Costs Associated With Second-Line Therapies for Lung Cancer

Lung cancer is one of the most prevalent malignancies in developed countries.¹ In the United States it has the highest incidence of any primary cancer site, accounting for 15% of all cancer diagnoses, and is the leading cause of cancer-relat

ed mortality for both men and women.²

Prognosis is poor; the 5-year survival rate is estimated at

16% when factoring in all stages, and as low as 4% when

distant metastases are present at diagnosis.² Over half of

cases are already at an advanced stage when diagnosed.^3,4

Lung cancer places a high economic burden on payers. Annual expenditures for lung cancer management in the United States have been estimated at $10.3 billion (2006 dollars).⁵ Furthermore, patients who experience lung cancer progression or treatment failure have significantly greater costs than patients with stable disease.^6,7 Most of the economic analyses involving patients with lung cancer pertain to the first-line treatment setting, and many of them predate recent major advances in lung cancer treatment.^8,9

Claims-based analyses complement the previously published economic evaluations by depicting actual rates of healthcare utilization and costs.^10,11 Such data can affect treatment provider decisions as well as payer decisions regarding coverage. In a claims-based study of Virginia Blue Cross and Blue Shield beneficiaries, Hillner et al¹¹ reported direct costs of US $47,941 per patient for 2 years after diagnosis or until death in a commercially insured population with advanced lung cancer. Patients who undergo multiple lines of treatment have costs that are 2 to 3 times higher than those with treatment limited to first-line therapy.^6,8 In another claims-based study (using data from MarketScan Commercial Claims and Encounters and Medicare Supplemental and Coordination of Benefits), patients who went on to receive second-line treatment accrued an additional US $10,370 per month in initial treatment phase costs and US $8779 more per month after starting second-line treatment and/or terminal care compared with those patients who underwent first-line treatment only.⁶

Approximately 85% of lung cancers are classified as non—small cell lung cancer (NSCLC).² First-line treatment for advanced NSCLC typically involves platinum-based chemotherapy.^12-15 Current practice guidelines recommend the use of second-line therapy in patients with advanced or metastatic lung cancer who experienced disease progression either during or after first-line therapy and who have good performance status.^12-15 Second-line therapy has been demonstrated in phase III trials to increase survival, reduce symptoms, and limit deterioration of quality of life for patients with metastatic lung cancer.^16-23

Specifically, docetaxel, pemetrexed, and erlotinib are recommended by the National Comprehensive Cancer Network for lung cancer patients after failure of first-line chemotherapy, and these agents are approved by the US Food and Drug Administration for use in this setting.^15,24-26 Erlotinib and pemetrexed (which is approved only for nonsquamous NSCLC) are also approved as maintenance therapy for patients whose disease does not progress following first-line treatment. These 3 agents differ with regard to mechanism of action, method of administration, and adverse event (AE) profile.^24-26 The cytotoxic agents docetaxel and pemetrexed are administered intravenously, whereas erlotinib, an epidermal growth factor receptor tyrosine kinase inhibitor, is an oral agent. While improved survival has been reported with each of these agents,^16,19-21,27 no head-to-head clinical trials of efficacy and toxicity of docetaxel, pemetrexed, and erlotinib have been conducted.²⁸

Economic models and cost comparisons of erlotinib, docetaxel, and pemetrexed have been published,^10,29-31 but 2 of these analyses were economic models based on clinical trial data available at the time of product launch and do not necessarily reflect real-world usage patterns.^29,30 In recognition of the scarcity of economic analyses based on actual lung cancer treatment patterns, the present study was designed with the objective of comparing real-world total healthcare costs among lung cancer patients receiving docetaxel, pemetrexed, or erlotinib monotherapy as second-line treatment. Furthermore, because toxicities associated with cancer therapy often necessitate supportive care, AE-related costs were also assessed. Specifically, we provide estimates of costs among patients receiving each treatment regimen as second-line therapy, including pharmacy costs, the costs of office visits associated with medication administration, AE-related care, and total healthcare consumption.

METHODS

In this retrospective claims analysis, data were obtained from a proprietary i3-affiliated claims database associated with a large US health plan. Approximately 14 million geographically diverse individuals with both medical and pharmacy benefit coverage were enrolled in the health plan during the study period. Data were deidentified and accessed in compliance with the Health Insurance Portability and Accountability Act³²; approval by an institutional review board was therefore not required.

Study Population

Patients were required to have at least 1 medical claim during the identification period of July 1, 2004, through December 31, 2006, with a diagnosis of lung cancer (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 162.xx, 163.xx, 231.xx) in any position (the service date of the first qualifying claim was defined as the initial diagnosis date); at least 2 medical claims with a lung cancer diagnosis on separate service dates within 6 months following the initial diagnosis date; and no more than 10 medical claims with diagnosis codes for other cancers on separate service dates in the period 30 days prior to and 6 months following the initial diagnosis date.

Patients were also required to be 18 years or older, to be continuously enrolled with both medical and pharmacy benefits from the initial diagnosis date through at least 60 days following the index date (defined as the service date on the first claim for a second-line agent), and to have received treatment with docetaxel monotherapy, erlotinib monotherapy, or pemetrexed monotherapy as secondline treatment (with or without third-line treatment).

Identification of Anticancer Therapy

First-line chemotherapy was identified as an anchor

from which to identify second-line therapy with docetaxel,

pemetrexed, or erlotinib. Medical and pharmacyclaims during the 30 days prior to and 6 months followingthe initial diagnosis date were examined for receiptof first-line chemotherapy. First-line chemotherapy wasdefined as the first chemotherapeutic agent received duringthe 213-day review period.

Second- and third-line agents were defined as agents different from the prior therapy. Docetaxel and pemetrexed were identified as second-line chemotherapy agents when they had an initial injectable administration date at least 28 days following the last delivery date of the first-line chemotherapy agent. Erlotinib, an oral medication, was counted as a second-line agent if the initial date on which the prescription was filled occurred on or after the last date of administration of first-line chemotherapy. The service date on the first claim (either pharmacy or medical) for one of the second-line agents was defined as the index date, and the identified agent was defined as the index therapy. Additional chemotherapy agents received within 8 days of the index therapy were considered part of the second-line treatment. Patients were assigned to 1 of 3 cohorts based on their index second-line treatment regimen: docetaxel monotherapy, erlotinib monotherapy, or pemetrexed monotherapy. Third-line therapy, identified in order to define the second-line treatment period, was defined as any new cancer therapy with an initial administration date at least 28 days following the last delivery of the second-line agent, with criteria analogous to those used to define second-line therapy.

Study Periods

All patients were enrolled for at least 6 months prior to and 2 months following the index date. The 6-month preindex period was defined as the baseline period and was used to assess patient demographic and baseline clinical characteristics, including age as of the index year, sex, geographic region, and Charlson-Deyo comorbidity score.³³ Patients were followed until disenrollment from the health plan or December 31, 2007, whichever occurred earlier. The second-line treatment follow-up period was defined as the time period from the index date to the end of the follow-up period (for patients without third-line treatment) or the index date to the third-line treatment date (for patients with third-line treatment). A second claim for the index therapy during the follow-up period was required for a patient to be included in the analysis, but continuation with therapy was not a criterion defining the duration of follow-up.

Study Variables

Adverse Events.

Occurrence of toxicities commonly associated with cancer therapy (which may or may not have been directly caused by the medications) and related utilization were identified through AE-related diagnosis, procedure, and medication codes (including supportive care) on claims dated within the follow-up period. Twelve AEs were selected for study based on severity grade 3/4 incidence greater than or equal to 5% as reported in the prescribing information for docetaxel, pemetrexed, and erlotinib.^24-26 These were nausea/emesis, anemia, neutropenia/leukopenia, selected pulmonary AEs, dyspnea, chest pain, dehydration, fatigue/asthenia, rash, anorexia, diarrhea, and febrile neutropenia. Codes indicative of these events were categorized according to the AE (

eAppendix A

, available at www.ajpblive.com). For example, codes for antiemetic drugs and nausea and vomiting diagnoses were included in the “nausea/emesis” category. Because AE occurrence was inferred from claims, the grades of the AEs experienced by patients included in the study sample were not known.

Costs. Healthcare costs were calculated during the second-line treatment follow-up period from the perspective of the patient’s health plan. Total medical costs were calculated as the costs associated with all medical claims during the period. Additionally, component medical costs (emergency department [ED], inpatient, outpatient [outpatient hospital and office-based encounters], laboratory, skilled nursing facility, and therapy costs) were calculated. Total pharmacy costs were calculated as the cost for all prescription claims. Total healthcare costs were calculated as the sum of medical and pharmacy costs. Adverse event—related costs were calculated on the basis of AE-related utilization. Specifically, costs associated with claims indicating AE codes along with hospitalization, ED visits, outpatient visits, prescription fills, or supportive care were considered AE-related costs. Only direct costs for services covered under the patient’s insurance benefit were included in the study.

Statistical Analyses

Comparisons of baseline characteristics among patients receiving erlotinib versus those receiving docetaxel or pemetrexed were done using the t test or c2 test. Unadjusted healthcare costs and costs related to AEs incurred during the follow-up period were calculated per patient per month (PPPM) to account for variable durations, and are presented as mean and standard deviation. All costs were summed for each patient for their specific

observation period. This sum was then divided by the

patient-specific days of enrollment and multiplied by 30.

Because healthcare costs were extremely skewed, the association between treatment and healthcare costs was estimated using a generalized linear model specified with a gamma error distribution and a log link.³⁴ This method avoids the potential difficulties introduced by transformation and retransformation.³⁵ The generalized linear models were used to predict adjusted costs, and bootstrapping with 1000 replications was used to estimate confidence intervals for adjusted incremental costs.³⁶ Five models were produced for total healthcare costs and for AE-related costs by sequentially adding variables to adjust for demographics (age, sex, geographic region), comorbid conditions (cardiovascular disease or congestive heart failure, diabetes), site/type of metastatic disease (visceral metastasis, brain metastasis, bone metastasis), and short follow-up (duration less than the median). Comorbid conditions and site of metastasis might influence therapy selection (eg, a patient with visceral metastasis might be more likely to receive an intravenous drug than an oral therapy, and a patient with diabetes might be less likely to be treated with docetaxel due to the requirement for concomitant steroids). Identification of comorbidities and metastasis sites were determined based on ICD-9-CM diagnosis codes.

Adverse events are presented as the percentage of patients with evidence of the 12 specific AEs during follow-up. Comparisons of AE incidence among patients receiving erlotinib versus those receiving docetaxel or pemetrexed were performed using a c2 test. Statistical significance was defined at the P <.05 level. All analyses were conducted using version 10 of the STATA software package (Stata Corp, College Station, TX).

RESULTS

Study Sample

Demographic characteristics of the patients who received second-line treatment with docetaxel, pemetrexed, or erlotinib and met the study criteria are summarized in

Table 1

. The heaviest concentration of patients resided in the Midwest and South, reflecting the geographic distribution of the health plan overall. The second-line erlotinib treatment cohort had a significantly higher percentage of female patients compared with the secondline docetaxel and pemetrexed cohorts (P <.01). There were no significant differences in mean Charlson-Deyo comorbidity scores between the treatment cohorts. The second-line treatment follow-up period was significantly shorter for the pemetrexed cohort than for the erlotinib cohort (P <.01).

Total Healthcare Costs

Total PPPM healthcare costs and costs for specific healthcare services during the second-line treatment follow-up period are shown in

Table 2

. Medical costs accounted for the majority of healthcare costs for all treatment cohorts. Office visit costs were greater for patients who received docetaxel or pemetrexed monotherapy than for those who received erlotinib (Table 2), due in part to inclusion of costs associated with intravenous infusions in medical claims instead of pharmacy claims. The erlotinib cohort had higher pharmacy costs than both the pemetrexed and docetaxel cohorts, but these appear to be offset by lower medical costs (Table 2).

The adjusted estimates for healthcare costs for each treatment cohort and differences in costs between the erlotinib cohort and the docetaxel and pemetrexed cohorts are shown in

Table 3

. Compared with patients receiving

erlotinib, patients who received pemetrexed had 50% higher PPPM total healthcare costs during the second-line treatment follow-up period (Table 3), but total healthcare costs for patients with docetaxel were not significantly different from those for patients with erlotinib (P = .216). Results were robust to various model specifications (

eAppendix B

, available at www.ajpblive.com).

Adverse Event—Associated Costs

Per patient per month healthcare costs associated with AEs during the second-line period are shown in

Table 4

. Compared with patients who received docetaxel or pemetrexed, patients who received erlotinib had lower total AE-related costs.

The adjusted estimates for healthcare costs associated with AEs for each treatment cohort and the difference in costs between patients receiving erlotinib and patients receiving docetaxel or pemetrexed are shown in Table 3. Total AE-associated costs in the second-line setting were more than 3 times greater for patients receiving docetaxel and more than 5 times greater for patients receiving pemetrexed than for patients receiving erlotinib (Table 3). Results were robust to various model specifications (

eAppendix C

, available at www.ajpblive.com). Adverse event—related costs comprised 18% of total healthcare costs for patients on second-line erlotinib and approximately 49% to 62% of total healthcare costs for patients receiving pemetrexed or docetaxel (

Figure

).

Incidence of Adverse Events

The percentage of patients in each cohort with evidence of any of the 12 adverse events is shown in

eAppendix D

, available at www.ajpblive.com. Nausea/emesis was the most common AE in all 3 cohorts during the second-line treatment follow-up period; 100% of the docetaxel cohort (P <.001 vs erlotinib), 97% of the pemetrexed cohort (P <.001 vs erlotinib), and 60% of the erlotinib cohort had claims-based evidence of nausea/ emesis. Consistent with the observed differences in AEassociated costs, patients who received erlotinib monotherapy had the lowest occurrence of AEs overall: 88% of patients on erlotinib had evidence of any AE versus 100% for both docetaxel (P <.001) and pemetrexed (P <.001). In addition to a lower incidence of nausea/emesis, erlotinib was associated with a significantly lower incidence of neutropenia/leukopenia (3%) compared with docetaxel (46%; P <.001) and pemetrexed (22%; P <.001) therapies. A significantly lower percentage of patients in the erlotinib cohort experienced anemia (41%) compared with the percentage of patients in the pemetrexed cohort (74%; P <.001), and dehydration (10%) compared with the percentage of patients in the docetaxel cohort (20%; P = .030).

DISCUSSION

In this study, total healthcare costs were greatest for patients receiving pemetrexed as second-line monotherapy. Pemetrexed-treated patients consistently incurred significantly greater costs than patients receiving erlotinib, whereas monthly costs for patients treated with docetaxel did not differ significantly from costs for those receiving erlotinib. The relative cost ranking of pemetrexed, docetaxel, and erlotinib is consistent with estimates derived previously.^10,29,30

A major strength of this study is that it utilizes population- based data; thus, the findings are relevant to actual treatment settings. Prior cost studies have been based on theoretical models and relied on externally drawn estimates and clinical trial data to calculate costs.^29,30 A budget impact model designed by Ramsey et al³⁰ found that healthcare costs incurred by lung cancer patients after failure of prior therapy for a health plan that included erlotinib on its formulary were similar to or lower than costs for a plan that did not include erlotinib. The model was based on assumptions regarding the incidence of AEs, care required for AEs, and drug dosing as indicated in the prescribing information for each medication. While these assumptions were reasonable given the available data, our analysis incorporates actual claims-based AE incidence, resource use, and cost data to produce a more accurate estimation. Notably, assumptions used in their model regarding incidence of AEs differ from the real-world evidence reflected in the present analysis. For instance, anemia was much more commonly observed in this study than was supposed in their model. During second-line treatment in the present study, 41% to 74% of patients had evidence of anemia, in contrast to the model assumptions of anemia incidence less than 10% for docetaxel and pemetrexed and no impact of anemia assumed for erlotinib treatment based on clinical trial data.³⁰ Carlson et al²⁹ used premarketing data to develop a decision analytic model to evaluate the incremental costs associated with erlotinib, docetaxel, or pemetrexed treatment in a cohort of patients with advanced-stage lung cancer. Their simulation suggested that erlotinib treatment was associated with the lowest total costs, a result supported by the retrospective claims data presented herein.

A previous retrospective data analysis of utilization and costs of second-line therapy for lung cancer patients from the perspective of a health insurer¹⁰ showed that total PPPM reimbursements were highest for patients receiving pemetrexed as second-line therapy and lowest for patients receiving erlotinib, but these investigators did not consider costs associated with AEs. Our results suggest that AE-related costs contribute substantially to total healthcare costs and are a differentiating factor among these 3 therapies. The differences in AE-related costs were reflected in the differences in AE incidence, with evidence of fewer AEs overall for erlotinib and, more specifically, fewer claims associated with neutropenia/leucopenia and nausea/emesis for patients on erlotinib compared with patients on either docetaxel or pemetrexed. Consistently, erlotinib monotherapy also had the lowest healthcare costs associated with AEs.

Study Limitations

Limitations inherent to claims-based analyses must be considered when interpreting these results. Claims data are collected for the purpose of payment, not research, and while they provide insight into real-world treatment patterns, they are subject to possible coding errors. Occurrence of an AE was inferred by coding related to that AE; thus, clinical information was insufficient to determine whether AEs were treatment related, and AEs that are not commonly associated with claims may have been underreported. Rare but potentially costly AEs might also have been missed due to the relatively small sample size or the method used for capturing AE-related utilization. Furthermore, it could not be determined whether AEs that occurred during second-line therapy were attributable to the therapy given during that period, or whether they were residual effects from a previous-line therapy. There may be a bias for greater detection of AE-related healthcare utilization among patients receiving infused therapies because they have more contact with healthcare professionals than patients receiving oral therapies do. In addition, patient expenses for copays or over-the-counter medications for AE management were not captured. Costs associated with AEs were substantial within the second-line setting, but other issues such as disease progression, patient survival outcomes, and costs of first-line treatment, which were not addressed in this analysis of second-line therapy, are also important considerations in the overall care of patients with lung cancer.

In the absence of clinical data, reasons for a change in drug therapy may be unclear, and we were unable to discern therapy changes that represented consolidation or maintenance regimens. In addition, the standard of care has evolved since the 2004-2006 study period. Unobserved factors may have confounded the relationships between treatment groups and costs due to lack of randomization, and clinical characteristics that could impact treatment selection and outcomes (eg, histology, disease stage, performance score, smoking history, race) were not available for evaluation in this analysis. However, adjusting for demographics, comorbidities, first-line chemotherapy (data not shown), and metastasis site revealed that these variables had little biasing influence on costs.

CONCLUSIONS

This study provides increased understanding of costs among lung cancer patients treated with second-line

therapy. Total healthcare costs associated with erlotinib

were significantly lower than costs associated withpemetrexed. Of the 3 therapies, pemetrexed was associatedwith the highest total healthcare costs. Our findingshighlight the economic benefits conferred by erlotinibmonotherapy due to lower costs of AEs and lower overallhealthcare costs.