the following coordination steps:
• Any revised Larkfield deliverables should be circulated simultaneously to both Calderwood & Harkness and Carraway for consistency and compliance review prior to incorporation into the final dossier.
• Larkfield should be provided with copies of the following Calderwood & Harkness work product for reference during its revision process: (a) the Clinical Evidence Gap Analysis Memorandum (April 15, 2024); (b) the Revised LCD/MolDx Strategy Memorandum (April 22, 2024); (c) the MolDx Dossier Bibliography (43 publications); and (d) the Reimbursement Rate Analysis Memorandum (May 14, 2024). Transmission of these materials should be coordinated through Kevin Driscoll and should be accompanied by appropriate confidentiality protections.
• A coordination call among all three advisory firms (Calderwood & Harkness, Larkfield, and Carraway) should be scheduled to align on the scope and timeline for Larkfield's revisions, the integration plan for the dossier, and the implications of the filing deferral recommendation.
Given the GDT board of directors' April 9, 2024 request that Calderwood & Harkness reassess the LCD filing strategy in light of the CLARITY-LUNG interim results, and our subsequent April 22, 2024 recommendation to defer the LCD filing until final CLARITY-LUNG data is available (projected Q2 2025), the urgency of finalizing the Larkfield revisions is reduced relative to the original August 2024 filing timeline. However, we recommend that Larkfield complete the revisions identified herein in a timely manner so that the HEOR analysis is substantially ready for incorporation into a revised dossier when the final clinical validation data becomes available. Completing the revisions now — while the analytical issues are fresh and the relevant personnel are available — will enable a more efficient and compressed timeline for finalizing the dossier once the CLARITY-LUNG trial results are in hand.
VIII. SUMMARY OF RECOMMENDED REVISIONS
The following is a consolidated list of all specific revisions recommended for the Larkfield HEOR analysis, as discussed in detail in Sections IV, V, and VI above:
1. Characterize CLARITY-LUNG interim data as preliminary. The analysis should clearly identify the interim PPV (0.68) and NPV (0.71) as preliminary estimates from an ongoing clinical validation trial. Model sensitivity scenarios should be run using both the observed interim values and the pre-specified target values (PPV ≥ 0.75, NPV ≥ 0.80), as well as a downside scenario with values below interim estimates.
2. Add sensitivity analyses for test cost. The cost-effectiveness model should be run at the weighted average crosswalk comparator rate ($3,622.33), the lowest comparator rate ($2,919), and the highest comparator rate ($4,206), in addition to the GDT target rate of $3,850. The base case assumption should be clearly documented and justified.
3. Strengthen direct clinical utility evidence. The Clinical Utility Evidence Report should strengthen the evidence specific to OncoPrecise GXP-47, including incorporation of CLARITY-LUNG trial design data and available retrospective analyses. Reliance on analogical evidence from other tumor types should be reduced and reframed as supporting precedent.
4. Source all Markov model inputs to peer-reviewed publications. Transition probabilities, cost inputs, utility values, and other model parameters cited to "internal GDT data" should be accompanied by methodology descriptions and, where possible, supplemented or replaced with published data. All proprietary data inputs should be clearly identified and labeled.
5. Update adverse event modeling. The adverse event component of the Markov model should be updated to reflect the most current published ICI safety data for NSCLC, including differentiation between ICI-associated irAEs and chemotherapy-related toxicities.
6. Add discount rate discussion. The model documentation should include an explicit discussion and justification of the discount rate assumptions (3% for costs and outcomes), with reference to applicable CMS and AHRQ guidance.
7. Remove all revenue projections and commercial data. All revenue projection, market sizing, commercial uptake, and volume forecast data should be removed from the dossier-facing sections of the HEOR analysis. These data should be maintained separately for GDT's internal commercial planning purposes.
8. Cross-reference the complete 43-publication bibliography. Larkfield should review and incorporate additional relevant publications from the master bibliography compiled by Calderwood & Harkness, particularly in the clinical utility and health economics categories. Any publications cited by Larkfield that are not in the master bibliography should be flagged for inclusion.
9. Map evidence to MolDx evidence hierarchy. The analysis should explicitly map its evidence base to the MolDx technical assessment evidence hierarchy, identifying where the current evidence falls for each domain.
10. Address the "reasonable and necessary" standard. The cost-effectiveness analysis should be framed within the context of the Social Security Act § 1862(a)(1)(A) "reasonable and necessary" standard, demonstrating that OncoPrecise GXP-47 is reasonable and necessary for the diagnosis or treatment of NSCLC.
11. Add limitations subsection. The Clinical Utility Evidence Report should include a limitations subsection acknowledging the interim nature of the CLARITY-LUNG data and the current evidentiary gaps.
12. Add budget impact analysis. A dedicated budget impact analysis from the Medicare payer perspective should be added to the cost-effectiveness section, using conservative volume assumptions.
13. Add tornado diagram. A tornado diagram (one-way sensitivity analysis summary plot) should be added to the cost-effectiveness section, illustrating the key drivers of ICER variability.
14. Note LDT regulatory status. The analysis should clearly note that OncoPrecise GXP-47 is marketed as a laboratory-developed test under CLIA authority without FDA clearance or approval, and should avoid any claims or language that could be construed as implying FDA authorization.
IX. CONCLUSION AND NEXT STEPS
Based on our review, the Larkfield HEOR analysis provides a reasonable foundation for the clinical utility and cost-effectiveness components of the MolDx Technical Assessment Dossier, but requires the material revisions summarized in Section VIII above before it is suitable for incorporation into the dossier. The most significant concerns relate to (a) the reliance on CLARITY-LUNG interim data that has not met pre-specified performance thresholds, (b) the use of an unestablished test cost assumption without adequate sensitivity analysis, (c) the insufficient direct clinical utility evidence specific to OncoPrecise GXP-47, and