Treatment patterns, comorbidities, healthcare resource use, and associated costs by line of chemotherapy and level of comorbidity in patients with newly-diagnosed Merkel cell carcinoma in the United States

Abstract

Aims: To examine the characteristics of patients with newly-diagnosed Merkel cell carcinoma (MCC), analyze their treatment patterns and comorbidities after diagnosis, and evaluate the economic burden on the MCC patient population in the US.

Materials and methods: This observational, non-interventional cohort study identified patients with MCC that were newly-diagnosed between January 1, 2010 through December 31, 2014, and whose data were either in the MarketScan Commercial Claims and Encounters (CCAE) or Medicare Supplemental and Coordination of Benefits databases. Standard descriptive statistics were used to describe patient demographics, clinical characteristics, treatment regimens, and healthcare resource use (HRU) and cost.

Results: Following MCC diagnosis, most patients in the study population (n = 2,177) received only surgery (34.5%) or surgery and radiotherapy without chemotherapy (22.0%), while 14.5% of patients received none of these treatments; 27.5% of patients received at least one line of chemotherapy as part of their treatment. Mean total healthcare costs per patient per year (PPPY), as well as mean inpatient, outpatient, and pharmacy costs, were significantly greater for patients who received chemotherapy compared with those who received other or no treatments. Higher HRU and mean costs were associated with increasing patient comorbidity burden, ranging from $62,401 PPPY in Deyo Charlson Comorbidity Index level 1 to $109,690 in level ≥3.

Limitations: The study used claims databases that were limited to patients who are covered by large employer-sponsored insurance and/or Medicare and did not provide information regarding the rationale for treatment choice or resource use.

Conclusions: The choice of treatment is a major factor in determining healthcare costs associated with MCC, with the highest costs in patients receiving chemotherapy. Patients with MCC often exhibit comorbidities, and both HRU and healthcare costs increase significantly with each comorbidity level.

Keywords:

• Merkel cell carcinoma
• MarketScan
• claims database
• comorbidities
• healthcare resource use
• line of treatment
• chemotherapy

Introduction

Merkel cell carcinoma (MCC) is a rare, neuroendocrine, cutaneous malignancy that occurs more frequently in elderly individuals and exhibits aggressive clinical features^1–4. Due to its rapid growth pattern and high risk of early metastasis, patients with MCC have comparatively poorer prognosis and diminished survival compared with patients with other aggressive skin malignancies, such as melanoma^5–10. The 5-year survival rate in patients with local disease is 64%, yet only 16% in patients with metastatic disease^6,11. The incidence of MCC has been linked to infection with Merkel cell polyomavirus (MCV), as ∼80% of MCC tumors are positive for MCV integration^12,13. However, in MCV-negative tumors, development of MCV may be associated with UV radiation exposure¹³. Chronic UV radiation may also be a contributing factor to the increased incidence of MCC in the elderly¹⁴. MCC occurs more frequently in immunosuppressed patients, such as those with organ transplants, multiple malignancies, or who are diagnosed with HIV/AIDS¹³. Recent findings indicate that immunosuppressed patients have a higher likelihood of mortality, while tumors with unknown primary sites have a lower risk for distant metastasis¹⁵. Primary MCC is often asymptomatic and painless¹⁶, making tumor diagnosis difficult and dependent on incisional tumor biopsy¹⁷.

In the US, according to Surveillance, Epidemiology, and End Results (SEER) tumor registry data, the incidence of MCC increased in recent years, from 0.5 cases per 100,000 person-years in 2000 to 0.7 cases per 100,000 person-years in 2013^18,19. MCC incidence increased exponentially with age, from 0.1 to 1.0 to 9.8 per 100,000 person-years among patients aged 40–44, 60–64, and ≥85 years, respectively. In this same study, men had a higher incidence of MCC compared with women across all age groups, and this effect was most pronounced among the oldest age groups¹⁸. MCC-related mortality in the US increased from 0.03 cases per 100,000 person-years in 1986 to 0.43 cases per 100,000 person-years in 2011⁵.

The choice of treatment for MCC depends on the stage of the disease, the location of the tumor, and comorbidities^20–22. Historically, prior to the advent of immune-oncology (IO), surgery had been the main treatment for localized MCC, with radiotherapy used to minimize local reoccurrence or when complete excision of the lesion was not feasible¹³. Surgery combined with adjuvant radiotherapy was shown to improve both localized relapse-free and distant metastasis-free survival²³. In patients with metastatic MCC (mMCC), chemotherapy and radiotherapy were considered primary treatment options²². Adjuvant radiotherapy has been recommended for patients with lymph-node positive MCC²². Commonly used chemotherapy regimens include cisplatin ± etoposide phosphate, carboplatin ± etoposide, and topotecan²². In a survey of 218 patients with MCC, 72.5% of patients underwent surgery as a part of their treatment regimen, while 13.4% of patients received chemotherapy¹⁵. While MCC is a chemosensitive tumor, patients with metastatic disease seldom have durable responses to chemotherapy, with studies reporting median progression-free survival ranging from 3.0–5.5 months, and duration of response at ≤8 months in both first- and second-line settings^24–28. Moreover, retrospective real-world studies indicate a lack of association between chemotherapy and MCC relapse or an improvement in survival^15,26. With the emergence of IO, immune checkpoint inhibitors have reported promising results in the form of durable disease responses not observed historically with chemotherapy²⁹. Current National Comprehensive Cancer Network (NCCN) guidelines reflect the recommendation to treat with immune checkpoint inhibitors if a clinical trial is unavailable²². There remains an unmet medical need for patients whose metastatic disease does not respond to IO, and this need necessitates ongoing clinical research for new agents or combinations²⁹.

While studies have outlined real-world treatment regimens and outcomes in patients with MCC, comprehensive analyses of the economic burden associated with MCC are sparse. Few studies have examined the healthcare resource use (HRU) and costs broken down by treatment type for patients with MCC, while economic analyses based on line of chemotherapy have not been performed. Furthermore, HRU and costs according to level of patient comorbidity are unknown. We aimed to characterize real-world longitudinal treatment patterns, identify comorbidities, and estimate HRU and healthcare costs for patients with newly-diagnosed MCC in the US using patient claims data collected over a 5-year span. Our study helps clinicians better understand how differences in treatment regimen and comorbidities might factor into treatment costs in patients with MCC.

Methods

Data source

This study is based on data from IBM Watson Health’s Truven MarketScan Commercial Claims and Encounters (CCAE) and the Medicare Supplemental and Coordination of Benefits datasets of patients whose MCC was diagnosed between January 1, 2010 through December 31, 2014³⁰. The CCAE is a medical and drug insurance claims database with ∼174 million unique de-identified patients (since 1995)—including active employees, early retirees, Consolidated Omnibus Budget Reconciliation Act (COBRA) continuers, and their dependents—who are insured by employer-sponsored plans. The database was established in 1988 and contains inpatient admission records, outpatient services, prescription drugs, populations, eligibility status, and costs of services. The Medicare Supplemental database contains the same data elements and profiles the healthcare experience of retirees with Medicare supplemental insurance paid for by employers. All costs were inflated to 2017 USD.

Study design

An observational, non-interventional cohort study was performed on secondary use of MarketScan claims for patients with newly-diagnosed MCC. Diagnosis information was collected from the Outpatient Services and the Inpatient Services databases using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes 209.31–209.36 and 209.75 for MCC. Patients with MCC were stratified according to four sub-groups of interest: age group, divided into patients <18 years old (child/adolescent cohort), 18–64 years old (adult cohort), and >64 years old (elderly cohort); treatment type; chemotherapy treatment line group, as defined by an algorithm described later in this article that was created to define chemotherapy treatment cycles and treatment lines; and comorbidity category.

Patient selection

The study population was identified according to the following inclusion criteria: patients with MCC newly-diagnosed between July 1, 2010 and December 31, 2014 (the date of first diagnosis of MCC was deemed the index date) and continuous enrollment for ≥6 months before the index diagnosis date (pre index period), without any diagnosis of MCC (incident cases). In addition to inclusion criteria, patients selected for follow-up-period analysis were required to have a follow-up period ≥30 days after the index date and no enrollment gaps >30 days after the index date (Figure 1). Patient claims data have been shown to be highly accurate in identifying the incidence and diagnosis dates of malignancies, and a 6-month cancer-free period prior to diagnosis was determined to be the minimum required to avoid possible misclassification of an incident case as a prevalent case³¹.

Figure 1. Patient selection and study design flowchart. MCC, Merkel cell carcinoma.

Study variables

Baseline socio-demographic and clinical characteristics described were sex, age, calendar year of the index date, region of residence, and health insurance plan. In addition, the comorbidity burden was evaluated using the Deyo Charlson Comorbidity Index (Deyo-CCI) score assessed by ICD-9-CM diagnosis codes recorded on the inpatient and outpatient services^32,33, and the most commonly recorded comorbidities based on inpatient and outpatient services were described. Increasing Deyo-CCI scores indicate increasing burden of comorbid conditions.

Two steps were followed to build chemotherapy treatment lines (Figure 2). First, treatment cycles were identified: for the first line (1L), the start of a cycle was defined as the first prescription or administration date of a chemotherapy. Chemotherapy treatments were identified with relevant antineoplastic agent-associated inpatient and outpatient procedures (Healthcare Common Procedure Coding System [HCPCS] codes) and chemotherapy agent prescriptions collected from outpatient pharmaceutical claims data using the FDA National Drug Code dictionary. For the following treatment lines, the cycle start date was set as the date of the first prescription of MCC treatment that follows the previous line’s end. The cycle length was set at 28 days. Second, treatment lines were identified based on treatment cycles defined as above. The maximum treatment-free period (period without chemotherapy) was set to 30 days. Consecutive cycles with the same combination of treatments were merged into one line if the duration between the end of the first cycle and the start of the second cycle did not exceed 30 days. In case of a gap exceeding 30 days and a change in the treatment defined as a switch or an add-on, the second cycle was identified as a new line (Figure 2).

Figure 2. Chemotherapy treatment sequences algorithm.

MCC management therapies of interest were surgery, radiotherapy, and chemotherapy after the index date. These treatments were described after the initial diagnosis of MCC and, therefore, were likely administered for MCC. Surgeries were identified from the inpatient and outpatient services using Current Procedural Terminology, 4^th Edition (CPT-4) and HCPCS classification systems. Radiotherapy was identified from the inpatient and outpatient services using the service sub-category code defined by Truven Health³⁰. The HRU and associated costs were assessed for inpatient, outpatient, and drug utilization. The considered inpatient utilization data included the number of admissions and the length of stay. Outpatient utilization included emergency department and office visits, radiology diagnostics, laboratory services, and all other outpatient services. Drug utilization data included prescriptions from the outpatient pharmaceutical claims data, outpatient office visits with administered antineoplastic agents (IV), and prescriptions of outpatient pharmacy antineoplastic agents.

Statistical analysis

Patient demographic and clinical characteristics, as well as study outcomes, were described using standard descriptive statistics. Number, mean, median, standard deviation, minimum, and maximum were provided for continuous variables, while frequency tables (with number and percentages) were presented for categorical variables. Descriptive statistics were performed for the overall population as well as for sub-groups of interest. Two-sided confidence intervals with a confidence probability of 95% were reported when necessary.

Descriptive analyses and Kaplan-Meier survival analyses were used to estimate the median duration of a treatment line. A log-rank test was performed to compare the Kaplan-Meier curves of time to chemotherapy by age group. HRU and costs were calculated per patient per year (PPPY). All costs were adjusted to 2017 US dollars. Statistical tests were performed to compare resource use and costs by Deyo-CCI severity level and by type of treatment received. Continuous characteristics were compared between groups using Student or Wilcoxon tests. Categorical characteristics were compared using χ² or Fisher tests. Analyses were performed using SAS Version 9.4 (SAS Institute, Cary, NC) and SAS/STAT Version 12.1 (SAS Institute).

Results

Patient selection

A total of 2,355 patients with MCC met all the inclusion criteria and constituted the overall MCC cohort, and 2,177 patients with MCC met the additional criteria to constitute the analysis population (Figure 2 and Table 1). The majority of selected patients (61.6%) were elderly patients, aged ≥65 years. Few patients aged <18 years were identified (n = 12). The mean age at first MCC diagnosis was 68.8 years (SD = 14.9), and patients were predominantly male (60.3%). The highest percentage of patients was from the southern region of the US (33.0%), and the majority of patients had a preferred provider organization (PPO) insurance plan (51.0%). The mean duration of follow-up was 522.9 days (SD = 428.2). The number of incident MCC cases decreased from 598 patients in 2011 (25.4% of the overall MCC cohort) to 422 patients in 2014 (17.9% of the overall MCC cohort). MCC was localized in the face, scalp and neck, upper limb, lower limb, trunk, and on other, undetermined sites for 25.3%, 10.5%, 14.3%, 7.9%, 5.8%, and 31.0% of patients, respectively. Secondary MCC was observed in 6.3% of patients (Table 1).

Table 1. Baseline socio-demographic and clinical characteristics of patients with newly-diagnosed Merkel cell carcinoma.

	Age <18 years (n = 12)	Age 18–64 years (n = 922)	Age ≥65 years (n = 1,412)	All patients (n = 2,355)
Sociodemographic characteristics
Age, mean (SD)	11.8 (4.4)	54.6 (8.7)	78.6 (7.9)	68.8 (14.9)
Female (n, %)	8 (66.7)	424 (46.0)	503 (35.4)	935 (39.7)
US region of residence (n, %)
Northeast region	1 (8.3)	190 (20.6)	301 (21.2)	492 (20.9)
North Central region	2 (16.7)	189 (20.5)	371 (26.1)	562 (23.9)
South region	8 (66.7)	334 (36.2)	435 (30.6)	777 (33.0)
West region	0	187 (20.3)	277 (19.5)	464 (19.7)
Unknown region	1 (8.3)	22 (2.4)	37 (2.6)	60 (2.6)
Insurance plan (n, %)
Comprehensive	0	38 (4.1)	560 (39.4)	598 (25.4)
EPO	0	14 (1.5)	16 (1.1)	30 (1.3)
HMO	3 (25.0)	92 (10.0)	109 (7.7)	204 (8.7)
POS (with or without capitation)	2 (16.7)	66 (7.2)	49 (3.5)	117 (5.0)
PPO	4 (33.3)	554 (60.1)	643 (45.3)	1,201 (51.0)
CDHP	2 (16.7)	44 (4.8)	4 (0.3)	50 (2.1)
HDHP	1 (8.3)	33 (3.6)	1 (0.1)	35 (1.5)
Missing/unknown	0	81 (8.8)	39 (2.7)	120 (5.1)
Health insurance: Medicare, n (%)	0	9 (1.0)	1,421 (100.0)	1,430 (60.7)
Duration of health plan eligibility  prior to index date, mean  days (SD)	845.0 (637.8)	749.6 (443.9)	780.9 (449.9)	768.9 (448.7)
Duration of follow-up from index  diagnosis, mean days (SD)	362.9 (384.7)	508.1 (421.0)	533.8 (432.9)	522.9 (428.2)
Median (Q1; Q3)	188.5 (151.0; 436.5)	391.0 (151.0; 792.0)	425.0 (170.0; 812.0)	416.0 (162.0; 804.0)
Deyo-CCI score, mean (SD)^a	0.58 (1.00)	1.43 (1.90)	2.61 (2.35)	2.14 (2.26)
Deyo-CCI category, n (%)^a
0	8 (66.7)	384 (41.6)	262 (18.4)	654 (27.8)
1	2 (16.7)	220 (23.9)	323 (22.7)	545 (23.1)
2	1 (8.3)	129 (14.0)	235 (16.5)	365 (15.5)
≥3	1 (8.3)	189 (20.5)	601 (42.3)	791 (33.6)
Tumor site at diagnosis by ICD-9-CM code, n (%)^b
Merkel cell carcinoma of the face (209.31)	2 (16.7)	190 (20.6)	403 (28.4)	595 (25.3)
Merkel cell carcinoma of the scalp and neck (209.32)	2 (16.7)	97 (10.5)	149 (10.5)	248 (10.5)
Merkel cell carcinoma of the upper limb (209.33)	0	132 (14.3)	205 (14.4)	337 (14.3)
Merkel cell carcinoma of the lower limb (209.34)	2 (16.7)	63 (6.8)	122 (8.6)	187 (7.9)
Merkel cell carcinoma of the trunk (209.35)	0	49 (5.3)	88 (6.2)	137 (5.8)
Merkel cell carcinoma of other sites (209.36)	6 (50.0)	329 (35.7)	395 (27.8)	730 (31.0)
Secondary Merkel cell tumor (209.75)	0	74 (8.0)	75 (5.3)	149 (6.3)

If not specified, all variables are presented as of the date of first diagnosis of MCC (index date).

^a Comorbidities that occur during the pre and post index periods (i.e. entire observation period) are included in the counts.

^b Patients may have multiple ICD-9-CM codes for MCC—for example, if MCC affects more than one anatomical site on the body during the course of the patient trajectory.

Abbreviations. CDHP, consumer-driven health plan; Deyo-CCI, Deyo Charlson Comorbidity Index; EPO, exclusive provider organization; HDHP, high deductible health plan; HMO, health maintenance organization; POS, point of service; PPO, preferred provider organization; Q1, first quartile; Q3, third quartile; SD, standard deviation.

Patient age and associated HRU and costs

Total mean costs were estimated at $86,227 PPPY (median = $38,977 PPY), with higher mean costs observed in the adult cohort (mean = $99,841; median = $52,560) than the elderly cohort (mean = $77,663; median = $34,708), driven mainly by higher costs of diagnostic radiology and other outpatient care observed in the adult cohort (mean diagnostic radiology costs = $8,282, mean other outpatient costs = $36,040) vs the elderly cohort (mean diagnostic radiology costs = $4,985; mean other outpatient costs = $26,803). Although only 12 patients under 18 years of age were evaluated, these patients were associated with the highest mean costs ($106,660); however, this figure was skewed by outliers because the median cost was only $8,006. Inpatient costs (mean = $86,866) were the main contributor to this high cost because total outpatient and pharmacy costs were lower in the child/adolescent cohort than in the other two cohorts. Inpatient, ER, office visit, radiotherapy, laboratory service, and pharmacy costs were all higher in the adult cohort than in the elderly cohort, while outpatient facility visit costs were the sole cost that was higher in the elderly cohort than in the adult cohort.

Treatment patterns in patients with MCC and associated HRU and costs

Treatments received following MCC diagnosis, including surgery, radiotherapy, and chemotherapy, rely on the sequence of treatment and the description of treatment lines among patients receiving specific chemotherapy treatment (Figure 3). The four most common treatment regimens were surgery only (34.5%), surgery and radiotherapy without chemotherapy (22.0%), surgery and radiotherapy with chemotherapy (14.7%), and no treatment (14.5%). Chemotherapy was administered following MCC diagnosis in 27.5% of the analysis population; most of these patients (53.4%) received chemotherapy in combination with surgery and radiotherapy. Additionally, similar chemotherapy regimens were used across both adult and elderly cohorts (most frequently carboplatin + etoposide; see below), indicating that limited options exist for patients with MCC.

Figure 3. Type of treatments received following new diagnosis of Merkel cell carcinoma, sorted by age.

Upon examination of resource use and costs estimated by type of treatment received following MCC diagnosis (Table 2), the highest total costs PPPY were observed in the four patient groups receiving chemotherapy: those receiving chemotherapy only (mean = $154,506; median = $101,487), those receiving chemotherapy and radiotherapy without surgery (mean = $150,157; median = $129,041), those receiving chemotherapy and surgery without radiotherapy (mean = $137,896; median = $63,063), and those receiving chemotherapy, surgery, and radiotherapy (mean = $134,730; median = $86,094). Among patients receiving treatment, costs were significantly lower in patients not receiving chemotherapy, with the lowest total healthcare costs PPPY being estimated for patients with surgery only (mean = $56,124; median = $22,999). A total of 316 patients (14.5%) did not receive any specified treatment. The total healthcare costs for those patients was estimated at a mean of $28,442 PPPY and at a median of $8,998 PPPY. Patients receiving chemotherapy, regardless of the presence of other combination therapies, had significantly higher mean costs PPPY in inpatient care, outpatient care, ER visits, office visits, laboratory services, and pharmacy usage than patients who were not treated with chemotherapy. Mean chemotherapy costs were estimated at $3,054 PPPY, or 3.5% of the total mean costs PPPY.

Table 2. All-cause healthcare utilization and healthcare costs PPPY, by type of treatment received following new diagnosis of Merkel cell carcinoma.

	No treatment (n = 316)	Surgery only (n = 751)	Radiotherapy only (n = 33)	Chemo only (n = 79)	Chemo + radiotherapyonly (n = 33)	Chemo + radiotherapy + surgery (n = 320)	Surgery + chemoonly (n = 167)	Surgery + radiotherapy only(n = 478)	p-value
Inpatient
Patients with inpatient admission, n (%)	46 (14.6%)	300 (40.0%)	18 (54.6%)	32 (40.5%)	17 (51.5%)	223 (69.7%)	99 (59.3%)	220 (46.0%)	<.0001^a
Inpatient costs (all), US $, mean (SD)	8,227 (39,793)	21,899 (85,833)	23,782 (54,030)	58,603 (208,632)	22,317 (39,393)	32,214 (81,778)	59,267 (242,537)	29,363 (112,449)	<.0001^b
95% CI	3,823–12,632	15,750–28,048	4,623–42,940	11,872–105,334	8,349–36,285	23,220–41,208	22,212–96,322	19,257–39,469
Median (IQR)	0 (0–0)	0 (0–9,515)	6,167 (0–26,855)	0 (0–31,742)	3,291 (0–19,844)	9,689 (0–29,883)	5,262 (0–41,449)	0 (0–15,586)
Outpatient
Patients with ER visit, n (%)	69 (21.8%)	301 (40.1%)	15 (45.5%)	35 (44.3%)	14 (42.4%)	197 (61.6%)	86 (51.5%)	205 (42.9%)	<.0001^a
ER costs (all), US $, mean (SD)	467 (1,540)	953 (3,400)	1,185 (2,510)	2,427 (5424)	2,312 (4,742)	1,870 (4,044)	2,660 (11,521)	1,122 (3,986)	<.0001^b
95% CI	297–637	710–1,197	295–2,075	1,212–3,642	631–3,994	1,425–2,314	899–4,420	764–1,480
Median (IQR)	0 (0–0)	0 (0–514)	0 (0–1,213)	0 (0–2,442)	0 (0–1,565)	239 (0–1,644)	55 (0–1,820)	0 (0–755)
Office visits
Patients with office visit, n (%)	291 (92.1%)	730 (97.2%)	30 (90.9%)	77 (97.5%)	33 (100.0%)	317 (99.1%)	162 (97.0%)	472 (98.7%)	<.0001^a
Office costs (all), US $, mean (SD)	1,205 (1,665)	1,645 (2,542)	1,670 (1,955)	2,372 (1,836)	2,249 (1,219)	2,749 (1,846)	2,797 (3,255)	2,245 (1,678)	<.0001^b
95% CI	1,021–1,390	1,462–1,827	977–2,363	1,961–2,783	1,817–2,681	2,546–2,952)	2,300–3,294	2,094–2,396
Median (IQR)	837 (437–1,479)	1,094 (619–1,988)	950 (400–1,958)	2,094 (992–3,025)	2,154 (1,384–2,523)	2,421 (1,569–3,509)	1,984 (1,080–3,491)	1,869 (1,196–2,750)
Facility visits
Patients with facility visit, n (%)	43 (13.6%)	221 (29.4%)	14 (42.4%)	22 (27.9%)	17 (51.5%)	142 (44.4%)	67 (40.1%)	136 (28.5%)	<.0001^a
Facility costs (all), US $, mean (SD)	329 (1,515)	612 (3,433)	1,170 (3,183)	1,257 (3,551)	1,259 (3,808)	619 (1,993)	620 (1,595)	361 (1,589)	.02^b
95% CI	162–497	366–858	42–2,299)	462–2,053	–91–2,609	400–839	377–864	218–504
Median (IQR)	0 (0–0)	0 (0–97)	0 (0–781)	0 (0–266)	55 (0–618)	0 (0–326)	0 (0–433)	0 (0–70)
Radiotherapy
Patients with radiotherapy, n (%)	0	0	32 (97.0%)	0	32 (97.0%)	319 (99.7%)	0	477 (99.8%)	<.0001^a
Radiotherapy costs (all), US $, mean (SD)	0 (0)	0 (0)	29,857 (31,890)	0 (0)	27,971 (35,388)	27,441 (49,251)	0 (0)	31,541 (62,863)	<.0001^c
95% CI	0–0	0–0	18,549–41,164	0–0	15,422–40,519	22,024–32,857	0–0	25,891–37,191
Median (IQR)	0 (0–0)	0 (0–0)	19,279 (6,406–45,103)	0 (0–0)	14,511 (8,180–33,128)	13,997 (5,684–28,176)	0 (0–0)	12,388 (5,394–34,292)
Diagnostic radiology
Patients with diagnostic radiology, n (%)	240 (76.0%)	671 (89.4%)	30 (90.9%)	70 (88.6%)	32 (97.0%)	319 (99.7%)	158 (94.6%)	463 (96.9%)	<.0001^a
Diagnostic radiology costs (all), US $, mean (SD)	3,027 (7,647)	4,101 (10,165)	11,396 (14,374)	8,401 (12,091)	8,572 (10,328)	10,618 (15,808)	6,986 (10,522)	7,533 (11,482)	<.0001^b
95% CI	2,180–3,873	3,373–4,829	6,299–16,493	5,693–11,110	4,910–12,234	8,879–12,356	5,379–8,594	6,501–8,565
Median (IQR)	524 (12–2,527)	870 (185–3,452)	5,733 (721–16,354)	4,433 (1,035–11,462)	4,965 (2,258–10,934)	5,270 (2,445–12,608)	2,528 (576–9,367)	3,332 (1,262–8,694)
Laboratory services
Patients with laboratory service, n (%)	234 (74.1%)	739 (98.4%)	25 (75.8%)	72 (91.1%)	30 (90.9%)	320 (100.0%)	165 (98.8%)	473 (99.0%)	<.0001^a
Laboratory costs (all), US $, mean (SD)	632 (2,112)	1,927 (3,983)	1,188 (2,522)	3,332 (10,457)	2,410 (3,557)	4,122 (5,810)	3,764 (7,360)	2,454 (3,916	<.0001^c
95% CI	399–866	1,642–2,213	294–2,083	990–5,675	1,149–3,671	3,483–4,761	2,639–4,889	2,102–2806
Median (IQR)	106 (0–482)	682 (263–1,739)	143 (0–953)	956 (242–2,737)	1,061 (135–3,345)	2,094 (971–4,659)	1,421 (535–4,047)	1,344 (713–2,833)
Other services
Patients with other service, n (%)	304 (96.2%)	749 (99.7%)	32 (97.0%)	79 (100.0%)	33 (100.0%)	320 (100.00%)	167 (100.00%)	478 (100.00%)	<.0001^a
Other outpatient costs (all), US $, mean (SD)	12,557 (30,358)	22,051 (60,764)	20,219 (32,681)	62,205 (60,170)	73,736 (87,304)	48,195 (72,180)	51,380 (71,642)	27,750 (43,154)	<.0001^b
95% CI	9,197–15,917	17,698–26,404	8,631–31,807	48,727–75,682	42,779–104,692	40,256–56,133	40,434–62,325	23,871–31,628
Median (IQR)	2,379 (571–9,088)	7,435 (2,887–20,251)	6,973 (2,335–24,997)	51,494 (19,307–82,360)	50,356 (13,941–87,370)	30,510 (11,241–56,417)	25,684 (6,877–66,636)	13,238 (6,409–30,675)
Total outpatient
Total outpatient costs (all), US $, mean (SD)	18,218 (34,816)	31,290 (68,099) |	66,685 (53,134)	79,995 (67,165)	118,508 (95,652)	95,613 (112,754)	68,207 (84,466)	73,005 (93,091)	<.0001^c
95% CI	14,364–22,071	26,411–36,168	47,844–85,526	64,951–95,039	84,591–152,425	83,212–108,014	55,302–81,112	64,639–81,372
Median (IQR)	5,797 (2,629–15,920)	13,335 (5,714–32,756)	46,056 (25,551–111,060)	72,815 (27,356–106,792)	94,986 (56,498–138,615)	66,773 (35,213–114,994)	36,691 (14,973–91,584)	40,699 (21,088–86,716)
Pharmacy
Patients with drug prescription, n (%)	208 (65.8%)	530 (70.6%)	24 (72.7%)	64 (81.0%)	27 (81.8%)	259 (80.9%)	154 (92.2%)	335 (70.1%)	<.0001^a
Pharmacy costs, US $, mean (SD)	1,997 (3,710)	2,935 (7,356)	3,364 (4,848)	15,908 (28,438)	9,332 (12,362)	6,903 (12,331)	10,421 (17,094)	2,612 (4007)	<.0001^c
95% CI	1,586–2,407	2,408–3,462	1,644–5,083	9,539–22,278	4,949–13,715	5,547–8,259	7,810–13,033	2,252–2,972
Median (IQR)	234 (0–2,345)	771 (0–3,354)	1,205 (0–4,724)	1,799 (184–15,715)	5,602 (819–13,319)	2,886 (637–6,648)	4,073 (1,115–11,132)	930 (0–3,444)
Total healthcare
Mean (SD)	28,442 (55,206)	56,124 (115,601)	93,830 (85,785)	154,506 (225,280)	150,157 (104,425)	134,730 (167,907)	137,896 (263,549)	104,980 (168,814)	<.0001^c
95% CI	22,332–34,552	47,843–64,405	63,412–124,248	104,046–204,966	113,130–187,185	116,263–153,197	97,631–178,161	89,808–120,152
Median (IQR)	8,998 (3,645–25,429)	22,999 (9,211–55,161)	71,790 (34,864–130,790)	101,487 (48,681–170,944)	129,041 (74,646–198,396)	86,094 (47,315–159,973)	63,063 (27,791–169,665)	55,984 (28,228–112,726)

Abbreviations. Chemo, chemotherapy; ER, emergency department; IQR, interquartile range (Q1–Q3); PPPY, per patient per year.

^a Categorical characteristics were compared using χ² test.

Continuous characteristics were compared between groups using ^bStudent or ^cWilcoxon test as appropriate.

Chemotherapy treatment lines in patients with MCC and associated HRU and costs

Of 2,177 patients with MCC, 599 patients (27.5%) received at least one line of chemotherapy (Table 3). Among these patients, 93 (15.5%) had at least a second line of chemotherapy; of these, 10 patients (10.8%) had a third line of chemotherapy and only one patient received four lines of chemotherapy. The Kaplan-Meier median duration estimate for each line of treatment (LOT) was 98 days (95% CI = 84–105) for first-line (1L) therapy, 84 days (95% CI = 56–119) for second-line (2L) therapy, and 94 days (95% CI = 28–112) for third-line (3L) therapy³⁴. The use of monotherapy or combination chemotherapy is described by line among chemotherapy-treated patients overall and by age group³⁴. In the overall chemotherapy-treated population (n = 599), monotherapy and combination regimens were equally distributed in 1L and 2L chemotherapy. Within 1L, patients in the elderly cohort were more likely to receive monotherapy than combination therapy (54.5% vs 45.5%), while patients in the adult cohort were more likely to receive combination therapy than monotherapy (57.7% vs 42.3%). These differences in proportions were reduced in 2L chemotherapy (n = 93). For 3L chemotherapy (n = 10), most patients in the elderly cohort received combination therapy (87.5%), while patients in the adult cohort were evenly split between those receiving monotherapy and those receiving combination therapy.

Table 3. All-cause healthcare utilization and healthcare costs PPPY, within lines of chemotherapy, by chemotherapy line.

	First line (n = 599)	Second line (n = 93)	Third line (n = 10)
Inpatient
Patients with inpatient admission, n (%)	151 (25.2%)	25 (26.9%)	2 (20.0%)
Inpatient costs (all), US $, mean (SD)	46,353 (333,249)	59,834 (201,845)	8,475 (24,217)
95% CI	19,612–73,094	18,264–101,403	−8,849–25,799
Median (IQR)	0 (0–1,987)	0 (0–10,337)	0 (0–0)
Outpatient
ER visits
Patients with ER visit, n (%)	168 (28.1%)	26 (28.0%)	5 (50.0%)
ER costs (all), US $, mean (SD)	3,130 (17,697)	1,819 (4,669)	5,333 (10,244)
95% CI	1,710–4,550	858–2,781	−1,995–12,661
Median (IQR)	0 (0–320)	0 (0–682)	835 (0–4,505)
Office visits
Patients with office visit, n (%)	555 (92.7%)	83 (89.3%)	10 (100.0%)
Office costs (all), US $, mean (SD)	3,153 (3,062)	3,357 (2,897)	4,343 (4,960)
95% CI	2,907–3,399	2,761–3,954	795–7,891
Median (IQR)	2,658 (1,398–4,122)	2,800 (1,195–4,900)	2,126 (1,742–3,493)
Facility visits
Patients with facility visit, n (%)	116 (19.4%)	17 (18.3%)	4 (40.0%)
Facility costs (all), US $, mean (SD)	737 (3,183)	1,069 (3,681)	1,555 (2,341)
95% CI	482–993	311–1,827	−119–3,230
Median (IQR)	0 (0–0)	0 (0–0)	0 (0–4,660)
Radiotherapy
Patients with radiotherapy, n (%)	185 (30.9%)	20 (21.5%)	2 (20.0%)
Radiotherapy costs (all), US $, mean (SD)	25,314 (79,096)	15,680 (42,713)	7,219 (15,231)
95% CI	18,967–31,661	6,883–24,476	−3,677–18,115
Median (IQR)	0 (0–8,772)	0 (0–0)	0 (0–0)
Diagnostic radiology
Patients with diagnostic radiology, n (%)	417 (69.6%)	63 (67.7%)	7 (70.0%)
Diagnostic radiology costs (all), US $, mean (SD)	7,789 (19,610)	6,055 (9,477)	5,994 (6.614)
95% CI	6,215–9,362	4,103–8,006	1,263–10,725
Median (IQR)	1,312 (0–7,243)	1,676 (0–8,462)	5,953 (0–8,216)
Laboratory services
Patients with laboratory service, n (%)	505 (84.3%)	81 (87.1%)	9 (90.0%)
Laboratory costs (all), US $, mean (SD)	4,273 (14,151)	4,319 (7,419)	3,118 (2,737)
95% CI	3,138–5,409	2,791–5,847	1,160–5,076
Median (IQR)	1,045 (159–3,642)	1,111 (317–4,463)	1,845 (1,476–4,842)
Other services
Patients with other service, n (%)	584 (97.5%)	91 (97.9%)	9 (90.0%)
Other outpatient costs (all), US $, mean (SD)	87,206 (139,454)	85,189 (97,735)	121,872 (115,151)
95% CI	76,016–98,397	65,060–105,317	39,498–204,245
Median (IQR)	54,275 (11,655–105,825)	55,272 (17,748–118,607)	107,859 (16,896–183,951)
Total outpatient
Total outpatient costs (all), US $, mean (SD)	131,604 (174,056)	117,488 (110,845)	149,434 (121,336)
95% CI	117,636–145,571	94,659–140,316	62,635–236,233)
Median (IQR)	87,801 (28,806–164,704)	86,583 (39,024–174,916)	138,315 (39,469–209,656)
Pharmacy
Patients with drug prescription, n (%)	494 (82.5%)	79 (85.0%)	10 (100.0%)
Pharmacy costs, US $, mean (SD)	16,507 (43,565)	18,932 (35,163)	26,162 (31,480)
95% CI	13,011–20,003	11,690–26,174	3,643–48,681
Median (IQR)	3,207 (557–9,393)	3,596 (758–13,855)	8,699 (2,625–63,571)
Total healthcare
Mean (SD)	194,463 (382,463)	196,253 (265,400)	184,071 (132,816)
95% CI	163,773–225,154	141,595–250,912	89,060–279,081
Median (IQR)	121,102 (48,369–226,013)	130,760 (57,794–209,568)	173,413 (78,685–281,994)

Abbreviations. ER, emergency department; IQR, interquartile range (Q1–Q3); PPPY, per patient per year.

The 10 most common chemotherapy agents by line of chemotherapy are presented in Figure 4. During 1L and 2L chemotherapy, etoposide and carboplatin were the most commonly used agents, with 39.7% and 32.7% of patients, respectively, receiving these agents in 1L, and 33.3% and 28.0% of patients, respectively, receiving these agents in 2L. The most common chemotherapy regimens used in 1L were carboplatin + etoposide (22.2%), fluorouracil (11.7%), cisplatin + etoposide (7.3%), and megestrol acetate (5.5%)³⁴. The proportions of patients receiving 2L chemotherapy were 17.9% of patients treated with combination therapy in 1L and 13.1% of patients with monotherapy in 1L. Among the 93 patients receiving ≥2L chemotherapy, the most common treatments used were 1L carboplatin + etoposide (2L topotecan (7.5%), followed by 1L cisplatin + etoposide (2L carboplatin + etoposide (4.3%), and 1L fluorouracil (2L carboplatin + etoposide (4.3%)³⁴.

Figure 4. Most commonly administrated chemotherapy agents by line of treatment among the overall chemotherapy-treated population. 1L, first line; 2L, second line; 3L, third line.

All resource use and costs were amplified within the lines of chemotherapy compared with the overall patients’ follow-up post-MCC diagnosis (Table 3). The overall mean costs PPPY across multiple LOTs were: 1L, $194,463 (median = $121,102), 2L, $196,253 (median = $130,760), and 3L, $184,071 (median = $173,413). Major cost components within lines of chemotherapy were primarily outpatient care (1L, 67.7% of total costs; 2L, 59.9%; 3L, 81.2%), followed by inpatient care in 1L and 2L (1L, 23.8% of total costs; 2L: 30.5%), or pharmacy prescriptions in 3L (14.2% of total costs). Most of the costs incurred within the chemotherapy lines were for outpatient care.

MCC comorbidity and associated HRU and costs

The mean Deyo-CCI score assessed over the entire observation period was 2.14 (SD = 2.26), with the highest percentage of patients having a score ≥3 (33.6%). During the entire observation period, lower mean comorbidity scores were observed in patients in the child/adolescent cohort compared with those in the adult and elderly cohorts (0.58, 1.43, and 2.61, respectively; Table 1). The most frequently occurring comorbidities within the total patient population were diabetes (27.8%), chronic pulmonary disease (23.1%), cerebrovascular disease (19.2%), congestive heart failure (14.2%), and renal disease (14.0%)³⁵.

After breaking down costs by Deyo-CCI level, assessed within the study observation period (Table 4), resource use and costs increased significantly with patients’ worsening comorbidities burden³⁵. Mean total healthcare PPPY costs increased from $62,401 (median = $27,163) in CCI level 0 to $73,537 (median = $32,929) in CCI level 1, $95,741 (median = $37,745) in CCI level 2, and $109,690 (median = $54,640) in CCI level ≥3. Mean inpatient care costs were estimated at $10,623 in level 0 and increased about $10,000 for each comorbidity level, up to $43,339 in level ≥3. Mean outpatient care costs were estimated at ∼ $48,000 in levels 0 and 1 and then increased up to ∼ $60,000 in levels ≥2, with a substantial increase in radiotherapy related costs between level 1 ($9,600) and level 2 ($15,406). Outpatient costs associated with ER visits and facility visits also showed a general trend toward higher costs with increasing comorbidity level. Mean pharmacy costs also increased with severity and ranged from $3,715 in level 0 to $5,368 in level ≥3.

Table 4. All-cause healthcare utilization and healthcare costs PPPY, by Deyo Charlson Comorbidity Index (CCI) level (n = 2,177).

	CCI level: 0 (n = 594)	CCI level: 1 (n = 502)	CCI level: 2 (n = 347)	CCI level: ≥ 3 (n = 734)	p-value
Inpatient
Patients with inpatient admission, n (%)	156 (26.3%)	176 (35.1%)	172 (49.6%)	451 (61.4%)	<.0001^a
Inpatient costs (all), US $, mean (SD)	10,623 (38,448)	21,144 (76,626)	30,844 (102,283)	43,339 (165,482)	<.0001^b
95% CI	7,525–13,721	14,425–27,864	20,044–41,643	31,348–55,331
Median (IQR)	0 (0–1,634)	0 (0–8,772)	0 (0–15,251)	6,778 (0–28,479)
Outpatient
ER visits
Patients with ER visit, n (%)	162 (27.3%)	187 (37.3%)	173 (49.9%)	400 (54.5%)	<.0001^a
ER costs (all), US $, mean (SD)	917 (3,647)	838 (2,595)	1,153 (2,891)	1,885 (6783)	.0001^b
95% CI	623–1,211	611–1,066	848–1,458	1,394–2,377
Median (IQR)	0 (0–96)	0 (0–502)	0 (0–944)	178 (0–1,347)
Office visits
Patients with office visit, n (%)	575 (96.8%)	489 (97.4%)	336 (96.8%)	712 (97.0%)	.9^a
Office costs (all), US $, mean (SD)	1,865 (2,301)	1,925 (2,722)	1,891 (1,730)	2,210 (2,032)	.02^b
95% CI	1,679–2,050	1,687–2,164	1,708–2073	2,063–2,357
Median (IQR)	1,172 (612–2,317)	1,350 (754–2,421)	1,448 (863–2,330)	1,768 (1,039–2,831)
Facility visits
Patients with facility visit, n (%)	82 (13.8%)	113 (22.5%)	111 (32.0%)	356 (48.5%)	<.0001^a
Facility costs (all), US $, mean (SD)	105 (891)	288 (1,248)	396 (1,420)	1,190 (4,034)	<.0001^c
95% CI	33–177	179–398	246–546	897–1,482
Median (IQR)	0 (0–0)	0 (0–0)	0 (0–161)	0 (0–783)
Radiotherapy
Patients with radiotherapy, n (%)	204 (34.3%)	196 (39.0%)	133 (38.3%)	327 (44.6%)	.002^a
Radiotherapy costs (all), US $, mean (SD)	10,127 (26,571)	9,600 (24,202)	15,406 (56,080)	13,059 (43,517)	.08^b
95% CI	7,986–12,268	7,477–11,722	9,485–21,327	9,906–16,213
Median (IQR)	0 (0–7,589)	0 (0–7,948)	0 (0–9,657)	0 (0–9,847)
Diagnostic radiology
Patients with diagnostic radiology, n (%)	516 (86.9%)	457 (91.0%)	321 (92.5%)	689 (93.9%)	.0001^a
Diagnostic radiology costs (all), US $, mean (SD)	6,430 (10,360)	6,417 (11,069)	7,338 (16,415)	5,364 (10,060)	.7^c
95% CI	5,595–7,265	5,446–7,388	5,604–9,071	4,635–6,093
Median (IQR)	1,879 (300–7,886)	2,055 (376–7,058)	2,321 (499–7,270)	2,037 (487–5,359)
Laboratory services
Patients with laboratory service, n (%)	554 (93.3%)	469 (93.4%)	329 (94.8%)	706 (96.2%)	.07^a
Laboratory costs (all), US $, mean (SD)	2,435 (5,146)	2,170 (5,222)	2,348 (4,756	2,451 (4,498)	.8^b
95% CI	2,021–2,850	1,712–2,628	1,845–2,850	2,125–2,777
Median (IQR)	772 (253–2,220)	748 (201–2,242)	908 (324–2,367)	1,062 (385–2,575)
Other services
Patients with other service, n (%)	587 (98.8%)	499 (99.4%)	343 (98.9%)	733 (99.9%)	.09^a
Other outpatient costs (all), US $, mean (SD)	26,184 (49,378)	27,225 (49,899)	31,786 (71,966)	34,823 (63,791)	.03^b
95% CI	22,205–30,163	22,850–31,601	24,187–39,385	30,201–39,446
Median (IQR)	8,030 (2,424–31,749)	8,973 (3,213–29,469)	11,012 (3,836–32,791)	15,966 (6,589–37,586)
Total outpatient
Total outpatient costs (all), US $, mean (SD)	48,064 (72,120)	48,463 (67,091)	60,317 (109,619)	60,983 (91,478)	<.0001^c
95% CI	42,252–53,875	42,580–54,346	48,743–71,891	54,354–67,612
Median (IQR)	18,771 (5,787–67,489)	23,736 (6,937–63,570)	25,341 (9,082–70,123)	33,336 (14,767–73,106)
Pharmacy
Patients with drug prescription, n (%)	437 (73.6%)	364 (72.5%)	259 (74.6%)	541 (73.7%)	.9^a
Pharmacy costs, US $, mean (SD)	3,715 (10,962)	3,930 (10,280)	4,580 (9,942)	5,368 (10,651)	.02^b
95% CI	2,831–4,598	3,028–4,831	3,531–5,630	4,596–6,140
Median (IQR)	481 (0–2,628)	940 (0–3,505)	1,570 (0–4,865)	2,305 (0–6,088)
Total healthcare
Total healthcare costs (all), US $
Mean (SD)	62,401 (92,420)	73,537 (115,361)	95,741 (180,415)	109,690 (200,280)	<.0001^c
95% CI	54,954–69,849	63,421–83,653	76,692–114,790	95,177–124,203
Median (IQR)	27,163 (8,281–80,595)	32,929 (11,047–91,342)	37,745 (15,953–99,068)	54,640 (24,460–116,744)

Abbreviations. ER, emergency department; IQR, interquartile range (Q1–Q3); PPPY, per patient per year.

^a Categorical characteristics were compared using χ² test.

Continuous characteristics were compared between groups using ^bStudent or ^cWilcoxon test, as appropriate.

Discussion

The purpose of this observational, non-interventional cohort study was to identify demographic and clinical characteristics, including comorbidities, to delineate real-world longitudinal treatment patterns, and estimate HRU and healthcare costs for patients with newly-diagnosed MCC in the US. In this study, we demonstrated that patients with MCC have a high comorbidity burden as well as high costs associated with chemotherapy treatment and increasing comorbidities. Our study is, to our knowledge, one of the first to describe HRU and costs for patients with MCC by age, type of treatment, line of chemotherapy treatment, and comorbidity burden. However, given that the diagnoses of MCC described in this study were between 2010 and 2014, our results are based on historic treatments, as there were no officially approved treatments for MCC during that time span.

Similar to previous reports^36,37, we found in this study that the majority of patients with newly-diagnosed MCC were elderly, with a mean age at diagnosis of 69 years. The majority of the patients with MCC were identified in the southern region of the US. This is likely due to elevated exposure to UV light, a well-known risk factor for MCC as well as other skin cancers³⁸. In addition, consistent with the higher incidence of MCC in men than in women, the majority of the study patients were male, and this gap in proportions increased in elderly patients¹³. When comparing MCC to other skin cancers, total mean costs associated with MCC ($86,227 PPPY) are similar to total mean costs associated with cutaneous squamous cell carcinoma ($88,630 PPPY), while total mean costs associated with melanoma vary based on the stage of cancer, ranging from less than $1,000 PPPY for Stage 0 to over $150,000 PPPY for Stage IV^39,40.

Real-world data describing treatment patterns and HRU and costs among patients diagnosed with MCC are scarce; furthermore, these studies do not cover details regarding the costs associated with line of chemotherapy or increasing comorbidities associated with MCC. To our knowledge, only two recent studies investigated treatment patterns and total healthcare costs of MCC specifically in patients with advanced MCC (aMCC; stage III or IV disease)^41,42. Steuten et al.⁴¹ described treatment patterns and total direct healthcare costs among 257 Medicare enrollees diagnosed with aMCC in the US between 2006 and 2013, using the SEER-Medicare database, of whom 51% had stage IIIb and 49% had stage IV disease. The authors considered 1L treatment to be surgery, radiotherapy, or chemotherapy received within 4 months after diagnosis. Kearney et al.⁴² described clinical characteristics, HRU, and management costs among 48 patients diagnosed with aMCC or mMCC in the US between 2010 and 2016 using the HealthCore Integrated Research Database (HIRD). Patients with stage IIIb or IV disease were confirmed by linking patient data to state registries or the HealthCore Integrated Research Environment-Oncology (HIRE-O) dataset.

Our study found a slightly higher percentage of elderly patients who received treatment defined as surgery, radiotherapy, or chemotherapy, compared with elderly patients in the SEER study (results from the HIRD study were not further broken down by age), with a total of 89.7% vs 84%; this finding may be related to the restriction used by Steuten et al.⁴¹ to treatment received within 4 months after diagnosis, while in our study we evaluated treatment at any time after diagnosis¹⁹. Of the treated patients, 37% (80 of 216) and 50% (24 of 48) had received chemotherapy, per the SEER and HIRD data, respectively, while our study found that chemotherapy was used in 32% (599 of 1,861) of treated patients^19,41,42. Thirty-three per cent of patients treated with chemotherapy received 2L chemotherapy according to Steuten et al.⁴¹, while in our study 15.5% of patients received 2L chemotherapy. However, the method for defining a 2L chemotherapy was not discussed in the Steuten et al.⁴¹ study, thus complicating the comparison. We found that carboplatin + etoposide, cisplatin + etoposide, and carboplatin were some of the most commonly prescribed regimens, but our study also highlighted the common use of fluorouracil and megestrol acetate, especially as monotherapy in 1L chemotherapy. Our findings also mirror the results from the HIRD study, in which carboplatin and etoposide were the most commonly administered chemotherapy agents, although breakdown by treatment line was not provided⁴².

Steuten et al.⁴¹ estimated the median total 12-month direct healthcare costs—including diagnostics and imaging, treatment procedures, inpatient and outpatient visits, hospice, home healthcare, and durable medical equipment for any condition—to be $29,680 in patients receiving no treatment, $43,384 in patients receiving surgery and/or radiotherapy but no chemotherapy, and $48,939 in patients receiving chemotherapy only or with other treatment⁴¹. Consistent with Steuten et al.⁴¹, we found that patients receiving no treatment had the lowest median costs and that costs were the highest in the patients receiving chemotherapy alone or combined with radiotherapy and/or surgery. However, our study estimated much higher total median healthcare costs PPPY in patients treated with chemotherapy, ranging from $63,063 in patients treated with surgery and chemotherapy only to $129,041 in patients treated with chemotherapy and radiotherapy only. This may be due to under-estimation of costs in the Steuten et al.⁴¹ study, which included patients who may not have had complete claims beyond a 4-month enrollment period or patients with variable follow-up duration.

This is, to our knowledge, one of the first studies quantifying the resource use and costs within the lines of chemotherapy in patients with newly-diagnosed MCC. Despite the high costs associated with treatment of MCC with chemotherapy, ORR and DOR after treatment remain low, thus highlighting the need for novel therapies that are both more economical and efficacious²⁸. Further analyses on the burden of antineoplastic chemotherapy agents and related toxicities, along with the burden of IO and related toxicities, and their impact on healthcare costs, should be investigated to further understand the differences in the economic impact between different treatment strategies used in the clinical management of MCC.

Data on the burden of comorbidities in patients with MCC are limited^19,30. We found that, during follow-up, 27.8% of the patients had been diagnosed with diabetes, 23.1% with chronic pulmonary disease, 19.2% with cerebrovascular disease, 14.2% with congestive heart failure, and 14.0% with renal disease⁴³. Overall, patients had a mean Deyo-CCI score of 2.14, with more than a third of patients scoring ≥3. The mean Deyo-CCI score was even higher in the HIRD study population (4.8), although all 50 patients had aMCC. These results illustrate that the prevalence of comorbidities is higher among patients with MCC than previously reported⁴⁴, and demonstrate the potential difference between evidence generated in the clinical trial setting in highly selected populations and the treatment challenges relevant to real-world patient populations⁴⁵. Because HRU and costs increase with worsening comorbidity level, further research may determine whether the presence and extent of comorbidities in patients with newly-diagnosed MCC affects their treatment options and tolerance, and whether these are associated with poor health outcomes. Furthermore, an increasing comorbidity burden, especially in elderly populations, may limit these patients’ treatment options. Although IO agents are promising new alternatives in the treatment of MCC^20,46,47, their economic impact remains to be determined.

Limitations

Some limitations of our study should be noted. The study population was limited to individuals with large employer-sponsored health coverage and/or Medicare supplemental coverage who are enrolled in the MarketScan database. Consequently, findings based on these patients’ data may not be generalizable to other populations in the US, such as people without insurance, those employed by small-to-medium companies, those enrolled for shorter periods, and those covered by Medicaid or military-based insurance programs. Moreover, claims data may not fully capture physician-, system-, and patient-level factors that may influence treatment choice and resource use. For example, we cannot directly capture the patient’s level of physical function, self-perceived health status, side-effects of medication, or reasons for discontinuing therapy. This also limits our ability to predict or explain the reasons for administering chemotherapy and/or external beam radiotherapy. In addition, because the database used in this study does not contain the clinical detail required to know the line of chemotherapy that was assigned by the treating physician, lines of chemotherapy were defined using algorithms based on the timing of administrative claims. Consequently, results from these algorithms may not reflect the definition of lines of therapy used in clinical practice. Moreover, we were unable to exclude the presence of concurrent malignancies in patients based on their claims data. The data used in our study also came from patients with MCC who were diagnosed from 2010 to 2014, and thus their treatment paradigms are reflective of practices prior to the recent emergence of IO. Our study also involved only descriptive analyses. Regression analyses on HRU and costs could be considered in the future, with adjustments based on insurance type and other potential confounders, to consider the skewed distribution of costs and confounders. Lastly, information regarding MCC disease staging was not provided in the database data. We were, thus, unable to determine HRU and costs by MCC stage, which would provide additional insight into the economic burden on patients with MCC.

Conclusions

Prior to the advent of IO, patients with MCC, especially those with advanced or metastatic disease, historically had limited treatment options. Our study is one of the first to document HRU and costs associated with MCC by age, type of treatment, line of chemotherapy treatment, and comorbidity burden. We found that overall costs are consistent with those observed in previous studies, and choice of treatment strategy is a main driver of the healthcare costs of MCC, with the highest costs incurred in patients receiving chemotherapy and radiotherapy without surgery and in those receiving chemotherapy only. Our results also suggest that there is a substantial comorbidity burden in patients with MCC and that HRU and costs significantly increase with patients’ worsening comorbidity burden, which may also limit their treatment options. IO represents a new option for the treatment of MCC, with the first immune checkpoint inhibitor therapy having been approved in 2017. Our results stress the importance of factoring in treatment line and comorbidity in patient cost/benefit analyses and will help inform clinicians in comparing historical treatment costs alongside new treatment strategies in determining patient options. However, further analysis of the real-world outcomes and economic impact of these new agents needs to be conducted.

Transparency

Declaration of funding

This study was sponsored by an alliance between Merck KGaA, Darmstadt, Germany, and Pfizer, New York, NY. Medical writing support was provided by ClinicalThinking, Inc, Hamilton, NJ, and funded by Merck KGaA, Darmstadt, Germany, and Pfizer, New York, NY.

Declaration of financial/other interests

M.K, E.B., and M.B. are employees of Merck KGaA. K.T. is an employee of Creativ-Ceutical, which received funding for this research from Merck KGaA. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in, or financial conflict with, the subject matter or materials discussed in the manuscript.

Acknowledgments

The authors would like to thank Dina Vemming Oksen and Michael Locker, former employees of Merck KGaA and EMD Serono, Ltd, respectively, and José Peñalvo, employee of Merck KGaA, for their contributions to the planning and design of this research. Data from this study were previously presented in poster format at the 2017 AMCP NEXUS, 2018 AAD, and 2018 ISPOR annual meetings, and as an abstract for the 2018 ASCO annual meeting.