Real-world treatment patterns, cost of care and effectiveness of therapies for patients with squamous cell carcinoma of head and neck pre and post approval of immuno-oncology agents

Abstract

Aims: In 2016, nivolumab and pembrolizumab were approved for the treatment of squamous cell carcinoma of the head and neck (SCCHN) following progression after initial platinum-based therapy. We sought to explore the uptake, effectiveness, and impact on healthcare resource utilization (HRU) and total costs of care pre and post introduction of immuno-oncology (IO) agents.

Materials and Methods: Recurrent/metastatic SCCHN patients were identified from a healthcare claims clearinghouse by selecting patients with a claim for distant metastases or who initiated systemic therapy at least 120 days following discontinuation of platinum-based therapy. Two cohorts were created according to the date of post-platinum therapy (PPT) initiation: pre-IO = 08/01/2014-07/31/2015; post-IO = 08/01/2016-07/31/2017. Treatment patterns and effectiveness (duration of treatment, time to next treatment) during first-line (1 L) PPT, HRU, and costs were compared between propensity-score matched patients from each cohort.

Results: Of 716 patients identified (pre-IO = 265, post-IO = 451) 46.3% of post-IO patients received IO post-platinum. In 229 matched patients 20.0% of the post-IO compared to 10.7% of the pre-IO (p=.02) had at least a 6 month duration of 1 L PPT. Inpatient admissions during 1 L PPT: 34.1% post-IO versus 48.0% pre-IO (p= <.01). PPPM total costs of care in 1 L PPT were significantly greater post-IO ($11,535) compared to pre-IO ($9,054, p=.002). Time to next treatment (from 1 L PPT start) was 6.1 months pre-IO versus 7.4 months post-IO (p=.046).

Limitations: Recurrent SCCHN patients were identified using a validated claims-based algorithm but misclassification may occur. Requiring patients to have received 1 L PPT the pre-IO cohort may be systematically different that the post-IO cohort as pre-IO patients were more likely to have not received further treatment beyond 1 L PPT.

Conclusions: The significant uptake of IO therapy resulted in longer durations of therapy, lower rates of hospitalizations although higher treatment costs. The results suggest IO treatment provides additional clinical benefits to recurrent/metastatic SCCHN patients.

Keywords:

• Head and neck cancer
• immuno-oncology
• healthcare costs
• treatment patterns
• effectiveness

JEL CLASSIFICATION CODES:

• I11
• I19

Introduction

Head and neck cancers (HNC) are tumors that arise in the oral cavity, pharynx, larynx, sinuses, and salivary glands. The majority of cases occur in men (74%), begin in squamous epithelial cells (squamous cell carcinoma of the head and neck [SCCHN]), and are caused by smoking, chewing tobacco, and human papilloma virus infection (HPV).¹ Collectively HNC accounts for ∼4% (64,690 individuals) of all incident cancer diagnoses each year in the United States.^1,2 Patients diagnosed with limited or early-stage disease (40% of cases) are treated with radiotherapy or surgery which results in 5-year survival rates above 80%.^2,3 Treatment for locally advanced disease (stage III, IVA/B) uses radiation therapy in combination or sequentially with platinum-based chemotherapy.³ When disease recurs or when diagnosed with stage IVC, often lumped as recurrent/metastatic disease (<5% of incidence cases), SCCHN is incurable.

Prior to 2016, treatment options for recurrent/metastatic SCCHN included radiation, surgery, chemotherapy, and targeted therapy in combination with chemotherapy. In 2016, immuno-oncology (IO) therapies including nivolumab and pembrolizumab ushered in a new era of treatment options for metastatic/recurrent SCCHN. Pembrolizumab was granted accelerated approval in August 2016 with nivolumab approved in November of that year, both received an indication for the treatment of recurrent or metastatic SCCHN with disease progression on or after platinum-based therapy.^4,5 Nivolumab’s approval was based on the results of the CheckMate-141 trial in which treatment with nivolumab improved median overall survival (OS) to 7.5 months compared to 5.1 months in the comparator arm (treatment of physician choice including cetuximab, methotrexate or docetaxel) which corresponded to a 30% reduction in the risk of death (HR = 0.70, p=.01).⁶ Pembrolizumab’s accelerated approval was based on the results of the non-randomized, single-arm KeyNote-012 trial which observed an objective response rate of 16% for pembrolizumab-treated patients.⁷ The protocol-specified final analysis of KEYNOTE-040, the phase III trial Pembrolizumab versus standard therapy (methotrexate, docetaxel or cetuximab), failed to meet its primary endpoint of OS (HR = 0.81; 95% CI: 0.66–0.99; p=.0204 [one-sided]).⁸ In an updated post hoc analysis with survival data from 12 previously censored patients, the results showed that the HR decreased to 0.80 (95% CI: 0.65–0.98 [p=.0161]) for OS in the ITT population.⁹

With chemotherapy providing limited benefit for recurrent/metastatic SCCHN, IO treatments hold a significant promise to improve and extend the lives of these patients. Several economic modeling studies have explored the cost-effectiveness (CE) of nivolumab for recurrent/metastatic SCCHN. Compared to cetuximab, nivolumab is cost-effective.¹⁰ However other models have shown that at current pricing, while nivolumab extends OS, it is not cost-effective at a standard willingness-to-pay threshold of $100,000 (no CE studies of pembrolizumab for SCCHN have been reported at the time of manuscript writing).^11–14 Compared to CE analyses, real-world costs derived from third-party payer claims allow estimation of the actual costs of care reflecting the real-world duration of treatment and rate of events because all billed procedures, treatments and events are captured. This approach may provide a more accurate and complimentary view of the value of a novel therapy to a third-party payer or society. Therefore, we undertook a third-party payer claims analysis to describe how the introduction of IO therapies to the treatment paradigm for recurrent/metastatic SCCHN has changed treatment patterns, impacted clinical outcomes and changed total costs of care from a healthcare payer perspective.

Methods

Study design

The objectives of this research were to: (1) describe the uptake of IO therapies for the treatment of recurrent/metastatic SCCHN, (2) describe the real-world effectiveness of IO therapies by evaluating treatment duration, and (3) compare the healthcare resource utilization (HRU, inpatient admissions and emergency department visits) and total costs of care in the post-platinum setting by line of therapy between pre- and post-IO treated patients.

We conducted a retrospective cohort study using payer claims data from the Symphony Health Integrated Dataverse (SHIDV). The SHIDV integrates data from physician practices, pharmacies and hospitals for a broad longitudinal view of a patient's disease history, treatment patterns and healthcare resource utilization. Individuals with any non-diagnostic medical claims containing an ICD-9/10 code for head and neck cancer (ICD 9: 140–149, 160–161, ICD 10: C00–C14, C30–C32), with a minimum of 6 months pre- and 6-months post diagnosis medical or pharmacy activity (i.e. at least one claim), who had received initial platinum-based chemotherapy, and who had initiated a subsequent post-platinum therapy (PPT) were selected. Patients who were <18 years of age at diagnosis, had a second primary cancer diagnosis or received care as part of a clinical trial (ICD-9: V70.7; ICD-10: Z00.6) were excluded.

Line of therapy determination was made using the following algorithm: systemic therapy agents administered within 30 days of each other were considered combination therapy; change in regimen (use of new agents, addition of an agent or discontinuation of an agent in a regimen for at least 60 days), or a gap (no claims for the agent(s)) of more than 12 weeks triggered a line-of-therapy increase. A switch from cisplatin to carboplatin or vice versa (as monotherapy or part of a combination) was not considered a new line of therapy unless there was a gap of 90 or more days between the platinum agents. To select only patients with recurrent disease, only those individuals who initiated PPT 120 days or more after discontinuing platinum-based therapy were selected.¹⁵ A live meeting of US-based oncologists (who had treated ≥5 SCCHN patients in the last 12 months) was convened to evaluate the validity of this previously published algorithm.¹⁶ Those patients with progression prior to 120 days (i.e. initiated PPT prior to 120 days) were required to have at least one medical claim with an ICD-9/10 diagnosis code for a distant metastatic site prior to being eligible for the analysis.

Healthcare resource utilization and costs

Acute care interventions were identified using admission codes for hospitalizations and emergency department (ED) visits. Paid costs were not available for medical claims. Standardized costs were created for both medical and pharmacy claims for drug costs (cost per administration of any antineoplastic agent or other pharmaceutical) by using the average wholesale price (AWP) for drugs coded with NDC codes and the Centers for Medicare & Medicaid Services (CMS) Average Sales Price (ASP) for drugs coded with J, Q, or C codes. The CMS Clinical Laboratory Fee Schedule (CLFS), the Medicare Physician Fee Schedule (MPFS), and Hospital Outpatient Prospective Payment System (OPPS) were used to standardize procedure costs. Inpatient costs were derived using the cost-to-charge ratio from HCUP (Healthcare Costs and Utilization Project). All costs were adjusted to 2017 US dollars (USD).

Statistical analyses

From the population of eligible patients with recurrent/metastatic SCCHN, we constructed two mutually exclusive cohorts. The pre-IO cohort consisted of patients who initiated PPT between 1 August 2014 and 31 July 2015, whereas the post-IO cohort consisted of patients who initiated PPT between 1 August 2016 and 31 July 2017. The end of the study period was 30 April 2018 (data cut-off). Patients who initiated PPT during the gap between the pre- and post-IO periods (1 August 2015–31 July 2016) were excluded in order to remove the effects of patient cross-over. Duration of treatment in the PPT setting was measured as the time from the first PPT treatment claim until the last PPT treatment claim(s) for that regimen; the proportion of patients still receiving treatment at 6, 9, and 12 months following the initiation of PPT was calculated among patients with at least those many months of follow-up from PPT start. Time to next treatment was assessed from the start of the PPT until treatment switch or until a gap of >90 days with no subsequent therapy (both considered events). The Kaplan–Meier method was used to estimate time to next treatment (patients censored on the last date of treatment in PPT if did not initiate subsequent line of PPT). The frequency of any inpatient admission or emergency department visit per line was assessed from the start of PPT to the initiation of the subsequent line or last follow-up date if no subsequent line. Descriptive comparisons between the pre- and post-IO rates of HRU were conducted using Student’s t-test or Kruskal–Wallis tests for continuous variables and χ² for categorical variables. Per patient per month (PPPM) costs of care were calculated to account for the variable follow-up of patients by line or cohort. Reported costs included inpatient costs, emergency department (ED) costs, and total costs (inpatient, ED, systemic therapy, office visits, laboratory and imaging tests, concomitant medications, drug administration costs, and other procedures). Statistical comparisons of costs were made using a generalized linear regression model with a log-link function and gamma distribution. Duration of treatment, time to next treatment, HRU and costs were compared between a set of propensity-score matched (PSM) patients using greedy nearest neighbor matching method with a 0.1 caliper width on age at diagnosis, sex, recurrent/metastatic disease, geographic region, comorbidities, and length of follow-up from initiation of systemic therapy. An estimate of the duration of treatment was conducted using the Kaplan-Meier method and comparisons of duration of treatment were made using the log-rank test.

Results

Study cohorts characteristics

Overall, 716 patients with recurrent/metastatic SCCHN were identified including 265 (37.0%) pre-IO and 451 (63.0%) post-IO patients (Table 1). The pre-IO cohort had a lower proportion of patients with oral cavity tumors (and higher proportion with pharynx as the primary site) compared to the post-IO cohort (p=.009). There were no significant differences in comorbidities at the time of diagnosis. No statistically significant differences were observed in the proportion of patients with distant metastatic sites in each cohort. The most common distant metastatic sites were lung, occurring in 24.9% and 18.6% of the pre- and post-IO cohorts, respectively, followed by bone (21.1% versus 22.6%, respectively). During first-line platinum-based therapy, 64.5% of the pre-IO and 66.5% of the post-IO cohort received radiation (at least one claim for radiation from 60 days prior to 60 days post the start of first line). Overall, the median length of follow-up from the initiation of 1 L PPT was 11.9 months in the pre-IO cohort and 9.9 months in the post-IO cohort (p<.0001).

Table 1. Demographic, clinical, and treatment related characteristics of the pre-IO and post-IO cohorts.

	Pre-IO n = 265		Post-IO n = 451		p-Value*
Male (n, %)	210	79.2	357	79.2	.965
Mean Age at Diagnosis (SD)	58.4	(9.0)	59.8	(9.4)	.054
Payer Type (n, %)					.018
Commercial	147	55.5	212	47.0
Medicare	49	18.5	131	29.0
Medicaid	50	18.9	78	17.3
Other (dual eligible, VA-military, etc.)	19	7.2	30	6.7
U.S. Geographic Region (n, %)					.827
Central/Midwest	64	24.2	100	22.2
Northeast	53	20.0	86	19.1
South	70	26.4	118	26.2
West	78	29.4	147	32.6
Primary Tumor Site (n, %)
Any oropharyngeal site	242	91.3	424	94.0	.172
Oral cavity	134	50.6	280	62.1	.009
Pharynx	76	28.7	88	19.5
Larynx	32	12.1	56	12.4
Paranasal sinuses and nasal cavity	11	4.2	8	1.8
Salivary glands	12	4.5	19	4.2
Initial Platinum-Based Therapy (n, %)					<.001
Platinum monotherapy	123	46.4	237	52.5
Platinum combination	26	9.8	150	33.3
Cetuximab-platinum	104	39.2	41	9.1
Platinum-cetuximab-5FU	12	4.5	23	5.1
Radiation treatment initial Platinum-based therapy (n, %)	171	64.5	300	66.5	.588
Metastatic disease at initiation PPT (n, %)^†
Bone	56	21.1	102	22.6	.212
Brain	9	3.4	11	2.4	.563
Liver	23	8.7	39	8.6	.987
Lung	66	24.9	84	18.6	.046
Charlson Comorbidity Index at initiation of PPT (mean, SD)	2.9	3.2	2.7	3.1	.329
Total lines of PPT					.823
1	265	100	451	100
2	78	29.4	143	31.7
≥3	25	9.4	39	8.6
Months of follow-up from start of PPT (mean, SD)	17.8	14.5	9.8	5.5	<.001

Abbreviations. 5-FU: 5-fluorouracil; SD: standard deviation; PPT: post platinum therapy.

^† Sites of disease not mutually exclusive.

* p-Value generated from Student’s test for continuous variables and χ² for categorical variables.

Treatment patterns

By definition, all patients initiated at PPT. In 1 L PPT 46.3% (n = 209/451) of patients in the post-IO cohort received IO therapy with either pembrolizumab or nivolumab. During the study period 29.4% (78/265) of the pre-IO cohort and 31.7% (143/451) of the post-IO cohort initiated 2 L PPT (Table 1). Among the 143 post-IO patients who had received a 2 L PPT, 54 (37.8%) had been previously exposed to IO (i.e. in 1 L PPT) and of the remaining 89, 65/89 (73.0%) received IO as 2 L PPT (data not shown). The use of IO replaced cetuximab use in 1 L PPT (monotherapy or in combination including with fluorouracil [5FU] and platinum) which declined from 47.9% (127/265) of regimens in the pre-IO period to 22.0% (99/451) in the post-IO period (Table 2). Sequential platinum monotherapy and platinum combination therapy as 1 L PPT also declined from 38.5% (102/265) in the pre-IO cohort to 28.2% (127/451) in the post-IO cohort. Taxanes use in 1 L PPT declined to <1.0% among patients who had received initial platinum monotherapy, platinum combination or platinum-cetuximab-5FU in the post-IO cohort.

Table 2. Treatment sequence from initial platinum-based therapy to 1 L PPT in the pre-IO and post-IO cohorts, respectively^†.

Pre-IO cohort	1L PPT
	Any IO	Cetuximab mono	Cetuximab combo	Platinum mono-therapy	Platinum combo	Platinum +  cetuximab +5FU	Taxane	Other chemo mono	Other chemo combo
(n = 265)	(n = 0)	(n = 68)	(n = 44)	(n = 34)	(n = 68)	(n = 15)	(n = 15)	(n = 19)	(n = 2)
Platinum monotherapy (n = 123)	0.0%	27.6%	17.9%	3.3%	30.1%	8.1%	5.7%	6.5%	0.8%
Cetuximab-platinum combination (n = 26)	0.0%	11.5%	19.2%	23.1%	23.1%	0.0%	7.7%	11.5%	3.8%
Platinum combination (n = 104)	0.0%	27.9%	15.4%	20.2%	22.1%	3.8%	3.8%	6.7%	0.0%
Platinum + cetuximab + 5FU (n = 12)	0.0%	16.7%	8.3%	25.0%	16.7%	8.3%	16.7%	8.3%	0.0%
(n = 451)	(n = 209)	(n = 57)	(n = 29)	(n = 42)	(n = 85)	(n = 13)	(n = 3)	(n = 11)	(n = 2)
Platinum monotherapy (n = 237)	39.7%	15.2%	7.6%	3.8%	26.6%	3.8%	0.8%	2.1%	0.4%
Cetuximab-platinum combination (n = 41)	68.3%	2.4%	7.3%	17.1%	2.4%	0.0%	2.4%	0.0%	0.0%
Platinum combination (n = 150)	46.0%	13.3%	5.3%	16.0%	13.3%	2.0%	0.0%	3.3%	0.7%
Platinum + cetuximab + 5FU (n = 23)	78.3%	0.0%	0.0%	8.7%	4.3%	4.3%	0.0%	4.3%	0.0%

Abbreviations. 5FU: 5-fluorouracil; PPT: post platinum therapy.

^†Percentages represent the row percent, that is, the proportion of patients receiving the 1 L PPT out of those who received platinum monotherapy, cetuximab-platinum combination therapy, platinum combination therapy (without cetuximab), or platinum + cetuximab + 5-FU as their initial platinum-based therapy.

Duration of treatment/time to next treatment

The duration of treatment and time to next treatment in the PPT setting was compared between the pre- and post-IO cohorts among the propensity score matched patients from the pre- and post-IO cohorts (n = 229 per cohort). No significant differences in the distribution of matching characteristics was found after conducting the match including in the median duration of follow-up from the initiation of 1 L PPT therapy (pre-IO = 10.1 months and post-IO = 10.8 months). The proportion of patients who remained on 1 L PPT therapy at 6, 9 and 12 months was 10.7, 4.8, and 3.8% in the pre-IO cohort, and 20.0, 11.5, and 8.4% in the post-IO cohort respectively (Table 3). Differences were significant at 6 and 9 months but not 12 months. Overall, the median time to next treatment in 1 L PPT was 6.1 months (95% CI: 4.9–6.9) in the pre-IO cohort and 7.4 (95% CI: 6.5–8.9) in the post-IO cohort (p=.046).

Table 3. Duration of treatment and time to next treatment in 1 L PPT between the pre-IO and post-IO propensity score matched patients.

	Pre-IO		Post-IO		p-Value^¥
Duration of treatment in 1L PPT (n of eligible, %)^†
≥6 months	159	10.7	170	20.0	.020
≥9 months	124	4.8	148	11.5	.050
≥12 months	106	3.8	95	8.4	.234
Time to next treatment from start of 1L PPT (median, 95% CI)	6.1 (4.9–6.9)		7.4 (6.5–8.9)		.046

Abbreviations. 1 L – first line; CI: confidence interval; NR: not reached; PPT: post platinum therapy.

^† Eligible patients are those with at least 6, 9 and 12 months of follow-up from the initiation of 1 L PPT, respectively.

¥ p-Value based on χ² test for duration of treatment comparison; log-rank test for time to next treatment.

Healthcare resource utilization and total costs of care

Significantly more patients experienced an inpatient hospitalization during 1 L PPT in the pre-IO cohort (48.0%) compared to the post-IO cohort (34.1%, p=.002) (Table 4). Over the course of all follow-up in the PPT setting, inpatient admission frequency remained higher in the pre-IO cohort versus the post-IO cohort (58.1 vs. 45.4%, respectively, p=.007) while specifically in 1 L PPT 48.0% of pre-IO and 34.1% of post-IO patients experienced an inpatient admission (p=.002) No differences in the rates of ED visits were observed in both 1 L PPT and overall follow-up post-platinum. Total costs of care were significantly higher in the post-IO cohort during 1 L PPT ($11,535 PPPM vs. $9,054 PPPM, p=.005) and during the entire PPT follow-up ($11,971 PPPM vs. $9,347, p=.005). Among patients who experienced inpatient care or ED visits costs were similar, while frequency of inpatient admissions was significantly lower post-IO. Systemic therapy costs were higher post-IO in both 1 L PPT and over entire PPT follow-up.

Table 4. Comparison of healthcare resource utilization and costs* of care in the post-platinum setting between the pre- and post-IO propensity score matched patients.

	Pre-IO N = 229		Post-IO N = 229		p-Value^†
All patients, all PPT
Any Inpatient (n, %)	133	58.1	104	45.4	.007
Any ED (n, %)	91	39.7	90	39.3	.924
Mean total PPPM cost (SD)	9,347	(15,061)	11,971	(11,113)	.005
Inpatient	6,281	(8,517)	8,014	(11,336)	.112
ED	903	(1,134)	867	(820)	.760
Systemic	1,695	(2,760)	4,802	(4,564)	<.001
1L PPT
Any Inpatient (n, %)	110	48.0	78	34.1	.002
Any ED (n, %)	78	34.1	76	33.2	.843
Mean total PPPM cost (SD)	9,054	(9,404)	11,535	(10,749)	.005
Inpatient	7,281	(10,147)	8,653	(11,872)	.317
ED	1,059	(1,222)	1,103	(1,070)	.777
Systemic	2,108	(3,443)	4,846	(5,047)	<.001

Abbreviations. 1 L: first line; ED: emergency department; IO: immuno-oncology; PPPM: per patient per month; PPT: post platinum therapy.

* All costs reported in 2017 US dollars. Total costs include additional categories not represented such as office visits, other prescription drug, laboratory tests, and other procedures not received in the inpatient or ED setting.

^† p-Values for HRU comparisons were generated using the χ² test statistic. p-Values for PPPM cost comparisons were generated by comparing costs through a generalized linear model with log-link function and gamma distribution.

Discussion

The treatment paradigm for recurrent/metastatic SCCHN has rapidly shifted since the introduction of IO therapies in 2016. In this first claims-based analysis since that time we observed a dramatic uptake in IO utilization as both 1 L (46.3%) and a not surprising increase in the total costs of care due to the costs of IO therapy (pre-IO PPPM 1 L PPT=$9,054 vs. post-IO=$11,535, p = 0.005). Novel findings from this research include the HRU benefit in the form of reduced inpatient admissions observed in the post-IO versus pre-IO periods (48.0 vs. 34.1%, p=.002) and the appearance of improved real-world effectiveness in the form of longer median time to next treatment in the post-IO (7.4 months) versus pre-IO (6.1 months) period. While total costs of care are likely to be greater due to the costs of IO therapy we hypothesize that PPPM costs in the post-IO period may decline as patients survive longer with IO-based therapy and may receive additional non-IO systemic therapy post disease progression (on IO).

Earlier studies of the costs of HNC are no longer relevant given they were conducted prior to the development and utilization of IO agents.^17–20 One recently published analysis of patients with recurrent/metastatic SCCHN who underwent disease progression between 1 January 2007 and 1 October 2015 observed a slightly higher rate (58%) of any inpatient admission during 1 L PPT compared to the 48.0% pre-IO and 34.1% post-IO observed in our analysis but a lower rate of ED visits (9%) in 1 L PPT compared to our estimate of 34.1% pre-IO and 33.2% post-IO).²¹ Using a similar standardized costing approach the prior work estimated PPPM costs of $14,270 (SD = 20,353) in 1 L PPT and $12,727 (SD=$20,955) in 2 L PPT. Our PPPM total costs of care estimates were lower in both the pre and post-IO period which likely reflects the lower observed rate of hospitalizations versus ED visits given our pre-IO systemic therapy costs were similar ($2,108) compared to previously published ($2,202).²¹ Of note, we relied upon evaluation and management codes from the payer claims database to measure the occurrence of inpatient admissions and ED visits as opposed to notations of these occurrences in unstructured data from the previously published chart review.

Our estimates of time to next treatment as a marker of clinical effectiveness are the first published from a third-party payer claims database. First, it should be noted that surrogate endpoints in general, such as progression-free survival (PFS), may not correctly predict either survival or quality of life benefit for patients.^22,23 Second, we examined time to next treatment as a surrogate for PFS but note that time to next treatment also includes discontinuation due to toxicity or any other unmeasured reason for discontinuing therapy. Thus, we acknowledge this as an imperfect marker of clinical effectiveness. However, taking this into consideration, and viewing time to next treatment as an upwardly-biased estimate of PFS, as next treatment must start post disease progression, our estimates do illustrate the potential for additional benefit that would need confirmation through more rigorous evaluation of clinical benefit in future studies. Our median time to next treatment estimates of 6.1 months pre-IO and 7.4 months post-IO are consistent (marginally longer) than that observed in a prior EMR-based analysis (5.0 months for chemotherapy only treated patients).²⁴ Our estimates are also longer than the median PFS observed in CheckMate-141: nivolumab-treated = 2.0 months, standard therapy = 2.3 months.⁶ Indirect comparison of these results is difficult. In CheckMate-141 more than 50% of patients had received at least two prior lines of therapy. In the prior EMR study treatment modalities were highly variable compared to our more controlled setting where all patients had to have received initial platinum-based therapy and subsequently initiated PPT after disease recurrence or the development of distant metastatic disease. Further, long-term real-world follow-up studies are needed to evaluate the clinical benefit of IO PPT.

Limitations

First, our analysis cohort does not provide a population-based estimate of the rate of PPT treatment pre- and more importantly post-IO. We required not only the receipt of PPT but also 6 months of follow-up and while it appears that the more patients are receiving IO therapy some proportion of the pre-IO cohort who went on to get PPT may have been excluded. Second, misclassification of locally advanced disease as recurrent/metastatic disease may have occurred as stage at diagnosis was inferred by algorithm. The algorithm had high specificity, thus limiting the potential inclusion of non-recurrent cases.¹⁵ However, the low sensitivity may lead to many true cases of recurrent disease not being captured. In particular, patients with rapid disease progression who do not respond to initial platinum-radiotherapy and therefore may have a worse clinical status (and have higher resource utilization and costs), may have been differentially excluded from the study cohorts. Next, the database used for this study is comprised of the longitudinally linked prescription, medical and hospital claims, but is dependent upon the facility being a part of its network for inclusion. Acute care or standalone outpatient radiation facility not in the SHIDV network would not be captured. This, to the extent that patients received care at multiple, non-integrated hospital systems, may in part explain lower rates of radiation and hospitalization observed compared to other studies. Finally, actual paid costs were not available and therefore standardized costs were used. Although this approach focuses on contribution of treatment patterns to change in costs, it ignores differences existing in allowed paid amounts across payers and underestimates the costs to commercial third-party insurers reimbursing select services above government payers.

Conclusions

Our research shows that the introduction of IO treatments has resulted in longer duration of therapy, lower rates of hospitalization, but higher treatment-related costs. While we show that third-party payer costs are higher in the post-IO period, the longer time to next treatment and lower rate of inpatient admissions lend further support to the value of IO agents established in randomized controlled trials. Further delineation of patient subtypes who may receive a higher benefit from therapy is needed along with research to understand how patient quality of life is affected by IO therapies compared to other treatment approaches.²⁵ Given the concern over drug pricing, the conflicting evidence as to the cost-effectiveness of IO therapies in recurrent/metastatic SCCHN, and variability in willingness-to-pay thresholds for end of life care, these data may help inform how to best utilize IO therapy for HNC patients.

Transparency

Declaration of funding

This research was funded by Bristol-Myers-Squibb.

Declaration of financial/other interests

PA, BK, PS, and JS are employees of Bristol-Myers-Squibb; JKK and BF are employees of Cardinal Health. At the time of data analysis and manuscript preparation JR was an employee of Cardinal Health. JME peer reviewers on this manuscript have received an honorarium from JME for their review work, but have no other relevant financial relationships to disclose.

Author contributions

JR and JKK developed the study analysis, conducted the analysis, and wrote the manuscript. PA, BK, PS, and BF contributed to the analysis and edited the manuscript. JS edited the manuscript. All study authors participated in the design, analysis, and interpretation of the findings.

Acknowledgements

None reported.