Real-world Direct Health Care Costs for Metastatic Colorectal Cancer Patients Treated With Cetuximab or Bevacizumab-containing Regimens in First-line or First-line Through Second-line Therapy

Accepted Manuscript Real-world direct healthcare costs for metastatic colorectal cancer patients treated with cetuximab or bevacizumab-containing regimens in first-line or first- through second-line therapy Stephen Johnston, MS, Kathleen Wilson, MPH, Helen Varker, BS, Elisabetta Malangone-Monaco, MS, Paul Juneau, MS, Ellen Riehle, MPH, Sacha SatramHoang, PhD, Nicolas Sommer, PhD, Sarika Ogale, PhD PII:

S1533-0028(16)30171-2

DOI:

10.1016/j.clcc.2017.03.014

Reference:

CLCC 371

To appear in:

Clinical Colorectal Cancer

Received Date: 30 September 2016 Accepted Date: 16 March 2017

Please cite this article as: Johnston S, Wilson K, Varker H, Malangone-Monaco E, Juneau P, Riehle E, Satram-Hoang S, Sommer N, Ogale S, Real-world direct healthcare costs for metastatic colorectal cancer patients treated with cetuximab or bevacizumab-containing regimens in first-line or first- through second-line therapy, Clinical Colorectal Cancer (2017), doi: 10.1016/j.clcc.2017.03.014. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Title Page

Title: Real-world direct healthcare costs for metastatic colorectal cancer patients treated with cetuximab

RI PT

or bevacizumab-containing regimens in first-line or first- through second-line therapy

Short Title: Cost of treating colorectal cancer patients

1

1

SC

Author names and affiliations: Stephen Johnston, MS [email protected]; Kathleen Wilson, 1

MPH [email protected]; Helen Varker, BS [email protected]; Elisabetta 1

1

M AN U

Malangone-Monaco, MS [email protected]; Paul Juneau, MS 1

[email protected]; Ellen Riehle, MPH [email protected]; Sacha Satram2

3

Hoang, PhD [email protected]; Nicolas Sommer, PhD [email protected]; Sarika 3

Ogale, PhD [email protected]

Truven Health Analytics Inc., 7700 Old Georgetown Rd, Ste 650, Bethesda, MD 20814, United States

2

Q.D. Research, Inc., 8777 Auburn Folsom Rd, Ste C501, Granite Bay, CA

3

Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080

TE D

1

EP

Corresponding author: Elisabetta Malangone-Monaco; [email protected]; Truven Health Analytics Inc., 7700 Old Georgetown Rd, Ste 650, Bethesda, MD 20814, United States; Phone

AC C

203-980-4054

1

ACCEPTED MANUSCRIPT

Conflict of Interest Page

Kathleen Wilson, Elisabetta Malangone-Monaco, Ellen Riehle, Helen Varker, and Paul Juneau are

RI PT

employees of Truven Health Analytics and Stephen Johnston is a former employee of Truven Health Analytics. Truven Heath Analytics was paid by Genentech to conduct this study. Sacha Satram-Hoang is a paid consultant to Genentech. Nicolas Sommer and Sarika Ogale are employees and stockholders of

AC C

EP

TE D

M AN U

SC

Genentech.

2

ACCEPTED MANUSCRIPT

MicroAbstract (current count = 60 words; limit = 60)

Cost considerations may factor into the choice of metastatic colorectal cancer (mCRC) treatments. This

RI PT

real-world observational study of 2,352 mCRC patients in the United States found that per-patient monthly healthcare costs in first-line or first- through second-line therapy were substantially higher for patients treated with first-line cetuximab-containing versus bevacizumab-containing regimens. Such cost

AC C

EP

TE D

M AN U

SC

implications may be meaningful in real-world clinical practice.

3

ACCEPTED MANUSCRIPT

Abstract (current count = 299 words; limit = 300) Purpose: This study examined real-world direct healthcare costs for metastatic colorectal cancer (mCRC) patients initiating a first-line (1L) bevacizumab (BEV)- or cetuximab (CET)-containing regimen in 1L or 1L-

RI PT

through-second-line (1L-2L) therapy. Patients and methods: Using a large U.S. insurance claims database, patients with mCRC initiating a 1L BEV- or 1L CET-containing regimen between 1/1/2008-9/30/2014 were identified. Per-patient per-month (PPPM) all cause healthcare costs (2014 US dollars) were measured during 1L therapy and, for patients

SC

continuing to a 2L biologic-containing regimen, during 1L-2L therapy. Multivariable regressions were used to compare PPPM total healthcare costs between patients initiating a 1L BEV- vs. 1L CET-containing

M AN U

regimen.

Results: A total of 6,095 patients initiating a 1L BEV- and 453 initiating a 1L CET-containing regimen were evaluated for 1L costs; 2,218 patients initiating a 1L BEV- and 134 initiating a 1L CET-containing regimen were evaluated for 1L-2L costs. In 1L therapy, using patients initiating a 1L BEV-containing regimen as reference, 1L CET had adjusted PPPM costs that were $3,135 (95% CI=$1,174-$5,040, P<0.001) higher on average. In 1L-2L therapy, using patients moving from 1L BEV-2L BEV as reference,

TE D

1L BEV-2L CET had adjusted PPPM costs that were $1,402 (95% CI=$1,365-$1,442, P=0.010) higher and 1L CET-2L BEV had adjusted PPPM costs that were $4,279 (95% CI=$4,167-$4,400, P=0.001) higher on average; adjusted PPPM cost differences for 1L BEV-2L other biologic or 1L CET-2L other

Conclusion:

EP

biologic were numerically higher but statistically insignificant.

AC C

PPPM total healthcare costs through 1L and 2L therapy tended to be higher for patients treated with 1L CET-containing regimens compared with 1L BEV-containing regimens and continuing treatment with BEV-containing regimens 1L though 2L was shown to be less costly than switching between BEV and CET. The cost differences between BEV and CET hold important implications for treatment decisions of mCRC patients in real-world clinical practice.

Keywords bevacizumab; biologics; cetuximab; healthcare cost; metastatic colorectal cancer

4

ACCEPTED MANUSCRIPT

Abbreviations 1L, first line 2L, second line

RI PT

BEV, bevacizumab CET, cetuximab CI, confidence interval CR, cost ratio

SC

CRC, colorectal cancer ED, emergency department

mCRC, metastatic colorectal cancer NDC, National Drug Code PAN, panitumumab PPPM, per-patient per-month

SD, standard deviation USD, United States dollars

AC C

EP

ZAF, ziv-aflibercept

TE D

REG, regorafenib

M AN U

ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification

5

ACCEPTED MANUSCRIPT

Introduction Colorectal cancer (CRC) is the third most commonly diagnosed cancer among men and women 1

in the United States. Approximately 20% of patients have metastatic colorectal cancer (mCRC) at the

RI PT

time of diagnosis, with more progressing to mCRC over time.1 Treatment options for mCRC have been increasing in part due to advances in new chemotherapy drugs and expanded indications for biologic agents. In the United States, five biologics are currently approved by the US Food and Drug Administration for first or later-line treatment of mCRC, including

(ZAF).

2,3,4,5,6

SC

bevacizumab (BEV), cetuximab (CET), panitumumab (PAN), regorafenib (REG), and ziv-aflibercept Of these biologics, BEV and CET have been indicated for the treatment of mCRC for the

M AN U

longest duration of time and are currently the most commonly used for the treatment of mCRC. BEV is a vascular endothelial growth factor (VEGF)-specific angiogenesis inhibitor that is indicated for first-line (1L) or second-line (2L) treatment of mCRC with intravenous 5-fluorouracil–based chemotherapy and for 2L treatment in patients who have progressed on a first-line BEV-containing 2

regimen. CET is an epidermal growth factor receptor (EGFR) antagonist indicated for treatment of RAS

TE D

wild-type, EGFR-expressing, mCRC in 1L with concomitant FOLFIRI and as single agent or in combination with irinotecan in refractory patients.

3

The CALGB 80405 trial, the largest Phase III trial designed to compare overall survival of chemotherapy + BEV regimens with chemotherapy + CET regimens for the 1L treatment of KRAS wild-

7

EP

type mCRC, demonstrated no difference in efficacy between BEV and CET in 1L KRAS wild-type patients. Aside from efficacy, other factors which may influence the choice between BEV and CET for 1L

AC C

treatment of mCRC may include tolerability, right-sided versus left-sided tumor location, physician and patient preferences, specific treatment goals, and cancer regimen costs. An economic analysis of CALGB 80405 showed that chemotherapy + BEV regimens cost less and achieved similar survival and quality adjusted survival as chemotherapy + CET regimens for 1L treatment of KRAS wild-type mCRC, but there is currently little published information regarding the real-world comparative costs of BEV and CET. Two prior observational studies compared healthcare costs between mCRC patients initiating BEV- vs. CETcontaining regimens in 2L and reported that patients initiating CET-containing regimens had monthly total 8,9

healthcare costs which were over $2,000 greater.

However, there is currently no published information 6

ACCEPTED MANUSCRIPT

comparing real world healthcare costs between BEV- vs. CET-containing regimens in 1L, nor is there information on how costs compare between these regimens through the 2L. Thus, this study addressed two objectives: 1) to compare healthcare costs during 1L therapy

RI PT

between mCRC patients treated with a 1L BEV-containing or 1L CET-containing regimen, henceforth referred to as the 1L comparison; 2) to compare healthcare costs during 1L through 2L therapy between mCRC patients treated with a 1L BEV-containing or 1L CET-containing regimen and continuing to a 2L biologic-containing regimen (BEV, CET, PAN, REG, or ZAF), henceforth referred to as the treatment

SC

sequencing comparison.

M AN U

Patients and Methods Data Source

This was a retrospective, observational cohort study based on U.S. administrative insurance claims data for the time period spanning July 1, 2007 through September 30, 2014. The data were extracted from the Truven Health MarketScan® Commercial and Medicare Supplemental databases (Ann Arbor, MI). These databases include inpatient medical, outpatient medical, and outpatient pharmacy

TE D

claims data, as well as insurance enrollment and demographic information collected from more than 300 large self-insured U.S. employers and more than 25 U.S. health plans. The Commercial database includes information for individuals with employer-sponsored health

EP

insurance who are under the age of 65 years, including the primary insured as well as spouses or dependents covered under the primary insured’s plan. The Medicare Supplemental database includes

AC C

information for individuals who are Medicare-eligible (most of whom are aged 65 years or older) and have both traditional Medicare coverage plus a Medicare supplemental insurance plan. The Medicare Supplemental database includes both the Medicare-paid and supplemental-paid components of reimbursed insurance claims. During the time period of study, the databases included insurance claims for over 100 million unique individuals. The study databases satisfy Sections 164.514 (a)-(b)1ii of the Health Insurance Portability and Accountability Act of 1996 privacy rule (HIPAA) regarding the determination and documentation of statistically de-identified data. This study did not involve the

7

ACCEPTED MANUSCRIPT

collection, use, or transmittal of individually identifiable data, and therefore Institutional Review Board review or approval was not required.

RI PT

Patient Selection Criteria Figure 1 depicts key elements of the study design and time periods that are relevant to the

process of patient selection. Patients meeting all of the following study eligibility criteria were included in the 1L comparison: had ≥1 non-diagnostic medical claim (i.e., excluding medical claims that may be

SC

associated with a diagnostic workup to rule out a given condition) with an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis code for colorectal cancer (ICD-9-

M AN U

CM 153.xx, 154.0x, 154.1x, 154.8x) between January 1, 2008, and September 30, 2014; had ≥1 nondiagnostic medical claim with an ICD-9-CM code for metastasis (ICD-9-CM 196.0x–196.1x, 196.3x– 196.5x, 196.8x, 197.0x–197.3x, 197.7x–198.xx) between January 1, 2008, and September 30, 2014 (date of first observed metastasis claim for each patient was designated as the patient’s metastasis date [Figure 1]); aged 18 years or older on metastasis date; had a 1L treatment regimen containing BEV or CET after the metastasis date (start of 1L therapy was designated as the 1L index date [Figure 1]); had 6

TE D

months of continuous enrollment in medical and pharmacy benefits immediately before the metastasis date through 1 month after 1L index date; and had no evidence of multiple biologics in the 1L of therapy. Patients were excluded if they had non-diagnostic medical claims with an ICD-9-CM diagnosis code for

EP

metastasis in the 6 months immediately before the metastasis date; had non-diagnostic medical claims with an ICD-9-CM diagnosis code for a non-CRC primary cancer in which BEV, CET, or REG are

AC C

indicated (ICD-9-CM 140.xx–146.xx, 148.xx–149.xx, 160.xx–161.xx, 162.2x–162.9x, 180.xx, 189.0x, 191.xx) in the 6 months immediately before the metastasis date; and had claims for any biologics (BEV, CET, PAN, REG, or ZAF) prior to the metastasis date; Patients meeting the following additional study eligibility criteria were included in the treatment

sequencing comparison: had a 2L treatment regimen containing BEV, CET, PAN, REG, or ZAF (start of 2L therapy was designated as the 2L index date [Figure 1]); had continuous enrollment in medical and pharmacy benefits from the metastasis date through 1 month after the 2L index date; and had no evidence of multiple biologics in the 2L of therapy.

8

ACCEPTED MANUSCRIPT

Lines of Therapy Construction and Follow-up This study’s algorithm for construction of lines of therapy was adapted from a validated 8

administrative claims-based algorithm. As shown in Figure 1, the first claim for a CRC treatment agent

RI PT

occurring after the metastasis date was designated the 1L index date. CRC treatments included chemotherapies (5-fluorouracil/capecitabine, oxaliplatin, irinotecan) and biologics (BEV, CET, PAN, REG, or ZAF). The 1L therapy regimen included all CRC treatment agents added within 28 days of the 1L index date, with the exception that BEV and ZAF could be added 28–120 days after the 1L index date. Addition

SC

of BEV and ZAF up 28-120 after the 1L index date was allowed to account for treatment delay or

treatment suspension that may have been the result of major surgery performed at the time of metastasis.

M AN U

The duration of the 1L of therapy began on the 1L index date and ended at the date of the earliest occurrence of: (1) appearance of a new agent not included in the first-line therapy regimen (which begins a 2L of therapy); (2) a gap of >120 days for all agents in the first-line therapy regimen (end date was set to last day with any line of therapy agent on hand, based on days supplied for oral agents and recommended treatment intervals for intravenous agents); (3) inpatient death; (4) disenrollment from

TE D

health insurance; or (5) end of study period (September 30, 2014).

For patients who initiated a 2L of therapy, the same rules for 1L of therapy construction and the duration of the 1L of therapy applied to the 2L of therapy.

EP

Outcome and Covariates

All study variables were measured based on insurance enrollment records or inpatient medical,

AC C

outpatient medical, and outpatient pharmaceutical claims data using ICD-9-CM diagnosis and procedure codes, Current Procedural Terminology® codes, Healthcare Common Procedure Coding System codes, and National Drug Codes (NDCs), as appropriate. The study outcomes were healthcare utilization and costs, measured during the lines of therapy

and expressed in per-patient per-month (PPPM) units to account for the variable-length of lines of therapy. Healthcare utilization and costs were categorized as follows: inpatient medical, outpatient medical (sub-divided by emergency department, office visits, and other outpatient), outpatient pharmacy excluding the cancer regimen, and cancer regimen costs (chemotherapy and biologic drug and

9

ACCEPTED MANUSCRIPT

administration costs). Healthcare costs were adjusted to 2014 U.S. dollars using the Medical Care component of the Consumer Price Index

10

and were measured using the financial fields on administrative

claims in the MarketScan Databases, and included: gross covered payments for all healthcare services or

RI PT

products; i.e., the amount eligible for payment after applying pricing guidelines such as fee schedules and discounts, but including deductibles, copayments, and coordination of benefits. Costs for services

provided under capitated payment arrangements were imputed using a Truven Health Analytics algorithm that computes a payment “proxy” for health care services used, based on the average payments for

SC

noncapitated claims at the region, year, and procedure level within the MarketScan databases. This algorithm has been used in the published health economic literature for multiple disease areas.

11

M AN U

Study covariates, including patients’ demographics and baseline clinical characteristics, were measured to describe the study sample and used for adjustment in the multivariable analyses as described below. Demographics were measured as of the index date using enrollment data and included: age; sex; United States Census Bureau geographic region of residence; health plan type; year of first-line index date. Unless otherwise noted, clinical characteristics were measured during the 6-month period immediately before the 1L index date using medical and prescription claims data and included: the

TE D

National Cancer Institute Comorbidity Index, which is an aggregate measure of comorbidity expressed as a numeric score based on the presence of select diagnoses for various conditions, excluding diagnoses 12

for cancer, each with specific weights ranging from 1 to 6 points ; the number of unique NDCs and the

EP

number of unique ICD-9-CM diagnosis codes at the 3-digit level of specificity, both of which have been 13

shown to be predictive of future healthcare costs ; total healthcare costs; site of metastasis; resection

AC C

surgery; radiation treatment; and chemotherapy.

10

ACCEPTED MANUSCRIPT

Statistical Analyses Descriptive statistics were used to summarize all study variables. In the 1L comparison, descriptive statistics were stratified by 1L BEV-containing versus 1L CET-containing regimens. In the

RI PT

treatment sequencing comparison, descriptive statistics were stratified by the following treatment sequences: 1L BEV–2L BEV; 1L BEV–2L CET; 1L BEV–2L PAN; 1L BEV–2L other biologic (i.e., REG or ZAF); 1L CET–2L BEV; 1L CET–2L CET; 1L CET–2L other biologic (i.e., PAN, REG, or ZAF). Categorical variables were summarized by providing the percentage of patients in each treatment category;

SC

continuous variables were summarized by providing the mean and standard deviation.

Multivariable ordinary least squares regression was used to compare log-transformed monthly

M AN U

total healthcare costs across the comparison groups, adjusting for all demographic and baseline clinical characteristics. Because some stratifications in the treatment sequencing comparison had very few patients, two versions of the outcome model were fitted: (1) one in which 1L CET-2L other biologic and 1L CET-2L CET were collapsed into a single category and 1L BEV-2L PAN and 1L BEV-2L other biologic were collapsed into a single category; (2) one in which all treatment sequences with 1L CET-containing regimens were collapsed into a single category. Recycled prediction was used to generate adjusted

TE D

predicted healthcare cost differences between the comparison groups. Five hundred bootstrap samples were used to calculate the 95% simultaneous confidence intervals around the predicted cost differences. A P≤0.05 was considered the threshold for statistical significance and statistical significance testing was

AC C

Results

EP

undertaken only in the context of the multivariable analyses.

Patients

After application of the study eligibility criteria, a total of 6,095 patients treated with 1L BEV-

containing regimens and 453 patients treated with 1L CET-containing regimens were included in the 1L comparison; 2,218 patients treated with 1L BEV-containing regimens and 134 patients treated with 1L CET-containing regimens were included in the treatment sequencing comparison. Supplemental Table S1 shows the sample attrition associated with the application of each study eligibility criterion and the sample sizes of the individual treatment sequences examined in the treatment sequencing comparison.

11

ACCEPTED MANUSCRIPT

In the 1L comparison, patients treated with 1L BEV-containing regimens were slightly younger, had a greater proportion of females, and had mean values of administrative claims-based indices of comorbidity and health status that were indicative of slightly better health status when compared with

RI PT

patients treated with 1L CET-containing regimens (Table 1). A similar general pattern was observed in the treatment sequencing comparison, though the differences in the indices of comorbidity and health status were attenuated (Table 1).

In both the 1L and treatment sequencing comparisons, BEV-containing regimens were more

SC

likely to contain oxaliplatin and less likely to contain irinotecan when compared with CET-containing regimens, in their respective lines of therapy (Tables 2 and 3). In the treatment sequencing comparison,

M AN U

the majority (72%) of patients with 1L BEV also received BEV in 2L. In the 1L comparison, patients were followed on 1L therapy for a median of 179 days for 1L BEV and 119 days for 1L CET. In the treatment sequencing comparison, patients were followed on 1L + 2L therapy for a median number of days ranging from 216 days in 1L CET-2L 'other' biologic to 363 days for 1L BEV-2L PAN.

Cancer Regimen Costs

TE D

In the 1L comparison, unadjusted PPPM cancer regimen costs were higher in 1L CET-containing regimens compared with 1L BEV-containing regimens (Figure 2, left panel). This pattern was maintained in the treatment sequencing comparison, with all 1L CET-containing treatment sequences having higher

EP

PPPM cancer regimen costs than 1L BEV-containing treatment sequences (Figure 2, right panel). In both comparisons, cancer regimen costs were primarily driven by the biologic drug costs, and the higher

AC C

biologic drug costs for CET compared to BEV was the primary driver of differences across the regimens and sequences

Total Healthcare Costs

Cancer regimen costs accounted for 54%-66% of total costs across the regimens, the largest

single component of total costs, followed by outpatient medical costs (24%-34%), inpatient medical costs (4%-15%), and outpatient pharmacy costs (1%-2%) (Table 4). In the 1L comparison, unadjusted PPPM total healthcare costs were higher in 1L CET-containing regimens compared with 1L BEV-containing

12

ACCEPTED MANUSCRIPT

regimens (Table 4). This pattern was maintained in the treatment sequencing comparison, with all 1L CET-containing treatment sequences having higher PPPM total healthcare costs than 1L BEV-containing treatment sequences (Table 4).

RI PT

In the multivariable analyses of the 1L comparison, PPPM total healthcare costs for were 15% higher (Cost Ratio [CR] = 1.15, 95% Confidence Interval [CI] = 1.09-1.22, P < 0.001) in 1L CET-

containing regimens compared with 1L BEV-containing regimens, equating to an excess cost of $3,135 (95% CI = $1,174-$5,040) per month (Figure 3, left panel). In multivariable analyses of the treatment

SC

sequencing comparison, in which 1L BEV-1L BEV was treated as the reference category, PPPM total healthcare costs were 7% higher (CR = 1.07, 95% CI = 1.02-1.13, P = 0.010) in 1L BEV-2L CET,

M AN U

equating to an excess cost of $1,402 (95% CI = $1,365-$1,442) per month, and 21% higher (CR = 1.21, 95% CI = 1.08-1.37, P = 0.001) in 1L BEV-2L CET, equating to an excess cost of $4,279 ($4,167-$4,400) (Figure 3, right panel) per month. PPPM total healthcare costs were higher for 1L BEV-2L other biologic and 1L CET-2L other biologic compared to 1L BEV-2L BEV, but the differences were not statistically significant. In a sensitivity analysis of the treatment sequencing comparison in which all 1L CETcontaining treatment sequences were compared with all 1L BEV-containing treatment sequences, PPPM

TE D

total healthcare costs were 14% higher (CR = 1.14, 95% CI = 1.05-1.24, P= 0.002), equating to an excess cost of $2,736 (95% CI = $953-$4,573) (data not shown).

EP

Discussion

To our knowledge, this is the first study to compare real world healthcare costs between BEV- vs.

AC C

CET-containing regimens in 1L and in 1L through 2L. As noted above, two prior observational studies compared healthcare costs between mCRC patients initiating BEV- vs. CET-containing regimens in 2L and reported that patients initiating CET-containing regimens had per-patient monthly healthcare costs which were over $2,000 greater.

8,9

This is generally consistent with the present study, which found that 1L

PPPM total healthcare costs were $3,135 higher for patients initiating 1L CET compared to 1L BEV. In both of those analyses as well as in the present analysis, a major driver of the differences in costs was the greater biologic drug costs for the CET groups.

13

ACCEPTED MANUSCRIPT

There are presently no real world observational data to which the present study’s 1L comparison results can be compared. A cost-minimization economic analysis of CALGB/SWOG 80405—which has thus far been published only in the form of an abstract—suggested that 1L BEV is preferable to 1L CET 14

Thus, the real-world

RI PT

from an economic standpoint due to differences in the cost of the cancer regimens.

evidence provided by the present study corroborates the collective findings generated from this clinical trial based economic modeling study.

Similar to other real-world observational studies evaluating the sequence of 1L through 2L

SC

treatments for mCRC, this study found the majority of patients initiating 1L BEV continued on a 2L BEVcontaining regimen, while fewer patients switched to a 2L CET-containing regimen or another targeted 15,16

Fewer patients were initiating 1L CET, which previous studies have suggested may be due,

M AN U

therapy.

in part, to later approval and slower uptake of CET as first line therapy.

15,16

This study found that 1L CET

patients were equally as likely to continue on a 2L CET-containing regimen as they were to switch to a 2L BEV-containing regimen.

As patients progressed from 1L through 2L treatment sequences, the differences in costs observed between 1L CET-containing regimens and 1L BEV-containing containing regimens persisted

TE D

through 2L. Adjusted PPPM total healthcare costs through 1L and 2L were $3,135 higher for patients initiating 1L CET-containing regimens compared to 1L BEV-2L BEV. Switching treatment regimens to a different 2L therapy corresponded to higher PPPM total healthcare costs 1L through 2L. Patients initiating

EP

1L BEV-containing regimens and proceeding onto a 2L CET-containing regimens had $1,402 higher PPPM healthcare costs compared to 1L BEV-2L BEV. Patients initiating 1L CET-containing regimens and

AC C

proceeding onto a 2L BEV-containing regimens had $4,279 higher PPPM healthcare costs compared to patients with 1L BEV-2L BEV. There were smaller differences in costs for patients switching to newer 2L treatments such as other EGFR antagonists or other targeted therapies; however, these sample sizes were limited due to the later approvals for these treatments for mCRC. Future research may be needed to investigate the long-term impact of newer treatments on differences in healthcare costs for patients with mCRC. This study is subject to limitations. First, because RAS testing results were not available in this dataset, the study population could not be restricted to RAS wild-type patients. As a consequence of the

14

ACCEPTED MANUSCRIPT

label restriction of CET to KRAS wild-type patients, it can be assumed that patients initiating 1L CET would have been predominantly KRAS wild-type, while patients initiating 1L BEV would have been a mixture of KRAS genotypes. Since KRAS wild-type patients have better prognosis and more treatment

RI PT

options, the results could be biased by the over-representation of KRAS wild-type patients in the CET lines of therapy. Second, our study period extended back to a time before CET was approved for 1L treatment of mCRC; a decision which was made to maximize sample size based on the knowledge that in real world populations, treatment patterns may not always follow labeled instructions. In a post-hoc

SC

sensitivity analysis, we analyzed the 1L comparison results when restricting the study population to

individuals initiating 1L therapy after the approval of CET for 1L treatment of mCRC and found results that

M AN U

were consistent with the primary analyses (data not shown). Third, earlier approval of BEV for 1L treatment of mCRC, KRAS genotype and other factors, such as physician prescribing preferences and biologic drug costs, may have contributed to differences in the number of patients initiating 1L BEV- vs. CET-containing regimens. Despite the differences in sample size, the demographic and baseline clinical characteristics were similar for both groups, suggesting these groups are largely representative of the general population of patients with mCRC initiating 1L therapy. Fourth, administrative claims data are not

TE D

collected for research purposes and are subject to coding inaccuracies that may introduce measurement error into variables. Fifth, there is potential for residual confounding even after multivariable adjustment as clinical information on performance status and stage are not available in this dataset. Finally, the present

EP

study results were derived from individuals with commercial or Medicare plus supplemental health insurance and may not be generalizable to the entire U.S. population, including individuals with only

AC C

traditional Medicare coverage, with Medicaid coverage, or the uninsured. Conclusion

This study found mCRC cancer regimens which contained bevacizumab in the first line of therapy

were associated with lower cancer regimen and total healthcare costs than those which contained cetuximab. These cost differences persisted into the second line of therapy. The cost differences between bevacizumab and cetuximab hold important implications for treatment decisions of mCRC patients in realworld clinical practice.

15

ACCEPTED MANUSCRIPT

Clinical Practice Points What is already known about this subject? Two prior observational studies compared healthcare costs between mCRC patients initiating

RI PT

•

bevacizumab- vs. cetuximab-containing regimens in the second line of therapy and reported that patients initiating cetuximab-containing regimens had healthcare costs which were over $2,000 greater. To our knowledge, this is the first study to compare real-world healthcare costs between

SC

bevacizumab vs. cetuximab-containing regimens in first line and in first line through the second

M AN U

line.

What are the new findings? •

This study found mCRC cancer regimens which contained bevacizumab in the first line of therapy were associated with lower cancer regimen and total healthcare costs than those which contained cetuximab. Continuing treatment with bevacizumab-containing regimens through the second line

TE D

was less costly than switching between bevacizumab and cetuximab.

How might it impact on clinical practice in the foreseeable future? The cost differences between bevacizumab and cetuximab hold important implications for

EP

treatment decisions of mCRC patients in real-world clinical practice.

AC C

•

16

ACCEPTED MANUSCRIPT

Acknowledgments Stephen Johnston, Kathleen Wilson, Helen Varker, Elisabetta Malangone-Monaco, Paul Juneau, Ellen Riehle, Sacha Satram-Hoang, Nicolas Sommer, and Sarika Ogale all made the following contributions: 1)

RI PT

substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data; 2) drafting the manuscript and revising it critically for important intellectual content; and 3) final approval of the version to be published. The authors wish to acknowledge Diana Stetsovsky for

assistance with statistical programming. Stephen S. Johnston is the guarantor for this article and takes

SC

full responsibility for the integrity of the work as a whole. Sponsorship for this study was funded by

AC C

EP

TE D

M AN U

Genentech Inc.

17

ACCEPTED MANUSCRIPT

References 1

Noone AM et al. Seer Cancer Statistics Review Based on November 2013 Seer Data Submission

(National Cancer Institute, 2013). Avastin [package insert]. South San Francisco, CA: Genentech, Inc.; 2015.

3

Erbitux [package insert]. Indianapolis, IN: Eli Lilly & Co; 2015.

4

Vectibix [package insert]. Thousand Oaks, CA: Amgen Inc.; 2015.

5

Stivarga [package insert]. Wayne, NJ: Bayer HealthCare Pharmaceuticals Inc.; 2015.

6

Zaltrap [package insert]. Bridgewater, NJ: sanofi-aventis U.S. LLC; 2015.

7

Venook AP, Niedzwiecki D, Lenz HJ, et al. CALGB/SWOG 80405: Phase III trial of irinotecan/5-

SC

RI PT

2

M AN U

FU/leucovorin (FOLFIRI) or oxaliplatin/5-FU/leucovorin (mFOLFOX6) with bevacizumab (BV) or cetuximab (CET) for patients (pts) with KRAS wild-type (wt) untreated metastatic adenocarcinoma of the colon or rectum (MCRC). Presented at the 2014 American Society of Clinical Oncology (ASCO) Annual Meeting; Abstract LBA3. 8

Dacosta Byfield S, Yu E, Morlock R, Evans D, Teitelbaum A. Corroboration of claims algorithm for

TE D

second-line costs of metastatic colorectal cancer treatment with targeted agents. J Med Econ. 2013 Aug;16(8):1071-81. 9

Yang H1, Yu AP, Wu EQ, Yim YM, Yu E. Healthcare costs associated with bevacizumab and cetuximab

in second-line treatment of metastatic colorectal cancer. J Med Econ. 2011;14(5):542-52. United States Bureau of Labor Statistics. Consumer Price Index. Available at: http://www.bls.gov/cpi/.

EP

10

Accessed June 1, 2015.

Truven Health Analytics. MarketScan Bibliography. Available at:

AC C

11

http://sites.truvenhealth.com/bibliography/2014TruvenHealthMarketScanBibliography.pdf Accessed June 1, 2015 12

National Cancer Institute, Division of Cancer Control & Population Sciences. SEER-Medicare:

calculation of comorbidity weights. Available at: http://healthcaredelivery.cancer.gov/seermedicare/program/comorbidity.html. Accessed June 1, 2015.

18

ACCEPTED MANUSCRIPT

13

Johnston SS. (Podium Presentation) Comparative performance of risk adjustment measures in a

sample of commercially-insured patients under age 65—Two simple measures outperform current standards. Value Health 2010; 13(3): A4. Fowler R. Schrag D, Dueck AC, Naughton MJ, et al. Cost of chemotherapy for metastatic colorectal cancer with

RI PT

14

either bevacizumab or cetuximab: Economic analysis of CALGB/SWOG 80405. J Clin Oncol 33, 2015 (suppl; abstr 6504). 15

Hess GP, Wang PF, Quach D, Barber B, Zhao Z. Systemic Therapy for Metastatic Colorectal Cancer:

2010 Nov;6(6):301-7. doi:10.1200/JOP.2010.000072.

Abrams TA, Meyer G, Schrag D, Meyerhardt JA, Moloney J, Fuchs CS. Chemotherapy usage patterns

M AN U

16

SC

Patterns of Chemotherapy and Biologic Therapy Use in US Medical Oncology Practice. J Oncol Pract.

in a US-wide cohort of patients with metastatic colorectal cancer. J Natl Cancer Inst. 2014

AC C

EP

TE D

Feb;106(2):djt371. doi: 10.1093/jnci/djt371.

19

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

20

ACCEPTED MANUSCRIPT

Tables 1

1L Comparison

Treatment Sequencing Comparison

1L CET

1L BEV2L BEV

1L BEV2L CET

1L BEV2L PAN

1L BEV2 2L Other

1L CET2L BEV

1L CET2L CET

1L CET3 2L Other

N = 6,095

N = 453

N = 1,606

N = 391

N = 108

N = 113

N = 63

N = 54

N = 17

58.8 (11.7)

60.3 (12.4)

57.7 (11.2)

58.0 (11.8)

54.4%

58.7%

46.5%

40.7%

Insurance plan type, % 14.0%

14.8%

14.5%

Managed Care

76.4%

72.0%

77.0%

Other

5.5%

4.6%

4.5%

Unknown

4.2%

8.6%

4.0%

27.2%

33.6%

23.3%

2008

6.5%

6.6%

2009

18.2%

17.4%

2010

16.9%

17.9%

2011

16.8%

12.6%

2012

17.5%

2013

14.6%

2014

9.5%

14.4%

Medicare payer, %

59.8 (12.8)

60.6 (11.8)

38.1%

51.9%

29.4%

19.4%

14.2%

15.9%

11.1%

11.8%

76.9%

76.1%

71.4%

72.2%

76.5%

6.4%

1.9%

8.8%

6.3%

1.9%

0.0%

5.9%

1.9%

0.9%

6.3%

14.8%

11.8%

27.1%

34.3%

26.5%

28.6%

35.2%

41.2%

7.7%

7.9%

4.6%

0.0%

6.3%

1.9%

5.9%

20.4%

19.4%

18.5%

0.9%

19.0%

18.5%

5.9%

20.6%

20.2%

22.2%

0.9%

25.4%

22.2%

23.5%

19.8%

17.9%

25.0%

20.4%

11.1%

14.8%

5.9%

19.9%

16.2%

18.7%

13.9%

33.6%

14.3%

29.6%

47.1%

17.9%

12.6%

14.8%

13.0%

38.1%

20.6%

7.4%

11.8%

7.7%

2.7%

1.0%

2.8%

6.2%

3.2%

5.6%

0.0%

21.2%

13.6%

15.9%

10.2%

8.0%

12.7%

29.6%

23.5%

EP

AC C

Northeast

59.0 (11.3)

46.0%

14.6%

Index year, %

Geographic region, %

58.5 (10.9)

47.2%

73.1%

TE D

Fee for service

60.3 (13.0)

M AN U

Female, %

SC

1L BEV

Demographics Age, Mean (SD)

RI PT

Table 1. Patient Characteristics

North Central

29.0%

30.7%

28.3%

32.5%

37.0%

21.2%

30.2%

25.9%

35.3%

South

40.2%

27.8%

41.2%

32.2%

39.8%

58.4%

28.6%

25.9%

29.4%

West

14.8%

19.2%

15.7%

18.4%

12.0%

11.5%

28.6%

16.7%

5.9%

1.6%

1.1%

1.2%

1.0%

0.9%

0.9%

0.0%

1.9%

5.9%

Unknown

21

ACCEPTED MANUSCRIPT

Baseline clinical characteristics 0.64 (1.12)

0.48 (0.84)

0.50 (0.86)

0.47 (0.75)

0.48 (0.81)

0.56 (0.80)

16.5 (7.2)

16.9 (8.8)

16.0 (6.7)

15.4 (7.1)

16.0 (6.0)

16.5 (7.2)

8.8 (6.0) 46,957 (44,514) 36,095

9.5 (7.1) 55,682 (64,834) 39,473

8.5 (5.7) 45,388 (42,283) 35,715

8.5 (5.6) 45,317 (42,716) 34,037

8.8 (5.6) 41,951 (39,404) 31,671

9.0 (5.5) 47,212 (47,534) 41,013

Liver

67.1%

59.6%

71.4%

69.1%

66.7%

Lung

20.4%

18.3%

19.9%

20.2%

Bone

6.1%

5.7%

4.6%

5.1%

35.0%

23.2%

40.7%

7.5%

15.0%

6.5%

Fluorouracil/Capecitabine

7.0%

8.8%

6.1%

Irinotecan

0.7%

2.2%

0.4%

Oxaliplatin

3.6%

3.5%

2.9%

Unique 3-digit ICD-9-CM Dx Unique NDCs Healthcare expenditures, 2014 USD Median

0.48 (0.79)

0.53 (0.72)

15.2 (6.5)

17.1 (7.9)

13.8 (6.8)

8.3 (5.2) 51,233 (53,397) 38,089

10.1 (7.1) 51,748 (66,295) 39,684

6.2 (5.1) 36,936 (28,831) 31,271

67.3%

71.4%

63.0%

58.8%

22.2%

23.0%

20.6%

16.7%

17.6%

9.3%

6.2%

3.2%

7.4%

0.0%

30.2%

27.8%

35.4%

23.8%

31.5%

5.9%

5.4%

3.7%

2.7%

11.1%

13.0%

17.6%

9.7%

10.2%

11.5%

9.5%

5.6%

17.6%

1.0%

3.7%

1.8%

4.8%

0.0%

0.0%

6.6%

6.5%

6.2%

4.8%

1.9%

11.8%

Sites of metastasis, %

Resection surgery, % Radiation treatment, %

TE D

Chemotherapy treatment, %

SC

NCI Index

RI PT

0.52 (0.91)

M AN U

Health indices, Mean (SD)

1

EP

1L, first line; 2L, second line; BEV, bevacizumab-containing regimen; CET, cetuximab-containing regimen; ICD-9-CM, International Classification th of Diseases, 9 Revision, Clinical Modification; NCI, National Cancer Institute; NDC, National Drug Code; PAN, panitumumab-containing regimen; SD, standard deviation; USD, United States dollars

AC C

Demographic characteristics measured on the 1L index date (date of 1L therapy initiation); clinical characteristics measured in the 6 months prior to the 1L index date (baseline period); see Figure 1 2 Regorafenib or ziv-aflibercept 3 Panitumumab, regorafenib, or ziv-aflibercept

22

ACCEPTED MANUSCRIPT

Table 2. First-line Therapy Attributes

1

Days of 1L therapy, Mean (SD) Median

1L BEV

1L CET

N = 6,095

N = 453

207 (160)

170 (150)

179

119

Chemotherapy used in 1L therapy, % 4.3%

4.2%

Irinotecan with or without other agents

21.7%

59.6%

Oxaliplatin with or without other agents

68.3%

19.2%

6.9%

17.9%

Appearance of new agent

31.3%

20.8%

Treatment gap (>120 days) of all agents

29.8%

33.8%

Biologic monotherapy

Discontinuation of all agents

2

M AN U

1L therapy end reason, %

18.6%

Inpatient death Disenrollment from health insurance End of study period (September 30, 2014)

SC

Fluorouracil/Capecitabine only

RI PT

1L Comparison

31.1%

0.7%

0.9%

12.4%

10.2%

7.2%

3.3%

1L, first line; BEV, bevacizumab-containing regimen; CET, cetuximab-containing regimen; SD, standard deviation

TE D

1L therapy attributes measured during 1L therapy follow-up period; see Figure 1 With no observed re-initiation of any agents from 1L therapy

EP

2

AC C

1

23

ACCEPTED MANUSCRIPT

Table 3. First- and Second-line Therapy Attributes

1

Treatment Sequencing Comparison 1L BEV2L CET

1L BEV2L PAN

1L BEV2 2L Other

N = 1,606

N = 391

N = 108

N = 113

Chemotherapy used in 1L therapy, %

N = 63

N = 54

N = 17

Irinotecan with or without other agents

17.1%

33.0%

34.3%

24.8%

51.9%

82.4%

Oxaliplatin with or without other agents

75.6%

66.0%

61.1%

70.8%

22.2%

22.2%

11.8%

3.4%

2.0%

1.9%

2.7%

7.9%

20.4%

5.9%

5.4%

5.4%

Irinotecan with or without other agents

68.1%

77.2%

Oxaliplatin with or without other agents

24.5%

9.5%

5.5%

8.7%

394 (220) 350

Biologic monotherapy Days of 1L + 2L therapy, Mean (SD) Median 2L therapy end reason, %

30.6%

Appearance of new agent

23.2%

Treatment gap (>120 days) of all agents

24.2%

4

1.4%

Inpatient death Disenrollment from health insurance

6.2%

7.9%

1.9%

11.8%

44.4%

72.6%

31.7%

72.2%

11.8%

7.4%

1.8%

54.0%

22.2%

11.8%

19.5%

6.3%

9.3%

64.7%

386 (215)

386 (198)

376 (226)

399 (240)

371 (243)

290 (213)

343

363

312

357

332

216

24.8%

24.1%

18.6%

19.0%

18.5%

11.8%

21.2%

17.6%

16.8%

25.4%

25.9%

29.4%

35.5%

42.6%

43.4%

28.6%

29.6%

41.2%

2.3%

0.0%

1.8%

0.0%

0.0%

0.0%

12.5%

11.8%

10.2%

8.8%

9.5%

13.0%

5.9%

8.1%

4.3%

5.6%

10.6%

17.5%

13.0%

11.8%

30.6%

24.8%

24.1%

18.6%

19.0%

18.5%

11.8%

AC C

End of study period (September 30, 2014)

0.0%

32.4%

EP

Discontinuation of all agents

73.0%

5.6%

15.7%

TE D

Fluorouracil/Capecitabine only

0.0%

SC

Chemotherapy used in 2L therapy, %

5.3%

1L CET3 2L Other

1.3%

Biologic monotherapy

3.7%

1L CET2L CET

4.2%

M AN U

Fluorouracil/Capecitabine only

1L CET2L BEV

RI PT

1L BEV2L BEV

1L, first line, 2L, second line; BEV, bevacizumab-containing regimen; CET, cetuximab-containing regimen; SD, standard deviation 1

1L and 2L therapy attributes measured during 1L and 2L therapy follow-up periods; see Figure 1 Regorafenib or ziv-aflibercept 3 Panitumumab, regorafenib, or ziv-aflibercept 4 With no observed re-initiation of any agents from 2L therapy 2

24

ACCEPTED MANUSCRIPT

Table 4. Unadjusted PPPM Healthcare Costs, 2014 USD1

Patients with inpatient admission, % Outpatient medical care costs, Mean (SD) Median

1L CET

1L BEV2L BEV

1L BEV2L CET

1L BEV2L PAN

1L BEV2 2L Other

1L CET2L BEV

1L CET2L CET

1L CET3 2L Other

N = 6,095

N = 453

N = 1,606

N = 391

N = 108

N = 113

N = 63

N = 54

N = 17

2,399 (9,885)

3,889 (11,376)

1,692 (5,655)

2,617 (4,750)

2,012 (3,679)

2,287 (4,242)

4,093 (14,697)

2,835 (6,178)

1,087 (1,772)

0

0

0

667

606

0

781

0

0

34.0%

39.5%

46.0%

60.1%

58.3%

48.7%

61.9%

48.1%

47.1%

6,541 (6,296)

7,135 (7,627)

6,574 (6,094)

6,068 (4,103)

6,134 (6,379)

7,265 (5,498)

7,129 (6,318)

5,750 (3,392)

7,126 (7,412)

4,850

5,229

4,929

5,139

4,419

6,307

5,518

4,909

4,717

111 (979)

73 (186)

67 (187)

116 (306)

100 (296)

94 (173)

97 (197)

71 (121)

22 (52)

57.3%

46.2%

57.3%

63.0%

46.0%

49.2%

53.7%

29.4%

350 (441) 6,711 (7,527) 562 (1,147)

340 (485) 6,167 (6,029) 401 (872)

311 (192) 5,642 (4,014) 513 (938)

300 (195) 5,733 (6,242) 419 (707)

468 (1,219) 6,702 (5,099) 322 (431)

382 (256) 6,650 (6,153) 661 (1,134)

409 (553) 5,270 (3,323) 661 (1,548)

222 (140) 6,882 (7,342) 418 (993)

130

164

161

191

143

149

197

201

112

13,291 (10,089)

15,481 (11,959)

11,016

12,930

12,862 (8,544) 11,076

13,646 (9,091) 11,653

12,773 (8,446) 10,191

11,690 (7,357) 10,302

16,136 (12,350) 13,111

14,464 (9,145) 12,370

16,629 (18,945) 10,880

22,655 (16,712)

27,067 (20,274)

21,530 (13,754)

22,844 (12,480)

21,338 (13,687)

21,564 (11,488)

28,018 (22,037)

23,710 (13,801)

25,260 (25,178)

18,825

21,162

18,034

20,151

16,920

18,919

22,274

20,345

15,061

ED visit costs, Mean (SD)

46.2%

Outpatient office visits costs, Mean (SD) Other outpatient care costs, Mean (SD)

Median Cancer regimen costs, Mean (SD) Median Total healthcare costs, Mean (SD) Median

AC C

Outpatient pharmacy costs, Mean (SD)

333 (690) 6,098 (6,090) 424 (1,005)

EP

Patients with ED visit, %

TE D

Sub-categories of outpatient medical care

SC

Median

1L BEV

M AN U

Inpatient medical care costs, Mean (SD)

Treatment Sequencing Comparison

RI PT

1L Comparison

25

ACCEPTED MANUSCRIPT

1L, first line, 2L, second line; BEV, bevacizumab-containing regimen; CET, cetuximab-containing regimen; ED, emergency department; PPPM, per-patient per-month; SD, standard deviation; USD, United States dollars 1

Costs are measured during 1L and 2L therapy follow-up periods, excluding time between lines of therapy; see Figure 1 Regorafenib or ziv-aflibercept 3 Panitumumab, regorafenib, or ziv-aflibercept

AC C

EP

TE D

M AN U

SC

RI PT

2

26

ACCEPTED MANUSCRIPT

Figures

Variable duration of time between metastasis date and 1L index date 1L index date

1

2L index date

2

2L therapy follow-up period

3

M AN U

1L therapy follow-up period

SC

6-months with no evidence of metastasis or a nonCRC primary cancer for which BEV, CET, or REG is indicated

RI PT

Figure 1. Study Follow-Up

First observed claim with a diagnosis of metastasis between January 1, 2008 and September 30, 2014 = metastasis date

1

TE D

1L, first line; 2L, second line; BEV, bevacizumab-containing regimen; CET, cetuximab; CRC, colorectal cancer; REG, regorafenib

AC C

EP

Demographic characteristics measured on the 1L index date (date of 1L therapy initiation); clinical characteristics measured in the 6 months prior to the 1L index date (baseline period) 2 1L of therapy included all CRC treatment agents added within 28 days of the 1L index date, with this exception: bevacizumab and cetuximab may have been added 28–120 days later; 1L therapy follow-up period extended from 1L index date until first of: (1) appearance of a new agent not in line of therapy (begins 2L therapy); (2) gap of >120 days for all agents in line of therapy (end date was set to last day with any line of therapy drug on hand); (3) inpatient death; (5) disenrollment from health insurance; or (4) end of study period (September 30, 2014) 3 2L of therapy and 2L therapy follow-up were established using same criteria as those of the 1L

ACCEPTED MANUSCRIPT

1

Figure 2. Unadjusted PPPM Cancer Regimen Costs, 2014 USD $18,000

$16,629

$1,195 $13,292

$11,040

$927 $12,000

$13,646 $12,862 $957

$6,561

$12,773

$1,038

$836

$8,077

$7,886

$6,738

$1,090

$1,285

$14,464

$10,067

$1,082

$11,690

$12,568

$10,380

SC

$15,000

$795

$9,000

$6,000 $1,087 $1,097

$1,066

$4,717 $3,000

$930

$4,070

$3,465

$2,317 $0 1L CET

1L BEV-2L BEV N = 1,606

1L BEV-2L CET N = 391

EP

1L BEV

M AN U

$7,140

$775

$3,276

TE D

PPPM Cancer Regimen Costs, 2014 USD

$15,482

RI PT

$16,136

1L BEV-2L PAN N = 108

$1,214

$923

$3,570

1L BEV-2L Other N = 113

AC C

Chemotherapy drug

$1,074

$2,297

1L CET-2L BEV N = 63

BEV-containing 1L

1L Comparison

$705

$2,831

1L CET-2L CET N = 54

$1,898 1L CET-2L Other N = 17

CET-containing 1L

Treatment Sequencing Comparison

Chemotherapy administration

Biologic drug

Biologic administration

1L, first line, 2L, second line; BEV, bevacizumab-containing regimen; CET, cetuximab-containing regimen; ED, emergency department; PPPM, per-patient per-month; USD, United States dollars 1 Costs are measured during 1L for the first-line comparison and during the 1L and 2L therapy follow-up periods, excluding time between lines of therapy for the treatment sequencing comparison; see Figure 1 2 Regorafenib or ziv-aflibercept 3 Panitumumab, regorafenib, or ziv-aflibercept

ACCEPTED MANUSCRIPT

RI PT

1

$4,500 $4,000

$3,000

$3,135

$2,000 CR2=1.04 95% CI=0.97-1.11 P=0.226

$1,500 $1,000

$845

$500 $0 1L CET

1L BEV-2L Other N = 221

CR2=1.07 95% CI=1.02-1.13) P=0.010

$4,279

1L BEV-2L CET N = 391

BEV-containing 1L

AC C

CR2=1.11 95% CI=0.99-1.24 P=0.071 $2,179

$1,402

Excess cost vs. 1L BEV-2L BEV

EP

Excess cost vs. 1L BEV

M AN U

$2,500

1L Comparison

CR2=1.21 95% CI=1.08-1.37) P=0.001

SC

$3,500

CR2=1.15, 95% CI=1.09-1.22 P<0.001

TE D

PPPM Excess Total Healthcare Costs, 2014 USD

Figure 3. Multivariable-Adjusted PPPM Excess Total Healthcare Costs, 2014 USD

1L CET-2L Other N = 71

1L CET-2L BEV N = 63

Excess cost vs. 1L BEV-2L BEV CET-containing 1L

Treatment Sequencing Comparison

1L, first line, 2L, second line; BEV, bevacizumab-containing regimen; CET, cetuximab-containing regimen; CR, cost ratio; PPPM, per-patient permonth; USD, United States dollars 1

Costs are measured during 1L and 2L therapy follow-up periods, excluding time between lines of therapy; see Figure 1; ‘Other’ in ‘1L BEV – 2L Other’ and ‘1L CET – 2L Other’ corresponds to panitumumab, regorafenib, or ziv-aflibercept; 2 Cost ratio uses 1L BEV as reference group in 1L Comparison, 1L BEV-2L BEV in Treatment Sequencing Comparison; CR>1 indicates lower costs in reference group

ACCEPTED MANUSCRIPT

Highlights

EP

TE D

M AN U

SC

RI PT

This study examined PPPM costs between 1L or 1L-2L BEV- vs CET- containing regimens PPPM costs 1L-2L were higher for 1L CET- vs 1L BEV- containing regimens. Cost differences may be meaningful in real-world clinical practice.

AC C

• • •

ACCEPTED MANUSCRIPT

Supplementary Material

1

RI PT

Table S1. Patient Selection Criteria and Attrition

SC

Patients with ≥1 non-diagnostic medical claim with ICD-9-CM diagnosis code for colorectal cancer between January 1, 2008, and September 30, 2014 2 + at ≥1 non-diagnostic medical claim with ICD-9-CM code for metastasis between January 1, 2008, and September 30, 2014; date of first observed metastasis claim referred to as metastasis date + aged 18 years or older on metastasis date + 6 months of continuous enrollment in medical and pharmacy benefits prior to metastasis date

Patients N 346,741 72,921 42,498 39,709

+ no non-diagnostic medical claims with ICD-9-CM diagnosis code for non-CRC primary cancer with indication for BEV, CET, or 3 regorafenib in the 6 months prior to metastasis date

35,888

+ no claims for any biologics (BEV, CET, panitumumab [PAN], ziv-aflibercept, or regorafenib) prior to metastasis date

31,770

M AN U

+ no non-diagnostic medical claims with ICD-9-CM diagnosis code for metastasis in the 6 months prior to metastasis date

4

15,108

+ continuous enrollment in medical and pharmacy benefits from metastasis date through 1 month after first-line index date

14,397

+ no evidence of multiple biologics in first line of therapy

14,365

TE D

+ 1L treatment regimen containing biologic and/or chemotherapy agent of interest

First-line Comparison

EP

1L BEV

AC C

1L CET

6,095 453

+ 2L treatment regimen containing CRC biologic

5,730

+ no evidence of other CRC biologics in second line of therapy

2,352

Treatment Sequencing Comparison

ACCEPTED MANUSCRIPT

1,606

1L BEV to 2L CET

391

1L BEV to 2L PAN

108

RI PT

1L BEV to 2L BEV

1L BEV to other biologic 1L CET to 2L BEV

SC

1L CET to 2L CET

M AN U

1L CET to other biologic

113 63 54 17

1L, first line; 2L, second line; ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification; CRC, colorectal cancer; BEV, bevacizumab-containing regimen; CET, cetuximab-containing regimen; PAN, panitumumab-containing regimen 1

Colorectal cancer ICD-9-CM diagnosis codes: 153.xx, 154.0x, 154.1x, 154.8x Metastasis ICD-9-CM diagnosis codes: 196.0x–196.1x, 196.3x–196.5x, 196.8x, 197.0x–197.3x, 197.7x–198.xx 3 Other primary cancers ICD-9-CM diagnosis codes: 140.xx–146.xx, 148.xx–149.xx, 160.xx–161.xx, 162.2x–162.9x, 180.xx, 189.0x, 191.xx; panitumumab and ziv-aflibercept do not have any other indications 4 CRC biologics and chemotherapy agents: bevacizumab, cetuximab, panitumumab, regorafenib, ziv-aflibercept, capecitabine/5-fluorouracil, irinotecan, oxaliplatin

AC C

EP

TE D

2