Real-world clinical and economic outcomes for patients with advanced non–small cell lung cancer enrolled in a clinical trial following comprehensive genomic profiling via liquid biopsy

Abstract

BACKGROUND:

Oncology clinical trial enrollment is strongly recommended for patients with cancer who are not eligible for established and approved therapies. Many trials are specific to biomarker-targeted therapies, which are typically managed as specialty pharmacy services. Comprehensive genomic profiling (CGP) of advanced cancers has been shown to detect biomarkers, guide targeted treatment, improve outcomes, and result in the clinical trial enrollment of patients, which is modeled to offset pharmacy costs experienced by US payers, yet payer policy coverage remains inconsistent. A common concern limiting coverage of CGP by payers is the potential of identifying biomarkers beyond guideline-recommended treatments, which creates a perception that insurance companies are being positioned to “pay for research.” However, these biomarkers can increase clinical trial eligibility, and specialty pharmacy management may have an interest in maximizing the clinical trial enrollment of members.

OBJECTIVE:

To investigate if clinical trial enrollment following liquid biopsy CGP for non–small cell lung cancer (NSCLC) is clinically and/or economically impactful from a payer claims perspective.

METHODS:

Clinical and economic outcomes were studied using a real-world clinical genomic database (including payer claims data) from patients with NSCLC who enrolled in clinical trials immediately following liquid biopsy CGP (using Guardant360) and matched NSCLC patient controls also tested with liquid biopsy CGP.

RESULTS:

Real-world overall survival was significantly (log-rank P < 0.0001) better for patients enrolled in clinical trials with similar costs of care, albeit with more outpatient encounters among those enrolled compared with matched controls.

CONCLUSIONS:

The results, together with previous analyses, suggest that, in addition to the clinical benefits associated with targeted therapies directed by CGP and other testing approaches, payers and specialty pharmacy managers may consider clinical trial direction and enrollment as a clinical and economic benefit of liquid biopsy CGP and adopt this into coverage decision frameworks and formularies.

Plain language summary

People with advanced lung cancer often have their tumors tested using a method called next-generation sequencing (NGS). In this study, patients had NGS testing that looked at the tumor’s DNA found in the blood. Of the people tested, those who took part in research studies after this testing lived longer compared with similar patients who did not join the studies. The costs of health care were about the same for both groups. These broad tests can help find treatments that work well for patients and may guide them toward research studies that can also be helpful.

Implications for managed care pharmacy

Clinical trials benefit patients and divert specialty drug costs, but barriers to enrollment include access, costs and coverage, and delays from tissue genomic tumor profiling. Our study shows that patients with non–small cell lung cancer enrolling in clinical trials following liquid biopsy had improved survival without increasing costs compared with matched controls. Specialty drug management should include coverage for testing likely to guide the appropriate use of therapies, which improves survival and drives clinical trial enrollment, also improving survival.

Oncology clinical trials are instrumental to evaluating therapeutic regimens in advanced cancer care. The National Comprehensive Cancer Network promotes clinical trials as the best management when appropriate for any patient with cancer and encourages patient participation.^1,2 Clinical studies suggest patient enrollment in clinical trials is associated with improved outcomes, although studies are heterogenous.^3,4 Clinical trials in the era of “precision medicine,” ie, tailoring patient therapy to their tumor’s molecular profile, have led to a growing class of trials for which the genomic profiling of a patient may be essential to determining eligibility.⁵ Indeed, matched biomarker-targeted treatment demonstrates increased survival compared with nonspecific systemic therapy in patients with advanced cancer, including metastatic non–small cell lung cancer (NSCLC) for which approved targeted therapies are now the standard of care in patients with appropriate biomarkers.^6,7 When examining clinical trial enrollment in real-world NSCLC populations, studies lack consensus regarding clinical outcome benefit for enrollment, although payer cost savings were seen; these represent older studies that would lack abundant precision oncology trials.^8,9 Targeted therapies are typically managed within specialty pharmacy, and the appropriate utilization of these drugs within or outside of clinical trial protocols potentially has a positive clinical and economic impact on oncology health care.

The advancement of genomic technologies has grown precision oncology initiatives dramatically, thus providing increased opportunities for clinical trials to improve patient outcomes using innovative biomarker-matched therapies. Indeed, the use of comprehensive genomic profiling (CGP) via next-generation sequencing (NGS) panels in clinical practice is included in clinical guidelines for many advanced solid tumors.^10,11 The use of genomic profiling is a driver of clinical trial enrollment in patients with advanced cancer, even doubling clinical trial accrual in one study.¹² Recent studies demonstrate improved outcomes for patients with NSCLC receiving timely genomic results via CGP NGS panels to guide treatment.^13,14 CGP may also identify patients for opportunistic enrollment in clinical trials.¹⁵ In 2022, the American Society of Clinical Oncology issued a Provisional Clinical Opinion recommending CGP NGS panels (defined as ≥ 50 genes) to not only identify actionable variants, which are associated with therapies approved by the US Food and Drug Administration (FDA), but also to enhance clinical trial enrollment, including trials recruiting patients with particular biomarkers.⁷ For instance, in 2017, 34% of industry-sponsored clinical trials used genomic biomarkers to stratify patients.¹⁵ NSCLC is a cancer type for which biomarker-driven clinical trials for precision oncology therapeutics are longstanding, and, with almost a dozen biomarker-targeted approved therapies, a significant proportion of metastatic patients harbor an actionable biomarker for therapy in a clinical care setting or for clinical trial.^16,17 Master protocols, such as basket trials and umbrella trials, have also grown in recent years and may enroll or stratify participants based on genomic biomarkers.^18-21 Such trials, in which the genomic marker rather than cancer type is the key factor, have resulted in several recent pan–solid-tumor approvals for biomarkers.²²

Despite the opportunity to be enrolled in a clinical trial and to receive the best available treatment, few eligible patients choose to enroll, facing systemic and individual barriers, such as financial concerns, failure of providers to offer trial opportunities, and uncertainty of risk associated with trials.²³ Costs associated with clinical trials fall broadly into 3 categories: investigational care borne by sponsors, routine care borne primarily by payers that may fall to the sponsor or patient as out-of-pocket costs, and nonmedical costs that primarily fall to the patient or may be offset by sponsor or philanthropic support.²⁴ Financial barriers to enrollment are complex but include insurance network agreements, commercial vs government insurance programs and administrators, patient concerns for costs of services, and insurance coverage.^9,25 Additionally, the unavailability, depletion, or quality of tumor tissue has been reported among the barriers to clinical trial enrollment for NSCLC, and delays associated with tumor tissue genotyping required for clinical trials limited some patients’ enrollment because of worsening performance status, even among biomarker-positive eligible patients.^26,27 Plasma-based circulating tumor DNA (ctDNA) testing (the most common type of liquid biopsy) has been adopted readily into clinical practice for NSCLC and many other advanced solid cancers, including several FDA-approved companion diagnostic assays; it has also been used effectively for clinical trial identification enrollment in study protocols.^18,28 Not only does CGP support the identification of patients who benefit from specialty targeted therapies, but CGP liquid biopsy may potentially allow for increased effective clinical trial direction, potentially diverting some specialty drug costs from payers.

In the last 2 decades, coverage has expanded for patients enrolled in clinical trials. Starting in the year 2000, Medicare reimbursement for routine costs in clinical trials was established and in 2007 was codified as a national coverage determination. In 2010, the Affordable Care Act was enacted and stipulated that health plans regulated under the statute may not keep patients from clinical trial enrollment, must pay for routine costs of clinical trials, and cannot increase rates for those patients.²⁴ Most recently, the Clinical Trials Act was passed in 2020, which guarantees coverage of the routine care costs of clinical trial participation for Medicaid enrollees with life-threatening conditions, like cancer. Although participation in clinical trials is generally expected to improve patient outcomes, it also follows that services rendered under clinical trial protocols could represent net savings for commercial insurers because investigational costs fall to the study sponsor.^29-31 Although studies investigating cost savings associated with clinical trial enrollment are heterogeneous, in precision oncology, cost savings have been demonstrated for patients with advanced solid tumors. For example, patients enrolled in clinical trials providing second-line therapeutic drugs for NSCLC resulted in an average savings to payers of $6,663 per patient per month (PPPM).⁹ Another analysis of drug cost diversion in precision oncology clinical trials estimated the specialty pharmacy savings to a health plan of $25,000 per patient (assuming a 3.23-month average treatment duration).³²

Although tissue or liquid biopsy NGS genomic profiling is routine clinical practice, supported by professional society and clinical practice guidelines for many advanced solid cancers, including NSCLC, under-genotyping or testing via limited panels may lead to underutilization of targeted treatments and, importantly, eligible clinical trial opportunities. Coverage decision frameworks for genomic tumor profiling used by payer policy makers do not typically include clinical benefits beyond the improved outcomes following biomarker-directed specialty drugs. Previous studies suggest that CGP via NGS panels results in clinical and economic benefits specifically related to clinical trial direction. The goal of this study was to understand the real-world clinical and economic impact of clinical trials following CGP liquid biopsy NGS in advanced NSCLC, including survival outcomes and health care resource utilization and costs. For this, we queried a real-world evidence database, GuardantINFORM, and compared patients with NSCLC enrolled in a clinical trial following CGP with patients with zero clinical trial claims as potential controls.

Methods

DATA SOURCE

The study used secondary data from the nationally representative clinical genomic GuardantINFORM database that comprised anonymized genomic data for more than 200,000 patients with advanced stage solid tumors in the United States who underwent tumor genomic profiling via a Guardant360 CDx or Guardant360 assay from June 1, 2014, to March 31, 2022. The assays are certified according to the Clinical Laboratory Improvement Amendments and accredited by the College of American Pathologists and the New York State Department of Health. Guardant360 CDx has been approved by the FDA for clinical testing of patients with advanced stage (stage III-IV) solid tumors. Guardant360 assays use hybrid capture technology and NGS to identify genomic alterations in 74 to 83 genes via liquid biopsy, including guideline-recommended biomarkers for NSCLC as well as microsatellite instability and tumor mutational burden (Guardant360 only). In a validation study of 10,593 samples from patients with solid tumors tested with liquid biopsy, the technical success rate was higher than 99.6% and the clinical sensitivity was 85.9%.³³ NSCLC studies comparing on-label biomarker detection between liquid and tissue biopsies report concordances between 90% and 100%.^34-36 In this study, patients were tested with 1 of the 2 panels. For the results of this study, both panels are combined and called “Guardant360.” Beyond ctDNA test results, the GuardantINFORM database includes structured commercial payer claims data collected from inpatient and outpatient facilities in both academic and community settings. The database does not include clinical features that are not coded as claims, such as the Eastern Cooperative Oncology Group score. Deaths are sourced from third-party providers and aggregated with administrative claims data; for at least half of the Centers for Disease Control and Prevention–reported deaths in the country, the parent dataset has an encounter within less than or equal to 1 month of the date of death.

ETHICS STATEMENT

The generation of deidentified data sets by Guardant Health for research purposes is approved by the Advarra institutional review board. The GuardantINFORM database is a fully deidentified database that complies with Sections 164.514 (a)–(b)1ii of the US Health Insurance Portability and Accountability Act regarding the determination and documentation of statistically deidentified data. As the data has been deidentified, informed consent was not obtained from patients whose information was used as part of the analysis.

STUDY DESIGN

A retrospective matched cohort design was used for this study (Supplementary Figure 1 ^{(481KB, pdf)} , available in online article). The index date was defined as the date that a patient’s first Guardant360 results were reported. The baseline period consisted of the 180 days before the index date. The follow-up period spanned from the index date to the earliest of the following: 360 days after and including the index date, the last claim activity date, or death. To ensure the last claim activity date was based on observance of continuous claims, patients were censored at the last claim date that occurred before a gap of more than 90 days between claims during the follow-up period.

To study the phenomenon of CGP leading to clinical trial enrollment most accurately, patients were included in the clinical trial arm if they had at least 2 claims with a clinical trial diagnosis code within 90 days after the index date. It was not possible to discern the nature of the specific trials; therefore, proximity to CGP testing is assumed to represent enrollment with the highest likelihood to be informed by Guardant360 testing.

MATCHING ADJUSTMENT

Using coarsened exact matching, clinical trial arm patients were matched at a ratio up to 1:10 (average matching rate was ~7:1 because not all clinical trial patients had 10 matched controls) with control arm patients who had no clinical trial diagnosis codes within 90 days after the index date. Matched patients with evidence of clinical trial participation were each assigned a weight of 1 and the corresponding control patients were each assigned a weight of 1 / (number of controls matched to a given clinical trial arm patient). For example, if there were 7 control patients that matched a clinical trial arm patient, each of these patients were given a weight of 1 / 7 in the matching adjusted analysis. The matching criteria included demographic characteristics (age ± 5 years, sex, and US Census region), an indicator of comorbidity burden (Elixhauser comorbidity index weighted score ± 0.2 SDs),^37-39 and an indicator of potential aggressiveness of metastatic disease (time from metastatic diagnosis to receiving first Guardant360 test results ± 30 days). Shorter time from diagnosis to test result was hypothesized to indicate expedited treatment decision-making in the context of aggressive metastatic disease.

STUDY POPULATION

Patients were required to meet the following inclusion criteria: have NSCLC reported on their Guardant360 test requisition form, at least 1 claim with a metastatic diagnosis code (the claim with the first metastatic diagnosis must occur before first Guardant360 report date), at least 1 medical claim and 1 pharmacy claim within 180 days before the index date, at least 1 claim with NSCLC diagnosis code and metastatic diagnosis codes during the 180 days before the index date, at least 90 days of follow-up time after the index date or until death (patients who died before 90 days were still included), and continuous enrollment (≥ 1 claim every 90 days). We excluded patients who died before the index date or continued to receive claims 180 days after the death date provided by the third-party data source. Patients who had claims with a clinical trial diagnosis code before the index date and those who had only 1 claim with a clinical trial diagnosis code in the 90 days after the index date were also excluded. Supplementary Table 1 ^{(481KB, pdf)} provides the list of diagnosis codes used to identify participation in a clinical trial, NSCLC, and metastatic disease.

STUDY MEASURES

For this study, we evaluated real-world overall survival (rwOS), payer claims costs, and health care resource utilization. The rwOS was defined as the time from the Guardant360 report (index) date to death. Costs were evaluated on a PPPM basis (30-day cycles) and were classified as either medical costs (eg, hospital) or pharmacy costs. Health care resource utilization was measured by the proportion of patients with inpatient visits and/or emergency department (ED) visits. Additionally, the number of days in the hospital and the number of outpatient visits were assessed on a PPPM basis (30-day cycles).

STATISTICAL ANALYSIS

All analyses were performed using SAS Enterprise Guide 8.3 (SAS Institute, Inc.) and R version 4.3.1. Patient characteristics that were reported during the baseline and follow-up periods were described using means, SD, and medians for continuous variables (also 25th and 75th percentiles for costs) and frequencies and proportions for categorical variables. Matching adjusted Kaplan-Meier survival curves were used to describe rwOS rates for matched patients with and without evidence of clinical trial enrollment within 90 days of their index date. P values for comparisons between cohorts were obtained from t tests for continuous variables (skewed variables, such as the time from metastatic diagnosis to receiving a Guardant360, number of days in the hospital, and costs, were assessed using the nonparametric Mood test for medians), chi-square tests for categorical variables, and log-rank tests for Kaplan-Meier rates. A P value less than 0.05 was considered statistically significant.

Results

In total, 18,008 patients met the study inclusion criteria (Figure 1). Of these patients, 393 were included in the clinical trial enrollment arm (approximately 2% of the cohort were enrolled in a clinical trial based on our definition); the remaining 17,615 patients with zero clinical trial claims served as potential controls. After matching, there were 323 patients in the clinical trial enrollment arm and 2,254 in the control arm.

FIGURE 1.

Identification of Patients With Metastatic NSCLC Who Enrolled in a Clinical Trial Arm and Corresponding Matched Controls

PATIENT BASELINE CHARACTERISTICS

After matching, patients in the clinical trial arm had a similar mean age of 62.9 years compared with 63.3 years in the control cohort at the index date, with a median age of 63 years for both cohorts (Table 1). For both groups, the mean (median) total number of comorbidities and the weighted Elixhauser score was 5.1 (5) and 22.5 (20), respectively, with no statistically significant differences for any included comorbidities (Supplementary Table 2 ^{(481KB, pdf)} ). Female patients represented 57% of each cohort (Table 1). The mean (median) time from metastatic diagnosis to receiving a Guardant360 test report was approximately 5 months (approximately 1 month) for both cohorts. The South was the most common US Census region, representing 41.5% of patients (Supplementary Table 3 ^{(481KB, pdf)} ).

TABLE 1.

Patient Baseline Characteristics by Cohort After Matching

Baseline characteristics a	Control arm		Clinical trial arm		P
		Median		Median
N (after weighting)b	323		323
Age, mean (SD), years	63.3 (3.6)	63	62.9 (9.6)	63	0.55
Sex, n (%)
Female	184 (57.0)		184 (57.0)		1
Male	139 (43.0)		139 (43.0)
Time from mDx to G360, mean (SD), d	151.4 (91.6)	31	150.5 (242.7)	30	0.47
Total comorbidities, mean (SD), n	5.1 (1)	5	5.1 (2.6)	5	0.96
Elixhauser score, mean (SD)	22.5 (3)	20	22.5 (7.9)	20	0.97
Total lines of therapy,c n (%)					0.48
0	213 (65.8)		224 (69.3)
1	91 (28.1)		85 (26.3)
2+	20 (6.1)		14 (4.3)
Inpatient visits (yes),c n (%)	153 (47.5)		140 (43.3)		0.29
ED visits (yes),c n (%)	86 (26.5)		70 (21.7)		0.15
Total days in hospital (PPPM), mean (SD)	0.6 (0.5)	0	0.5 (1.2)	0	0.19
Total outpatient visits (PPPM), mean (SD)	2.2 (0.6)	1.8	2.4 (1.6)	2	0.15
Medical cost (PPPM), mean (SD), USD	3,327 (1,897)	1,624	3,439 (4,614)	1,830	0.20
25th percentile	737		831
75th percentile	3,901		3,701
Pharmacy cost (PPPM), USD	939 (1,156)	144	845 (2,483)	135	0.62
25th percentile	40		24
75th percentile	461		385

^aBaseline period is 180 days before the first G360 final report date.

^bMatching was performed up to 10:1, resulting in a total unweighted control arm of 2,254 patients.

^cThe weighted number of patients in the control arm is rounded to the nearest whole number.

ED = emergency department; G360 = Guardant360; mDx = metastatic diagnosis; PPPM = per patient per month; USD = US dollars.

The baseline characteristics not included in the matching algorithm were also comparable. The year of the index date and first metastatic diagnosis was similar between the cohorts, with the majority occurring in 2019 or later. During the baseline period, 69.3% of the clinical trial arm and 65.8% of the control arm had no prior lines of therapy. The cohorts had similar health care resource utilization at baseline, with most patients not having an inpatient or ED visit. The clinical trial arm had 2.4 outpatient visits PPPM compared with 2.2 for the control arm, but the difference was not statistically significant. The mean and median medical costs and pharmacy costs PPPM were not statistically significantly different between the 2 arms.

CLINICAL AND ECONOMIC OUTCOMES AT FOLLOW-UP

Clinical trial patients had longer rwOS compared with patients in the control arm (log-rank P < 0.0001) (Figure 2); however, the median survival was not reached for either cohort (80th percentile landmark, clinical trial arm: 333 days vs control arm: 140 days). The CI for the control arm is tighter than that of the clinical trial arm because of its higher effective sample size because of the up to 10 (control):1 (clinical trial) matching process. In both cohorts, most patients had at least 1 inpatient visit during the follow-up period (Table 2). However, most patients in each group did not have an ED visit during this time. Patients in the clinical trial arm had a similar number of median days in the hospital PPPM compared with the control group but had more outpatient visits PPPM (3.3 vs 2.6, P < 0.001). There were not statistically significant differences in median PPPM medical costs between the cohorts (clinical trial arm: $4,968 vs control arm $4,001). The mean PPPM pharmacy costs were lower for the clinical trial arm (clinical trial arm: $1,371 vs control arm: $2,126); the median PPPM pharmacy costs were similar between the groups (clinical trial arm: $265 vs control arm: $259). Lastly, the clinical trial arm patients had greater mean (clinical trial arm: 286 days vs control arm: 244 days) and median follow-up time (clinical trial arm: 360 days vs control arm: 279 days).

FIGURE 2.

Real-World Overall Survival by Cohort After Matching

TABLE 2.

Cost and Utilization Outcomes by Cohort After Matching

Outcomes a	Control arm		Clinical trial arm		P b
		Median		Median
N (after weighting)c	323		323
Inpatient visits (yes),c n (%)	171 (53.1)		187 (57.9)		0.22
ED visits (yes),c n (%)	119 (36.8)		112 (34.7)		0.57
Total days in hospital (PPPM), mean (SD)	1.3 (0.9)	0.2	0.9 (1.6)	0.3	0.75
Total outpatient visits (PPPM), mean (SD)	2.6 (0.7)	2.3	3.3 (1.4)	3.2	<0.01d
Medical cost (PPPM), mean (SD), USD	7,176 (3,044)	4,001	6,674 (5,905)	4,968	0.06
25th percentile	1,190		2,037
75th percentile	11,351		10,159
Pharmacy cost (PPPM), mean (SD), USD	2,126 (1,816)	259	1,371 (3,315)	265	0.93
25th percentile	66		79
75th percentile	1,002		688
Follow-up time, mean (SD), d	244 (46)	279	286 (98)	360	<0.0001

^aBaseline period is 180 days before the first Guardant360 final report date.

^bThe P value compared here was for the mean, including in rows in which both the mean and median are reported.

^cMatching was performed up to 10:1, resulting in a total unweighted control arm of 2,254 patients. The weighted number of patients in the control arm is rounded to the nearest whole number.

^dThis denotes that the statistical significance was at the α = 0.05 level.

ED = emergency department; PPPM = per patient per month; USD = US dollars.

Discussion

Clinical trial enrollment within 90 days of CGP using liquid biopsy for advanced NSCLC resulted in improved rwOS at no increased medical or pharmacy costs compared with matched patients who did not enroll in a clinical trial. The utilization of health care services was similar overall, albeit with more outpatient visits in the clinical trial arm vs the matched controls. To our knowledge, this study represents the first claims-based analysis of the impact of clinical trial enrollment following liquid biopsy CGP, providing insights into the potential benefits of comprehensive (NGS) liquid biopsy testing to drive clinical trial enrollment.

Direction to clinical trials may be an underreported benefit of CGP, however many other benefits are well established. In addition to increased biomarker detection and increased direction to targeted therapies, CGP NGS panel testing is also guideline recommended and supported by the American Society of Clinical Oncology because of the improved outcomes among biomarker-positive patients treated with targeted treatments (Figure 3).^1,7,40 Although CGP is growing in clinical practice, most national payers are hesitant to provide formal insurance coverage policies and reimbursement for testing. A 2015 survey found that US national payers considered the clinical utility of pan–cancer tumor profiling via NGS a “reasonable hypothesis,” yet 60% considered it investigational, citing the concern of dramatically increased off-label drug use.⁴¹ In reality, studies demonstrate that oncologists utilizing CGP rarely use results to prescribe off-label therapies (5%) but are, in fact, likely to recommend clinical trial enrollment (43%).⁴² Professional society opinions also guide clinicians away from off-label prescription when a clinical trial is available.⁷ This suggests that the concerns raised by payers regarding CGP coverage may be overcome by the potential benefit of clinical trials. Payers have also expressed that a limiting factor of coverage for broad genomic testing is a general objection to “paying for research” despite the well-validated assays used in studies like this one.⁴³ This is rooted in concerns that increased CGP may result in “therapeutic anarchy.”⁴⁴ The present study and other analyses suggest that coverage of testing that may drive clinical trial enrollment does not significantly increase the cost of specialty pharmaceutical treatments, which are a large driver of cancer care cost in NSCLC, nor do overall costs of care increase significantly.⁴⁵ Traditional evidence frameworks employed by commercial payers typically evaluate the benefit of a test, treatment, or device in terms of clinical outcome or change to medical management.⁴¹ However, in oncology, the ability of a comprehensive test to direct a patient toward clinical trials may warrant inclusion in the analysis by payer policy writers and specialty pharmacy managers, given the demonstrable clinical and economic benefits. Oncologists report that insurance coverage is a key barrier to the use of CGP testing, along with turnaround time, patient concern for out-of-pocket costs, the Medicare 14-day rule, and inadequate reimbursement.^46,47

FIGURE 3.

Conceptualization of the Clinical Benefits, Costs, or Harms Associated with Comprehensive Genomic Profiling Next-Generation Sequencing Panels

Such inclusion is particularly important for tumor types such as NSCLC, which has a number of clinically actionable biomarkers as well as multiple targeted therapies in development and in clinical trials.⁴⁸ Targeted therapies have demonstrated dramatic improvements in patient outcomes (including OS) and genomic profiling via liquid biopsy enables direction to targeted therapy vs systemic treatment in advanced NSCLC.^40,45 In cost analyses of tumor profiling approaches, the most impactful driver of cost is the cost of targeted therapies.^45,49 Although targeted therapies in biomarker-positive NSCLC can have considerable costs, in the absence of molecular genomic biomarker results, the alternative treatment paradigm of chemotherapy with immunotherapy (accompanied by infusion and time costs) present a similar, if not greater, economic impact.⁵⁰ A study in patients with advanced NSCLC treated with tyrosine kinase inhibitors demonstrated that the largest drivers of costs were inpatient services and hospitalizations, and total costs were significantly higher for patients treated with a tyrosine kinase inhibitor in the second line compared with in the first line.^51,52 Because of the improved outcomes, multiple guidelines recommend broad panel-based molecular profiling by ctDNA and/or tumor tissue in advanced NSCLC to detect targetable alterations and identify patients eligible to enroll in clinical trials.^1,7,10,53 Specialty pharmacy managed care may support the appropriate utilization of targeted therapies by adjusting formularies and policies to encourage biomarker testing results before approval (Supplementary Figure 2 ^{(481KB, pdf)} ). The appropriate utilization of immunotherapy and chemotherapies in NSCLC may be effectively managed by encouraging genomic assessment by CGP and prioritizing these treatments on tumors that are biomarker-negative without clinical trial eligibility. The collaboration between specialty pharmacy management and genomic testing coverage policies for advanced NSCLC may enhance the appropriate utilization of therapies as well as drive clinical trial enrollment.

LIMITATIONS

There are limitations of the present study that may warrant consideration before generalized application. Limitations are inherent to the dataset. (1) This is an observational dataset with potentially unobserved confounding variables, such as the Eastern Cooperative Oncology Group score, the severity of disease, and provider type. (2) Because of the patient dynamics of health insurance, missing claims data may exist; however, we attempted to ensure continuous enrollment by ensuring patients had a claim at least every 90 days during the follow-up period. (3) The patients were not individually staged for this project; however, Guardant360 testing requires a diagnosis of advanced/metastatic solid cancer by the ordering provider, and this study required patients to have at least 1 claim with a metastatic diagnosis code during the 180-day baseline period. (4) Ethnic and racial data were not available. (5) Access to clinical trials may occur outside of claims billed, such as patient access programs through pharmaceutical sponsors. (6) Although the analysis was limited to clinical trial enrollment within 90 days of Guardant360 testing, the exact nature of the clinical trial was unknown. However, the observed percentage of clinical trial enrollment is similar to previous studies in oncology that observed rates between 2% and 8%.^54,55 (7) All patients in this study received Guardant360 testing, which may present a biased selection of patients and providers with interest and comfort using CGP via plasma-based technologies or the size of testing may impact clinical trial identification for biomarkers not tested. However, the population studied reflects diverse practitioners who vary in the treatment of patients with advanced cancer and coding for services/therapies. Furthermore, the patient population is younger than the median NSCLC age at diagnosis in the United States.⁵⁶ Lastly, the data period spanned the coronavirus disease 2019 pandemic, and thus the findings may be affected by its impacts on health care utilization. Future studies should expand this analysis to other cancer types. Clinical cancer centers may explore specific clinical trial enrollment in ways that the current real-world evidence was limited.

Conclusions

Patients with advanced NSCLC enrolled in clinical trials within 90 days of CGP by Guardant360 were shown to have improved rwOS without increased medical or pharmaceutical payer claim costs as compared with matched patients not enrolled in a clinical trial. Although future studies may clarify the types of trials in which patients enroll following testing and what trials may impact survival, the current study not only supports guideline recommendations for the use of liquid biopsy CGP for clinical trial enrollment but also supports economic parity or even benefit for payers whose members enroll in clinical trials.