Real world treatment patterns, healthcare resource use and costs in patients with advanced or metastatic non-small cell lung cancer by EGFR mutation type

Abstract

Aims

This study described treatment patterns, healthcare resource utilization (HRU) and costs among advanced or metastatic non-small cell lung cancer (a/mNSCLC) patients with different epidermal growth factor receptor (EGFR) mutation types.

Materials and methods

This retrospective study leveraged NeoGenomics NeoNucleus linked with IQVIA PharMetrics Plus between 01 January 2016 to 30 April 2021 (study period). Patients with evidence of a/mNSCLC between 01 July 2016 to 31 March 2021 (selection window) with EGFR test results indicating exon 19 deletion (exon19del), exon 21 L858R (L858R), or exon 20 insertion (exon20i) mutations were included; date of first observed evidence of a/mNSCLC was the index date. Treatment patterns, all-cause HRU and costs during ≥1 month follow-up were reported for each cohort (exon19del, L858R, and exon20i).

Results

A total of 106 exon19del, 75 L858R, and 13 exon20i patients met the study criteria. The prevalence of hospitalization was highest in the exon20i cohort (76.9%), followed by L858R (62.7%) and exon19del (55.7%) cohorts. A higher proportion of patients had evidence of hospice/end-of-life care in the exon20i (30.8%) and L858R (29.3%) cohorts relative to the exon19del cohort (22.6%). The exon20i cohort had higher median total healthcare costs per patient per month ($27,069) relative to exon19del ($17,482) and L858R ($17,763). EGFR tyrosine kinase inhibitors (TKI) were the most frequently observed treatment type for exon19del and L858R cohorts, while chemotherapy was the most observed treatment in exon20i cohort.

Limitations

The sample size for the study cohorts was small, thus no statistical comparisons were conducted.

Conclusions

This is one of the first real-world studies to describe HRU and costs among a/mNSCLC patients by specific EGFR mutation type. HRU and costs varied between EGFR mutation types and were highest among exon20i cohort, potentially reflecting higher disease burden and unmet need among patients with this mutation.

PLAIN LANGUAGE SUMMARY

Patients with non-small cell lung cancer (NSCLC) in an advanced or metastatic stage (a/mNSCLC) where cancer has spread to other parts of the body have high chance of dying within five years. Treatment and management of a/mNSCLC also incurs significant healthcare resource utilization (HRU) and costs. Patients with a/mNSCLC may have their epidermal growth factor receptor (EGFR) gene mutated with different variations. Our study described what a/mNSCLC patients were treated with, their HRU and healthcare costs separately for the following three types of EGFR mutations: exon 19 deletion (exon19del), exon 21 L858R (L858R), or exon 20 insertion (exon20i). Our study found that patients with exon19del or L858R mutation were commonly treated with EGFR tyrosine kinase inhibitors (TKIs), while exon20i patients were mostly treated with chemotherapy due to lack of targeted treatment for exon20i during the time when the study was conducted. HRU and healthcare costs were highest for patients with exon20i, which shows that patients with exon20i face high burden and have a need for new treatment options.

Keywords:

• Exon 20 insertion
• EGFR mutations
• exon 19 deletion
• L858R mutation
• non-small cell lung cancer
• treatment pattern
• healthcare resource utilization

JEL CLASSIFICATION CODES:

• I11
• I1
• I
• I10

Introduction

Lung cancer is the third most common cancer in the United States (US) and the leading cause of cancer mortality, with an estimated 238,340 new cases and 127,070 deaths in 2023, accounting for 12.2% of all new cancer cases and 20.8% of all cancer deaths¹. Non-small cell lung cancer (NSCLC) accounts for 80–90% of all lung cancers and is mostly diagnosed at an advanced stage^2,3. In addition, the five-year survival rate is approximately 34% for patients initially diagnosed at regional stage and merely 7% for patients diagnosed at distant stage⁴. In the US, approximately 24% of patients with NSCLC have a sensitizing mutation in the epidermal growth factor receptor (EGFR) gene that makes the tumor cells dependent on EGFR signaling and thus, a therapeutic target for molecular inhibition via TKIs⁵. Previous studies have suggested that EGFR mutation was associated with improved overall survival compared to EGFR wild type partly due to availability of TKIs targeting EGFR and better clinicopathologic characteristics associated with EGFR mutation^6–8. To that end, the 2023 National Comprehensive Cancer Network (NCCN) guidelines recommend molecular testing for all advanced or metastatic NSCLC (a/mNSCLC) cases, especially those of adenocarcinoma subtype, to ensure the most appropriate treatment strategy⁹.

The most common EGFR mutations are in-frame exon 19 deletion (exon19del) and exon 21 L858R point mutation (L858R), comprising approximately 45% and 44% of all EGFR mutations detected, respectively¹⁰. The 2023 NCCN guidelines recommend EGFR TKIs for treatment of a/mNSCLC with EGFR exon19del or L858R mutations⁹. Previous studies have shown better survival outcomes and response rate for exon19del compared to L858R mutations after EGFR TKI therapy^11–13. EGFR exon 20 insertion mutations (exon20i) are more rare (5–12%)¹⁴ and are associated with de novo resistance to TKIs, limiting the treatment options for these patients¹⁵.

Prior studies have shown that healthcare costs remain a cornerstone in oncology^16,17. In 2020, the estimated national expenditures for care of lung cancer was $23.8 billion¹⁷, with attributed costs of $109,103 per NSCLC patient in the last year of life¹⁷. A recent claims-based retrospective study found total all-cause costs of $25,499 per patient per month (PPPM) during the first line of therapy (1LOT), which consisted of drug-related costs ($15,356), other outpatient costs ($5400), inpatient costs ($4167), and other costs ($576) among EGFR mutated advanced NSCLC patients¹⁸. Similarly, there are many studies that describe cost-effectiveness of NSCLC treatment or associated healthcare costs among EGFR mutated NSCLC patients receiving treatments^19–21. Within EGFR mutated a/mNSCLC, differences in disease subtype based on EGFR mutation types or differences in responsiveness to available targeted therapies may lead to differences in HRU and healthcare costs²²; however, there is a lack of real-world evidence on the HRU and costs for specific EGFR mutation types.

This retrospective study leveraged a large claims database linked to genomic data to describe patient characteristics, treatment patterns, HRU and costs of a/mNSCLC patients by the specific EGFR mutation type: exon19del, L858R and exon20i.

Methods

Study design and data sources

This retrospective, descriptive cohort study leveraged laboratory-based genomic testing data from NeoGenomics NeoNucleus (NeoGenomics) linked to medical and pharmacy claims from IQVIA’s PharMetrics Plus database from 1 January 2016 to 30 April 2021 (study period; Supplementary S1). NeoGenomics is the largest cancer testing database in the US with approximately 1 million oncology tests performed per year. NeoGenomics includes data from >4400 practices across the US from all practice types and representation in all states. IQVIA PharMetrics Plus database comprises of fully adjudicated medical and pharmacy claims for more than 210 million unique enrollees and is representative of the commercially insured US population under 65 years old. The linkage of patient data between the two databases was achieved by a HIPAA-compliant deterministic matching algorithm which utilized de-identified data. An encryption algorithm was applied at the source to de-identify patient-level information by creating Patient IDs which were matched across datasets²³.

Study population

Three mutually exclusive study cohorts were created: exon19del, L858R, and exon20i. The exon19del and L858R cohorts included adult patients (≥18 years) with a/mNSCLC between 1 July 2016 to 31 March 2021 (selection window) in PharMetrics Plus with common EGFR mutation (exon19del or L858R) positive test results between 1 January 2016 and 31 December 2020 in NeoGenomics (Figure 1). Patients were considered to have lung cancer if they had ≥1 inpatient or ≥2 outpatient diagnoses of lung cancer in any position of the claim on separate days in PharMetrics Plus¹⁸. Patients were also required to have evidence of advanced or metastatic cancer, defined as ≥1 antineoplastic therapy or ≥1 diagnosis for metastatic cancer (Supplementary S2) in any position, on or within 30 days following the first lung cancer diagnosis. The date of earliest observed evidence of advanced or metastatic cancer was defined as the index date. All patients were required to have ≥6 months of continuous enrollment before index date and ≥1 month after the index date. Other inclusion and exclusion criteria are detailed in Figure 1. Patients were stratified into exon19del or L858R cohorts based on their EGFR mutation type.

Figure 1. Patient selection for common EGFR mutation cohorts (exon19del and L858R). *Advanced or metastatic cancer was defined as evidence of ≥1 antineoplastic therapy or ≥1 diagnosis code for metastatic cancer in any position. Abbreviations. EGFR, epidermal growth factor receptor; TKI, tyrosine kinase inhibitor.

The exon20i cohort was selected separately to maximize sample size due to the low prevalence of this mutation type using similar eligibility criteria (Figure 2) yet al.so allowing for evidence of lung cancer to be defined as a sample collected from the lung in NeoGenomics.

Figure 2. Patient selection for exon 20 insertion cohort (exon20i). *Advanced or metastatic cancer was defined as evidence of ≥1 antineoplastic therapy or ≥1 diagnosis code for metastatic cancer in any position. Abbreviation. EGFR, epidermal growth factor receptor.

Study measures

Patient demographic and baseline clinical characteristics were assessed during the 6-month baseline period; treatment patterns including LOT, all-cause HRU and healthcare costs, and location and number of incident metastatic sites were reported during ≥1 month follow-up for each cohort. EGFR testing patterns were also evaluated over the study period. The end of follow-up was defined as the earliest of end of continuous enrollment or end of the study period. LOT was defined by systemic treatment regimens using the dates of initiation and discontinuation of drugs of interest. Treatments started ≤30 days of another agent were counted within the same LOT (i.e. combination regimen). A new LOT was identified upon one of the following events: (1) switch or augmentation (i.e. additional therapies initiated after the first 30-day window within the LOT) or (2) evidence of treatment after discontinuation, defined as >60-day gap in treatment sessions after the date of last claim plus the days’ supply for oral drugs, and the last date of administration + 28 days for injectable drugs.

Statistical analysis

Study measures were described for each cohort using descriptive statistics. Categorical variables were reported in frequency (n) and percentage (%), while continuous variables were reported with mean, standard deviation (SD), median, first quartile (Q1), third quartile (Q3), minimum (min) and maximum (max). All costs included the amount paid by the insurer/payers and out-of-pocket costs by patients and were adjusted to 2021 US dollar using the medical care component of the US Consumer Price Index and reported PPPM over the follow-up for each cohort²⁴. All analyses were conducted using SAS Release 9.4 (SAS Institute Inc., Cary, NC).

Results

Study sample

A total of 106 patients with a/mNSCLC had positive results for exon19del mutation (i.e. exon19del cohort), 75 for L858R mutation (i.e. L858R cohort) (Figure 1), and 13 for exon20i mutation (i.e. exon20i cohort) (Figure 2).

Patient characteristics

On average, patients in L858R cohort were older (mean [SD] 63.3 [8.3] years) relative to exon19del (59.9 [8.6]) or exon20i (58.0 [10.7]) cohorts (Table 1). Patients were predominantly female across cohorts, ranging from 62.3% in exon19del cohort to 76.9% in exon20i cohort. For all cohorts, the majority of patients were located in the South or West region of the US (69.2–73.6%); the exon20i cohort had higher proportion of patients located in the Northeast (23.1%) relative to exon19del (7.6%) and L858R (5.3%) cohorts. The most prevalent payer type was commercial payers (61.5–68.9%) across the cohorts.

Table 1. Baseline characteristics.

Baseline characteristics	Exon 19 deletion	Exon 21 L858R	Exon 20 insertion
	N = 106	N = 75	N = 13
Age at index (years)
Mean (SD)	59.9 (8.6)	63.3 (8.3)	58.0 (10.7)
Median (Q1, Q3)	60.0 (54.0, 64.0)	62.0 (58.0, 67.0)	62.0 (50.0, 65.0)
Min, Max	41.0, 85.0	41.0, 85.0	39.0, 75.0
Gender, n (%)
Male	40 (37.7)	21 (28.0)	3 (23.1)
Female	66 (62.3)	54 (72.0)	10 (76.9)
Geographic region, n (%)
Northeast	8 (7.5)	4 (5.3)	3 (23.1)
Midwest	19 (17.9)	18 (24.0)	1 (7.7)
South	42 (39.6)	33 (44.0)	5 (38.5)
West	36 (34.0)	20 (26.7)	4 (30.8)
Other	1 (0.9)	0 (0.0)	0 (0.0)
Payer type, n (%)
Commercial	73 (68.9)	47 (62.7)	8 (61.5)
Self-insured	20 (18.9)	18 (24.0)	4 (30.8)
Medicare	7 (6.6)	7 (9.3)	1 (7.7)
Medicaid	6 (5.7)	3 (4.0)	0 (0.0)
Other/unknown	0 (0.0)	0 (0.0)	0 (0.0)
Index year, n (%)
2016	32 (30.2)	19 (25.3)	1 (7.7)
2017	19 (17.9)	12 (16.0)	3 (23.1)
2018	20 (18.9)	16 (21.3)	4 (30.8)
2019	16 (15.1)	16 (21.3)	4 (30.8)
2020	19 (17.9)	12 (16.0)	1 (7.7)
Quan CCI excluding cancer, n (%)
Mean (SD)	0.7 (1.1)	0.9 (1.3)	1.1 (2.0)
Median (Q1, Q3)	0.0 (0.0, 1.0)	0.0 (0.0, 1.0)	0.0 (0.0, 1.0)
Min, Max	2.0, 11.0	2.0, 12.0	3.0, 13.0
CCI categories excluding cancer, n (%)
0	67 (63.2)	41 (54.7)	7 (53.8)
1	20 (18.9)	21 (28.0)	4 (30.8)
2	14 (13.2)	5 (6.7)	0 (0.0)
3	1 (0.9)	2 (2.7)	1 (7.7)
≥4	4 (3.8)	6 (8.0)	1 (7.7)
Comorbidities, n (%)
Infection	70 (66.0)	45 (60.0)	10 (76.9)
Hypertension	45 (42.5)	35 (46.7)	6 (46.2)
Chronic pulmonary disease	25 (23.6)	19 (25.3)	3 (23.1)
Anemia	24 (22.6)	15 (20.0)	2 (15.4)
Fatigue/ asthenia	18 (17.0)	17 (22.7)	6 (46.2)
Anxiety or Depression	18 (17.0)	12 (16.0)	3 (23.1)
Mild liver disease	16 (15.1)	12 (16.0)	3 (23.1)
Obesity	17 (16.0)	8 (10.7)	4 (30.8)
Diabetes without complications	13 (12.3)	5 (6.7)	2 (10.0)
Cerebrovascular disease	5 (4.7)	8 (10.7)	3 (23.1)
Coronary artery disease	5 (4.7)	8 (10.7)	2 (15.4)

Abbreviations. CCI, Charlson Comorbidity Index; Q1, quartile 1; Q3, quartile 3; SD, standard deviation.

The exon20i cohort had a higher proportion of patients with Charlson Comorbidity Index (CCI) score (excluding cancer) of ≥3 (15.4%) relative to the L858R (10.7%) or the exon19del (4.7%) cohorts. The most commonly observed baseline comorbidities/conditions across cohorts were infection (60.0%–76.9%), hypertension (42.5–46.7%) and chronic pulmonary disease (23.1–25.3%). There was a higher proportion of patients in the exon20i cohort with evidence of fatigue/asthenia (46.2%) relative to exon19del (17.0%) and L858R (22.7%); a similar pattern was observed for obesity (exon20i: 30.8%; exon19del: 16.0%; and L858R: 10.7%).

During baseline, bone and bone marrow was the most common location of metastasis across all cohorts, with the highest prevalence observed in the exon20i (53.8%), followed by L858R (44.0%) and exon19del (42.5%) cohorts (Table 2). Other common locations of metastasis at baseline included central nervous system (CNS; 15.4–33.0%), lymph nodes (15.1–30.7%) and the respiratory system (7.7–33.0%). There was a higher proportion of patients having ≥2 metastasis sites during baseline in the exon19del (43.4%) and L858R (40.0%) cohorts than the exon20i cohort (30.8%).

Table 2. Metastasis during baseline and follow-up.

Metastasis	Exon 19 deletion		Exon 21 L858R		Exon 20 insertion
	N = 106		N = 75		N = 13
	Baseline	Follow-up	Baseline	Follow-up	Baseline	Follow-up
Location of metastasis (incident metastasis for follow-up) (not mutually exclusive), n (%)
Lymph nodes	16 (15.1)	19 (17.9)	23 (30.7)	15 (20.0)	3 (23.1)	2 (15.4)
Respiratory system	35 (33.0)	28 (26.4)	17 (22.7)	23 (30.7)	1 (7.7)	8 (61.5)
Central nervous system	35 (33.0)	23 (21.7)	22 (29.3)	16 (21.3)	2 (15.4)	6 (46.2)
Bone and bone marrow	45 (42.5)	27 (25.5)	33 (44.0)	17 (22.7)	7 (53.8)	4 (30.8)
Other	16 (15.1)	21 (19.8)	14 (18.7)	10 (13.3)	1 (7.7)	7 (53.8)
Unspecified	17 (16.0)	18 (17.0)	7 (9.3)	19 (25.3)	2 (15.4)	4 (30.8)
Number of metastasis sites at the baseline, n (%)
0	13 (12.3)	–	7 (9.3)	–	2 (15.4)	–
1	47 (44.3)	–	38 (50.7)	–	7 (53.8)	–
2	26 (24.5)	–	17 (22.7)	–	3 (23.1)	–
3	15 (14.2)	–	8 (10.7)	–	1 (9.2)	–
≥4	5 (4.7)	–	5 (6.7)	–	0 (0.0)	–
Number of incident metastasis sites during follow-up, n (%)
0	–	35 (33.0)	–	25 (33.3)	–	1 (7.7)
1	–	34 (32.1)	–	18 (24.0)	–	3 (23.1)
2	–	17 (16.0)	–	17 (22.7)	–	4 (30.8)
3	–	13 (12.3)	–	12 (16.0)	–	2 (15.4)
≥4	–	7 (6.6)	–	3 (4.0)	–	3 (23.1)

During ≥1 month of follow-up, the prevalence of incident metastasis was higher in the exon20i cohort (92.3%) than the exon19del (67.0%) and L858R (66.7%) cohorts. Respiratory system was the most common incident metastasis site over follow-up for all cohorts (exon19del: 26.4%; L858R: 30.7%; exon20i: 61.2%), followed by bone and bone marrow for exon19del (25.5%) and L858R (22.7%) and CNS in exon20i (46.2%).

EGFR testing patterns

All patients in the exon20i cohort and the majority of L858R (86.7%) and exon19del (69.8%) cohorts had EGFR test results within one month after the index date or before the index date. Among patients with systemic therapies observed during follow-up, all patients in exon20i cohort and most (89.4%) in L858R cohort had EGFR test results before initiation of first post-index therapy, while one-third of patients in exon19del cohort (32.0%) had EGFR test results after the initiation of first post-index therapy (Table 3).

Table 3. EGFR testing and treatment patterns during the follow-up.

	Exon 19 deletion	Exon 21 L858R	Exon 20 insertion
	N = 106	N = 75	N = 13
Duration of follow-up (months)
Mean (SD)	16.9 (14.6)	14.0 (11.7)	11.9 (9.9)
Median (Q1, Q3)	11.5 (6.0, 24.1)	9.7 (5.6, 20.1)	10.9 (3.3, 16.9)
Min, Max	1.0, 58.6	1.3 50.7	1.3, 33.3
Timing of EGFR testing results relative to index date, n (%)
Before index date (< 0 months)	47 (44.3)	47 (62.7)	9 (69.2)
≤1 month after the index date (0 to 1)	27 (25.5)	18 (24.0)	4 (30.8)
>1 and ≤3 months after the index date	3 (2.8)	4 (5.3)	0 (0.0)
>3 and ≤6 months after the index date	6 (5.7)	2 (2.7)	0 (0.0)
>6 and ≤12 months after the index date	7 (6.6)	2 (2.7)	0 (0.0)
>12 months after the index date	16 (15.1)	2 (2.7)	0 (0.0)
Timing of EGFR mutation test results relative to initiation of first post-index therapy, n (%)
Among patients that had systemic therapy during follow-up	N = 100	N = 66	N = 11
Before initiation of first therapy	68 (68.0)	59 (89.4)	11 (100.0)
After initiation of first therapy	32 (32.0)	7 (10.6)	0 (0.0)
Total LOTs^a observed during follow-up, n (%)
0	6 (5.7)	9 (12.0)	2 (15.4)
1	60 (56.6)	38 (50.7)	6 (46.2)
2	26 (24.5)	19 (25.3)	2 (15.4)
≥3	14 (13.2)	9 (12.0)	3 (23.1)
Treatments received anytime during follow-up, n (%)
No treatment^b	3 (2.8)	5 (6.7)	1 (7.7)
Systemic treatment^c	100 (94.3)	66 (88.0)	11 (84.6)
Platinum-based chemotherapy	30 (28.3)	21 (28.0)	8 (61.5)
Non-platinum-based chemotherapy	31 (29.2)	23 (30.7)	9 (69.2)
IO therapy	18 (17.0)	13 (17.3)	6 (46.2)
EGFR TKI	90 (84.9)	60 (80.0)	4 (30.8)
Afatinib	15 (14.2)	7 (9.3)	0 (0.0)
Erlotinib	31 (29.2)	19 (25.3)	0 (0.0)
Gefitinib	5 (4.7)	1 (1.3)	0 (0.0)
Osimertinib	65 (61.3)	48 (64.0)	4 (30.8)
VEGF therapy	10 (9.4)	6 (8.0)	3 (23.1)
Combination therapy^d	35 (33.0)	26 (34.7)	9 (69.2)
Radiation therapy	48 (45.3)	35 (46.7)	6 (46.2)
External beam radiation therapy	42 (39.6)	32 (42.7)	6 (46.2)
Stereotactic body radiation	12 (11.3)	5 (6.7)	3 (23.1)
Lung surgery	8 (7.5)	8 (10.7)	0 (0.0)
Lobectomy	5 (4.7)	6 (8.0)	0 (0.0)
Wedge resection	4 (3.8)	3 (4.0)	0 (0.0)
Among patients with evidence of surgery	N = 8	N = 8	–
Evidence of neoadjuvant treatment^e	0 (0.0)	0 (0.0)	–
Evidence of adjuvant treatment^f	6 (75.0)	6 (75.0)	–
Evidence of adjuvant osimertinib	4 (50.0)	4 (50.0)	–
Top 5 drug class-level therapies used during 1LOT, among patients with ≥1LOT, n (%)	N = 100	N = 66	N = 11
EGFR TKI	75 (75.0)	48 (72.7)	2 (18.2)
Non-platinum-based chemotherapy, platinum-based chemotherapy	12 (12.0)	9 (13.6)	4 (36.4)
IO therapy, non-platinum-based chemotherapy, platinum-based chemotherapy	2 (2.0)	2 (3.0)	1 (9.1)
IO therapy	2 (2.0)	3 (4.5)	1 (9.1)
EGFR TKI, non-platinum-based chemotherapy, platinum-based chemotherapy	4 (4.0)	1 (1.5)	0 (0.0)
Top 5 drug class-level therapies used during 2LOT, among patients with ≥2LOT, n (%)^g	N = 40	N = 28	N = 5
EGFR TKI	26 (65.0)	14 (50.0)	1 (20.0)
IO therapy	5 (12.5)	1 (3.6)	1 (20.0)
EGFR TKI, non-platinum-based chemotherapy, platinum-based chemotherapy	3 (7.5)	1 (3.6)	0 (0.0)
EGFR TKI, IO therapy, non-platinum-based chemotherapy, platinum-based chemotherapy	0 (0.0)	3 (10.7)	0 (0.0)
EGFR TKI, IO therapy, non-platinum-based chemotherapy, platinum-based chemotherapy, VEGF therapy	2 (5.0)	1 (3.6)	0 (0.0)
IO therapy, non-platinum-based chemotherapy, platinum-based chemotherapy	2 (5.0)	0 (0.0)	1 (20.0)
Non-platinum-based chemotherapy, platinum-based chemotherapy, VEGF therapy	2 (5.0)	0 (0.0)	1 (20.0)
Drug-level TKIs used during 1LOT (n, %)^a	N = 100	N = 66	N = 11
Osimertinib	43 (43.0)	30 (45.5)	3 (27.3)
Erlotinib	29 (29.0)	16 (24.2)	0 (0.0)
Afatinib	10 (10.0)	5 (7.6)	0 (0.0)
Gefitinib	3 (3.0)	1 (1.5)	0 (0.0)

Abbreviations. 1LOT, first line of therapy; 2LOT, second line of therapy; EGFR, Epidermal growth factor receptor; IO, immuno-oncology; LOT, line of therapy; Q1, 1st quartile; Q3, 3rd quartile; SD, standard deviation; TKI, tyrosine kinase inhibitors; VEGF, vascular endothelial growth factor.

^aLOT was defined by systemic treatment regimens using the dates of initiation and discontinuation of drugs of interest, with treatments started ≤30 days of another agent to be counted within the same LOT (i.e. combination regimen). A new LOT was identified upon one of the following events: (1) switch or augmentation (i.e. additional therapies initiated after the initial 30-day window within the LOT) or (2) evidence of treatment after discontinuation, defined as >60-day gap in treatment sessions after the date of last claim plus the days’ supply for oral drugs, and the last date of administration + 28 days for injectable drugs. LOTs in this table were defined at drug-level unless specified otherwise.

^bNo treatment was defined as not having evidence of systemic treatments, radiation therapy or lung surgery.

^cTop 5 systemic treatments were listed.

^d“Combination therapy” was defined as patients receiving ≥2 drug classes during the follow-up.

^eNeoadjuvant treatment was defined as any evidence of chemotherapy (platinum-based or non-platinum-based) or radiation therapy ≤6 months before evidence of surgery.

^fAdjuvant treatment was defined as any evidence of chemotherapy (platinum-based or non-platinum-based), osimertinib or radiation therapy ≤6 months after evidence of surgery.

^gTop 5 therapy regimen were defined based on the sum of patients across all cohorts on each regimen, thus more than five regimens are displayed due to regimens with same number of total patients.

Treatment patterns

The median (Q1–Q3) duration of follow-up was 11.5 (6.0–24.1) months for exon19del, 9.7 (5.6–20.1) months for L858R, and 10.9 (3.3–16.9) months for exon20i cohorts (Table 3). The exon20i cohort had the highest proportion of patients with ≥3 LOTs observed during follow-up (23.1%) relative to exon19del (13.2%) and L858R (12.0%). The most commonly observed drug class during follow-up in the exon19del and L858R cohorts was EGFR TKI (84.9% and 80.0%, respectively). Osimertinib was the most commonly used EGFR TKI for exon19del cohort (61.3%) and L858R cohort (64.0%), followed by erlotinib (exon19del: 29.2%; L858R: 25.3%) and afatinib (exon19del: 14.2%; L858R: 9.3%). The only EGFR TKI used in the exon20i cohort was osimertinib. In the exon20i cohort, non-platinum-based chemotherapy (69.2%) and platinum-based chemotherapy (61.5%) were the most commonly observed classes, while EGFR TKI was used by 30.8% of patients. Among patients with ≥1LOT (94.3% of exon19del, 88.0% of the L858R, and 84.6% of the exon20i), the majority of patients had EGFR TKI as 1LOT in the exon19del and L858R (75.0% and 72.7%, respectively) cohorts while combination therapy of non-platinum and platinum-based chemotherapy was the most frequently observed (36.4%) 1LOT for the exon20i cohort. Similar patterns were observed for 2LOT (Table 3). The median duration of 1LOT and 2LOT was shorter in the exon20i cohort (91.0 days and 167.0 days, respectively) relative to exon19del (146.0 days and 192.0 days, respectively) and L858R (115.0 days and 192.5 days, respectively) cohorts (Figure 3).

Figure 3. Treatment duration (days) for 1LOT and 2LOT.

All-cause HRU and healthcare costs

Overall, HRU was higher in the exon20i cohort. The proportion of patients with all-cause hospitalization during follow-up was higher in the exon20i cohort (76.9%) relative to exon19del (55.7%) and L858R (62.7%) cohorts (Figure 4). Approximately half of patients in exon20i (53.8%) and exon19del (47.2%) cohort and 60.0% of the L858R cohort had ≥1 all-cause ED visit during follow-up, and the proportion of patients with ≥1 office visit associated with antineoplastic treatment administration was higher in exon20i cohort (53.8%) relative to exon19del (30.2%) and L858R (21.3%). Lastly, evidence of hospice care or end of life care was also higher in the exon20i (30.8%) and L858R (29.3%) cohorts than the exon19del cohort (22.6%). The frequency of all-cause HRU per person was also relatively higher in the exon20i cohort across different types of HRU categories (Supplementary S3).

Figure 4. All-cause healthcare resource utilization during follow-up. Hospice care was defined using place of service codes and does not include home visits with a hospice nurse; end-of-life care was identified using an exploratory claims-based proxy, with evidence of any of the following during the last month in which medical and pharmacy claims were available prior to disenrollment: cerebral death, cardiac event including resuscitation, cardiac arrest/failure, respiratory arrest, hepatic coma, acute myocardial infarction, intensive care unit, critical care, do not resuscitate status, palliative care or death.

The exon20i cohort had the highest median (Q1–Q3) total all-cause healthcare costs PPPM of $27,069 ($16,206–$39,662), followed by $17,763 ($11,537–$25,748) in L858R and $17,482 ($9557–$27,521) in exon19del cohorts (Figure 5). The mean (SD) total healthcare costs and breakdown of cost components are displayed in Figure 6 and Supplementary S4. In the exon20i cohort, 74.4% of the total mean healthcare costs were medical services, including antineoplastic treatment administration (24.2%), hospitalization (21.1%) and other outpatient services (25.9%). In the exon19del and L858R cohorts, the main drivers of total healthcare costs were pharmacy costs (43.5% and 43.7%, respectively), followed by hospitalizations (26.7% and 28.4%).

Figure 5. Median total all-cause healthcare costs per patient per month during follow-up.

Figure 6. Mean total all-cause healthcare costs per patient per month during follow-up.

Discussion

To our knowledge, this is one of the first claims-based real-world studies to describe patient characteristics, treatment patterns, and economic burden among a/mNSCLC patients by specific EGFR mutation types, which are known to have important implications on treatment decisions and disease outcomes⁹. Although a/mNSCLC typically affects older patients (mean age at NSCLC diagnosis 70 years²⁵), the mean age of our study cohorts was around 60 years old; reflective of the PharMetrics Plus study database which includes a younger, commercially insured US population. The majority of patients in our study were female (across all cohorts), with the highest proportion observed in the exon20i cohort (76.9%). Our findings align with previous studies by Arcila et al. which reported median age of 66 years and 66.7% females, and by Oxnard et al. which reported median age of 60 years and 70.4% female among NSCLC patients with exon20i^26,27. Similarly, a systematic literature review found that pooled prevalence of EGFR mutation was more common among female (43.7%) compared to male patients (24.0%)⁵. In our study, the exon20i cohort had a higher comorbidity burden relative to the other cohorts with common EGFR mutations, with higher prevalence of certain comorbidities including fatigue/asthenia and obesity. Across all cohorts in our study, the most commonly observed location of metastasis during baseline was bone and bone marrow. Multiple previous retrospective database studies have also shown bone and brain as the common location of metastasis among NSCLC patients at baseline^28–30.

Our study found that the majority of patients in all cohorts had EGFR test results within one month after the first evidence of a/m NSCLC and before the initiation of first systemic therapy post-index. This finding is aligned with the 2023 NCCN treatment guideline where EGFR test result plays an important role in treatment decision making⁹. However, despite all patients in the exon20i cohort having their EGFR mutation test results before initiating therapy, nearly one-third of patients continued to be treated with EGFR TKIs not indicated for exon20i mutation during the study period. This treatment decision underscores the lack of other targeted therapy options for patients with exon20i mutations where chemotherapies were the most commonly observed treatment over the follow-up, which could have contributed to higher total healthcare costs from potential adverse events. Amivantamab (a bispecific EGFR-directed and MET receptor-directed antibody) and mobocertinib (an irreversible TKI), have been approved by FDA for treatment of a/mNSCLC with EGFR exon20i mutation in 2021 and therefore, were not included within the scope of current study³¹. However, mobocertinib has been withdrawn from the market following failure to meet primary endpoints in the Phase 3 EXCLAIM-2 study in 2023³².

In our study, the majority of patients in exon19del and L858R cohorts had EGFR TKI as 1LOT (75.0% and 72.7%, respectively) and as 2LOT (65.0% and 50.0%), which is consistent with a previous retrospective real-world study that found 82.2% and 61.1% of EGFR positive advanced NSCLC patients had EGFR TKI utilization in their 1LOT or 2LOT, respectively³³. On the other hand, non-platinum and platinum-based chemotherapy were the most commonly observed treatment classes for the exon20i cohort and TKIs were only observed in 18.2% of patients’ 1LOT and 20.0% of the 2LOT. This aligns with the NCCN treatment guidelines where systemic therapy serves as the 1LOT for exon20i mutation before subsequent targeted therapy, given the general lack of response to EGFR TKI therapy⁹. This finding also resonates with a previous systematic literature review assessing clinical burden of exon20i mutation, which found a trend of greater overall survival and response for chemotherapy compared with TKIs as 1LOT¹⁴. Furthermore, the shorter treatment duration observed for 1LOT and 2LOT in the exon20i cohort relative to the exon19del and L858R cohorts may indicate the lack of effective targeted treatment during the study period and worse prognosis among patients with exon20i mutation compared to the common EGFR mutations, which has been suggested in previous studies^34,35. Exon19del and L858R cohorts had similar use of EGFR TKIs, which aligns with a prior study assessing treatment patterns by EGFR mutation types; the higher prevalence of osimertinib use observed in our study (61.3–64.0%) compared to the prior study (24.8% and 35.2% in 2LOT and 3LOT, respectively) may be due to our study including more recent years of data as osimertinib was approved as a 1LOT for mNSCLC with EGFR exon19del or L858R mutation in 2018^35,36. Osimertinib was the only EGFR TKI used in the exon20i cohort in our study, which is different from the previously published studies where other EGFR TKIs (afatinib or erlotinib) were observed; this may be due to differences in study time period and the small sample size of exon20i cohort in our study (n = 13)^34,35.

To that end, in our study, the exon20i cohort had the highest proportion of patients with incident metastasis during follow-up, particularly in respiratory system and CNS. The higher proportion of patients with multiple incident metastatic sites observed during follow-up in this cohort may be one of the contributing factors of poorer prognosis among patients with exon20i mutation compared to the common EGFR mutations³⁷. Along these lines, we found that the exon20i cohort had higher prevalence of hospitalizations and evidence of hospice or end of life care relative to the exon19del and L858R cohorts; total all-cause healthcare costs PPPM were also higher in the exon20i cohort relative to exon19del and L858R cohorts. Within the common EGFR mutation cohorts in our study, HRU for the L858R cohort was higher relative to that observed for the exon19del cohort, including hospitalization and hospice/end of life care, which aligns with previous studies reporting worse clinical outcomes for patients with L858R mutation compared to those with exon19del mutation^5,38.

While prior studies have demonstrated HRU and costs among mNSCLC patients with EGFR mutations³⁹, our study is one of the first claims-based studies to describe economic burden by different mutation types. A recent retrospective claims database study reported mean total all-cause healthcare costs of $25,499 PPPM among mNSCLC patients with EGFR mutation, of which pharmacy costs account for nearly half of the total costs ($11,752)¹⁸. Our study found similar results for exon19del and L858R cohorts, where pharmacy costs were the main driver of total healthcare costs, which may be associated with use of EGFR TKIs in over 80% of these cohorts. In contrast, the main cost driver for the exon20i cohort was medical service costs, particularly costs associated with antineoplastic treatment administration and hospitalization, which reflect the high proportion of exon20i cohort receiving chemotherapy along with the high prevalence of hospitalization observed.

The strengths of our study include use of fully adjudicated medical and pharmacy claims from PharMetrics Plus linked to NeoGenomics data. The PharMetrics Plus database provides comprehensive capture of HRU and healthcare costs across all settings of care. Additionally, while most previously published studies that used administrative claims databases in the NSCLC population relied on a combination of lung cancer diagnoses and initiation of NSCLC-related treatments to proxy patients with EGFR mutation^21,40, the NeoGenomics data provides confirmed EGFR mutation status and mutation types; a novelty of the study at-hand.

Our study has limitations inherent to a retrospective study using claims data, including potential misclassification of diagnosis records, data entry error and lack of social determinants of health or clinical data to confirm disease stage. However, evidence of cancer treatment and diagnosis for metastatic disease were used as a proxy for advanced or metastatic cancer stage. Future studies including social determinants of health, such as race/ethnicity and education level, may provide more insight into potential disparities in healthcare costs and HRU among this patient population. Similarly, data on mortality was not available in the claims database. To address this limitation, proportion of patients with evidence of hospice or end-of-life care were reported, although this is likely to be underreported. While costs were reported PPPM, the interpretation of the HRU over the follow-up period should also consider different duration of follow-up between cohorts. In addition, the study population is primarily commercially insured and therefore the findings may not be generalizable to patients with a/mNSCLC who are uninsured or are covered by Medicaid or traditional Medicare. Finally, sample sizes for the study cohorts were small, and findings should be reassessed as more patients with longer follow-up become available. Future studies with larger sample size, including recently approved targeted therapy for exon20i (i.e. amivantamab), are warranted to confirm our findings and understand the impact of amivantamab on HRU and costs.

Conclusions

This is one of the first real-world studies to describe treatment patterns, HRU and healthcare costs among a/mNSCLC patients with specific EGFR mutation types. Patients with EGFR exon20i mutations had higher HRU, healthcare costs and evidence of disease progression relative to the more common EGFR mutations (i.e. exon19del and L858R), suggesting worse prognosis and higher disease burden. Furthermore, while EGFR TKIs approved for exon19del and L858R mutations are not recommended for the treatment of exon20i mutation, nearly one-third of exon20i patients were treated with EGFR TKIs, highlighting the need for therapy options targeting exon20i and the importance of understanding physician treatment decisions in this group of patients.

Transparency

Declaration of funding

This study was supported by Janssen Scientific Affairs, LLC (Horsham, PA, USA).

Declaration of financial/other relationships

PV, DW and JV are employees of Janssen Scientific Affairs10.13039/100017183, LLC. IL, JC, DT and AN are employees of IQVIA, which received funding for this study from Janssen Scientific Affairs10.13039/100017183, LLC. DL is employee of NeoGenomics, which received part of the research funding Janssen provided to IQVIA. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

All named authors meet the International Committee of Medical Journal Editors (ICMJE) criteria for authorship for this article and take responsibility for the integrity of the work as a whole. IL and JC drafted the manuscript, and all authors critically revised the manuscript. All authors read and approved the final manuscript.

Previous presentations

Part of this research was presented at the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) 2023.

Acknowledgements

No assistance in the preparation of this article is to be declared.

Data availability statement

The original de-identified data used in this analysis were obtained from and are the property of IQVIA and NeoGenomics. IQVIA and NeoGenomics have restrictions prohibiting the authors from making the data set publicly available. Interested researchers may contact IQVIA/NeoGenomics to apply to gain access to the study’s data in the same way the authors obtained the data (see https://www.iqvia.com/contact/sf or https://neogenomics.com/contact-us).