Economic burden of toxicities associated with metastatic colorectal cancer treatment regimens containing monoclonal antibodies

Abstract

Objectives:

Little is known about toxicity-related costs of monoclonal antibody treatments in metastatic colorectal cancer. This study aimed to identify toxicities associated with bevacizumab, cetuximab, and panitumumab and estimate the direct costs of these toxicities.

Methods:

Grade 3 and 4 toxicities were identified by a comprehensive literature search. Inpatient costs were estimated using ICD-9 codes and 2007 Medicare payments from the Healthcare Cost and Utilization Project database; costs were converted to 2010 dollars. Outpatient costs were estimated by applying 2010 Medicare reimbursement rates to resource use assumptions (based on in-depth clinical interviews).

Results:

Toxicities associated with bevacizumab included hypertension, arterial thrombosis, hemorrhage, gastrointestinal (GI) perforation, fistula, and wound-healing complications; toxicities associated with cetuximab and panitumumab included skin rash, hypomagnesemia, and infusion reactions. The inpatient cost per event was highest for GI perforation (USD 32,443), followed by fistula (USD 29,062), arterial thrombosis (USD 20,346), and wound-healing complications (USD 13,240), while inpatient costs per event for hypomagnesemia and skin rash were among the lowest. The cost per event of toxicities treated in the outpatient setting included USD 185 for skin rash up to USD 585 for wound-healing complications.

Limitations:

Treatment costs of toxicities for the outpatient setting were determined using assumptions validated by clinicians, and unit costs were based on Medicare reimbursement rates, which are often lower than the reimbursement rates for commercial health insurance plans. Toxicities included were only grades 3 and 4 adverse events and might be limited by differences between clinical studies.

Conclusions:

Monoclonal antibodies have different toxicity profiles and the costs associated with managing these toxicities vary greatly.

Keywords::

• Colon cancer
• Clinical trials
• Monoclonal antibody

Introduction

Advances in the treatment of metastatic colorectal cancer (mCRC) have improved median survival from 6–9 months to almost 2 years, largely as a result of improvements in systemic therapy1–3. The introduction of monoclonal antibodies targeting the vascular endothelial growth factor (VEGF; bevacizumab) or the epidermal growth factor receptor (EGFR; cetuximab and panitumumab) is considered a significant contributor to these improved outcomes4–8.

The economic evaluation of monoclonal antibodies in the treatment of mCRC has been based primarily on effectiveness estimates, such as anticipated improvements in survival or tumor response rate9,10. While demonstrated survival benefits are key drivers for economic analysis, treatment-related toxicities are also important determinants of value. In particular, serious toxicities that lead to subsequent morbidity and mortality can place a significant burden on resources. The US and European regulatory authority-approved product information suggests that gastrointestinal (GI) perforation, surgical and wound-healing complications, hypertension, arterial and venous thromboembolism, and hemorrhage are serious events associated with bevacizumab treatment (Avastin Prescribing Information, Genentech Inc, San Francisco, CA, 2009; Avastin Summary of Product Characteristics, Roche Products Ltd, Basel, Switzerland, 2010). Inhibitors of the EGFR have been associated with skin rash, hypomagnesemia, and infusion reactions, although the rates differ between agents. Based on regulatory authority-approved product information, cetuximab (a chimeric monoclonal antibody) is associated with severe infusion reactions in ∼3% of patients (Erbitux Prescribing Information, Imclone Systems Inc and Bristol Myers Squibb Company, New York and Princeton, 2009; Erbitux Summary of Product Characteristics, Merck KGaA, Darmstadt, Germany, 2010), and such reactions are observed in ∼1% of patients with the fully human monoclonal antibody panitumumab (Vectibix Prescribing Information, Amgen Inc, Thousand Oaks, CA, 2008; Vectibix Summary of Product Characteristics, Amgen Ltd, Cambridge, UK, 2010).

Limited information on the cost of treatment-related toxicities associated with chemotherapies in patients with CRC is available11,12, and, in particular, little is known about the economic burden of toxicities related specifically to monoclonal antibodies for mCRC. Knowledge of these costs are important because it can help inform evaluations of the clinical and cost-effectiveness of monoclonal antibodies for the management of patients with mCRC. The primary objective of this study was, therefore, to estimate the direct cost of toxicities associated with the use of monoclonal antibodies in the treatment of mCRC.

Materials and methods

Data selection

A comprehensive and systematic literature search was performed to identify clinical studies evaluating the use of bevacizumab-, cetuximab-, or panitumumab-containing treatment regimens recommended for the treatment of mCRC by the NCCN Colon Cancer and NCCN Rectal Cancer Guidelines. Studies of recommended regimens were initially identified by an ancestral search of reference lists from the 2010 NCCN Clinical Practice Guidelines in Oncology for colon cancer and for rectal cancer13,14. A targeted PubMed literature search, limited to phase 3 clinical trials published in the English language between 2000 and 2010, was performed. Phase 2 trials were only included when Phase 3 data were not available for that recommended regimen. PubMed search terms used included ‘bevacizumab’ or ‘cetuximab’ or ‘panitumumab’ combined with ‘colorectal neoplasms’ or ‘colonic neoplasms’, or ‘rectal neoplasms’ combined with ‘neoplasm metastasis’ or ‘metastatic’ or ‘advanced’. To ensure that we included the most recent and late-breaking bevacizumab, cetuximab, and panitumumab studies, abstracts from the American Society of Clinical Oncology and European Society of Medical Oncology annual meetings (years 2008–2010) were reviewed and considered for inclusion. Studies meeting the inclusion criteria were reviewed to extract information on treatment setting, study phase, number of patients receiving monoclonal antibody treatment, treatment history, and reported grade 3–4 adverse events. Finally, regulatory-approved labels were reviewed to identify additional toxicities. Specifically, adverse events were identified from the ‘Warning and Precautions’ section of the Food and Drug Administration-approved prescribing information and the ‘Special Warning and Precautions for Use’ section of the European Medicines Agency-approved Summary of Product Characteristics for bevacizumab, cetuximab, and panitumumab (Avastin PI 2009; Erbitux PI 2009; Vectibix PI 2008; Avastin SPC 2010; Erbitux SPC 2010; Vectibix SPC 2010).

Determination of the costs of treating toxicities

Since the same or similar toxicities can be reported using different terminology in different clinical trials or product labels, the adverse events identified were placed into representative groupings based on similarities in the event type and, hence, anticipated similarities in the treatment approach, as determined by clinical expert opinion from a panel of two oncologists and one cardiologist. Within each group, the most clinically significant toxicity (as identified by clinical experts) was selected as the ‘representative’ adverse event, and this was then used to estimate the direct cost of treatment. Inpatient and outpatient costs were estimated for each representative adverse event. The exception to this was for those events which, according to clinician opinion, would always be treated on an inpatient basis when presenting acutely, and for which only inpatient costs were calculated. The costs were based on a US payer perspective and were estimated from reimbursed amounts for services delivered, reported in 2010 dollars.

Inpatient costs

The costs of toxicities if they were treated in the inpatient setting were estimated using inpatient costs from the 2007 Healthcare Cost and Utilization Project (HCUP) nationwide inpatient sample (NIS) database, the largest US all-payer inpatient care database15. The NIS contains discharge-level records of the total cost of each hospital stay, including drugs and devices. The primary International Classification of Diseases, 9th Revision (ICD-9)16 codes were identified for the ‘representative’ toxicities using Flash Code™, a medical billing/coding database, and all codes were checked and validated by a clinician. The mean 2007 unit cost of each inpatient event, based on ICD-9 codes, was then identified from the HCUP NIS database. Costs were converted to 2010 dollars using the Consumer Price Index for medical care services (Bureau of Labor Statistics website).

Outpatient costs

The outpatient cost of each representative toxicity was based on the total cost of healthcare resources used for treating that event if the toxicity were to be treated in the outpatient setting. Use of resources was determined by in-depth clinical interviews with a panel of two oncologists and one cardiologist. Resources included office visits, laboratory tests, and procedures. However, costs of drugs that were used in the outpatient setting to treat toxicities were not included. Current Procedural Terminology (CPT) codes were determined for each resource and were used to identify the corresponding unit costs of those resources, based on the Medicare physician fee schedule, identified from the 2010 national average payment amount17. Outpatient treatment costs were calculated as the sum of the unit costs of all resource assumptions for treating that toxicity. For example, when an anemia event was assumed to be treated in the outpatient setting, an initial physician visit, two complete blood counts, and a follow-up physician visit would be required to treat this event. The outpatient cost for an anemia event would therefore be calculated as (2× unit cost for a physician visit) + (2 ×unit cost for a complete blood count laboratory test).

Results

Data sources and severe toxicities identified

Twenty publications of clinical trials met the inclusion criteria, of which 16 were phase 3 clinical trials, one reported a combined analysis of phase 2 and 3 studies, and three were phase 2 (Table 1). The way in which grade 3–4 adverse events were reported varied between clinical trials and publications: some reported ‘adverse events of interest’, while others reported ‘selected adverse events’, ‘most frequently observed adverse events’, adverse events reported in a specified period of time, and adverse events reported in ≥3% of patients. Among all studies, 16 reported grade 3 and 4 adverse events together and four reported grade 3 and 4 adverse events separately.

Table 1.  Summary of clinical studies of bevacizumab, cetuximab, or panitumumab for the treatment of mCRC in which associated grade 3–4 toxicities were reported.

Regimen	Reference	Study setting	Phase	Sample size	Treatment history
Cetuximab	Pessino et al.18	1st line	2	39	NA
	Cunningham et al.19	2nd line/3rd line	3	111	Irinotecan-based regimen
	Jonker et al.20	2nd line/3rd line	3	288	Fluoropyrimidine, irinotecan and oxaliplatin
Panitumumab	Van Cutsem et al.21	3rd line	3	231	Fluoropyrimidine, irinotecan and oxaliplatin
5-FU/LV + bevacizumab	Hurwitz et al.22	1st line	3	110	NA
	Kabbinavar et al.5	1st line	2/3	249	NA
IFL + bevacizumab	Hurwitz et al.2	1st line	3	402	NA
Irinotecan-based chemotherapy + bevacizumab	Hecht et al.23	1st line	3	113	NA
Irinotecan + cetuximab	Sobrero et al.24	2nd line	3	648	Fluoropyrimidine and oxaliplatin
	Cunningham et al.19	2nd line/3rd line	3	218	Oxaliplatin and irinotecan
FOLFIRI + cetuximab	Van Cutsem et al.6	1st line	3	599	NA
	Koo et al.25	3rd line	2	31	Fluoropyrimidine, irinotecan and oxaliplatin
FOLFIRI + panitumumab	Karthaus M et al. (Presented at Gastrointestinal Cancers Symposium 2009. Abstract 388)	1st line	2	154	NA
	Peeters M et al.26	2nd line	3	541	Oxaliplatin and bevacizumab
Oxaliplatin-based chemotherapy + bevacizumab	Hecht et al.23	1st line	3	397	NA
CapeOx + bevacizumab	Hochster et al.27	1st line	3	72	NA
	Saltz et al.28	1st line	3	349	NA
	Tol et al.29	1st line	3	368	NA
CapeOx + cetuximab	Adams et al.30	1st line	3	166	NA
FOLFOX4 + bevacizumab	Saltz et al.28	1st line	3	349	NA
	Giantonio et al.31	2nd line	3	287	Fluoropyrimidine and irinotecan
FOLFOX4 + cetuximab	Bokemeyer et al.4	1st line	3	170	NA
FOLFOX4 + panitumumab	Douillard et al.8	1st line	3	593	NA
mFOLFOX6 + bevacizumab	Hochster et al.27	1st line	3	71	NA

NA, not available.

A total of 60 grade 3–4 adverse events were identified from clinical studies and product labels (Table 2). A number of these (including thrombocytopenia, diarrhoea, vomiting, abdominal pain, and hand–foot syndrome) were reported in association with regimens containing each of the monoclonal antibodies. Severe toxicities associated with bevacizumab included GI perforation, fistula, hemorrhage, wound-healing complications, hypertension, reversible posterior leukoencephalopathy syndrome (RPLS), arterial and venous thromboembolic events, congestive heart failure, proteinuria, neutropenia and infections, and hypersensitivity reactions/infusion reactions (Avastin PI 2009; Avastin SPC 2010)2,28,29,31. Severe toxicities associated with the use of the EGFR inhibitors, cetuximab and panitumumab, included infusion reactions, interstitial lung disease/pulmonary fibrosis, dermatologic toxicity (skin rash), and hypomagnesemia, although rates of these toxicities differed between the two drugs (Erbitux PI 2009; Vectibix PI 2008; Erbitux SPC 2010; Vectibix SPC, 2010)6,20,21,24,26.

Table 2.  Grouping of severe toxicities identified in association with bevacizumab, cetuximab, and panitumumab.

Representative event^a	Other adverse events within the group^b
Anemia	Thrombocytopenia
Arterial thromboembolic event	Cardiac ischemia, cerebrovascular ischemia, left ventricular dysfunction, thromboembolism
Hemorrhage
Deep vein thrombosis	Deep vein thrombophlebitis, venous thrombotic event, thrombotic event
Dyspnea
Failure to thrive	Anorexia, fatigue, weight loss
Febrile neutropenia	Infection (NOS), leukopenia, neutropenia
Fistula	Intestinal obstruction, fistula/intra-abdominal abscess, intra-abdominal thrombosis
Gastrointestinal perforation
Hypertension
Hypomagnesemia
Infusion reaction	Hypersensitivity reaction
Interstitial lung disease
Peripheral edema	Edema, hypocalcaemia, hypokalemia
Peripheral neuropathy	Asthenia, neuropathy, neurologic toxicity paresthesia, sensory neuropathy
Proteinuria
Pulmonary embolism
Renal failure
Reversible posterior leukoencephalopathy syndrome
Skin rash	Mucositis, stomatitis, skin toxicity, skin exfoliation, hand–foot syndrome, paronychia, skin and subcutaneous (all)
Urinary tract infection
Vomiting	Abdominal pain, constipation, diarrhea, dehydration
Wound-healing complication

^aThe representative toxicity was the most clinically significant event within that grouping (as determined by clinical opinion) on which the costs of the toxicities in the group were based.

^bToxicities were grouped based on anticipated similarities in treatment approach, as determined by clinical opinion.

NOS, not otherwise specified.

Costs of treating severe toxicities

The severe toxicities identified were categorized into 23 groups for the determination of treatment costs based on clinical expert opinion from a panel of two oncologists and one cardiologist, as illustrated in Table 2. Some events were grouped together (for example, cardiac and cerebrovascular ischemia, left ventricular dysfunction, thromboembolism, and arterial thromboembolic event), while others (such as wound-healing complication and hypomagnesemia) were considered individually.

Inpatient costs

The estimated inpatient costs of treating each toxicity are shown in Table 3. The highest inpatient cost per event was for GI perforation (USD 32,443), followed by fistula (USD 29,062), arterial thromboembolic event (USD 20,346), wound-healing complication (USD 13,240), hemorrhage (USD 12,956), and infusion reaction (USD 10,877). The inpatient cost per event for hypomagnesemia and skin rash were among the lowest (USD 6174 and 4424, respectively).

Table 3.  Inpatient costs of treating severe toxicities associated with the use of the monoclonal antibodies bevacizumab, cetuximab, and panitumumab in metastatic colorectal cancer.

Toxicity	ICD-9 code	Inpatient costs (2010 USD)
Anemia	285.22	6219
Arterial thromboembolic event	444.22	20,346
Deep vein thrombosis	451.41	8748
Dyspnea	786.05	5582
Failure to thrive	783.7	8104
Febrile neutropenia	288.03	12,606
Fistula	537.4	29,062
Gastrointestinal perforation	569.83	32,443
Hemorrhage	459	12,956
Hypertension	405.99	8453
Hypomagnesemia	275.2	6174
Infusion reaction	N/A	10,877*
Interstitial lung disease	515	13,001
Peripheral edema	782.3	5695
Peripheral neuropathy	357.6	13,113
Proteinuria	791.0	4618
Pulmonary embolism	415.19	11,411
Renal failure	584.9	10,688
Reversible posterior leukoencephalopathy syndrome	348.5	10,116
Skin rash	782.1	4424
Urinary tract infection	599.0	6890
Vomiting	787.01	5559
Wound-healing complication	998.32	13,240

*Costs based on Foley et al.32 as no ICD-9 code was available for this event.

Outpatient costs

Unit costs for the individual resources used for the treatment of toxicities in the outpatient setting are provided in Table 4. These varied from laboratory fees of USD 4.54 for a urinalysis test to USD 370.59 for professional fees associated with computed tomography scanning.

Table 4.  Unit costs for resources used to determine the costs of toxicities in the outpatient setting.

Resource	CPT code	Unit cost (2010 USD)	Fee type
Basic metabolic panel	80047	12.12	Lab fee
Complete blood count	85025	11.14	Lab fee
Chest X-ray	71030	46.17	Professional fee
Computed	70450	193.05	Professional fee
tomography scan	72125	253.31	Professional fee
	70488	370.59	Professional fee
Electrocardiogram	93000	30.92	Professional fee
	93010	9.02	Professional fee
Electrolytes	80051	10.05	Lab fee
Medical stocking	A6531	45.43	Durable medical equipment fee
Physician visit, moderate complexity	99212	38.97	Professional fee
	99214	92.33	Professional fee
	99215	132.79	Professional fee
Platelets	85025	11.14	Lab fee
Protime	85610	5.62	Lab fee
Renal function test	78596	342.63	Professional fee
Spirometry	94010	32.84	Professional fee
Thyroid function test	78006	240.78	Professional fee
	78007	124.49	Professional fee
	84479	9.27	Lab fee
Tissue culture	88233	201.57	Lab fee
Ultrasound	37250	111.45	Professional fee
	37251	85.52	Professional fee
Urinalysis	81001	4.54	Lab fee
Vacuum closure system	97606	40.05	Professional fee
	A7043	36.02	Durable medical equipment fee

The resources used for determining the costs of treating each toxicity in the outpatient setting (based on in-depth clinical interviews) and the corresponding total costs are shown in Table 5. Outpatient costs of toxicities depended on the nature and extent of the resources used; for example, skin toxicity and hypertension (each requiring two physician visits) were estimated to cost USD 185, whereas a wound-healing complication requiring more substantial resources, including imaging equipment and devices, was estimated to cost USD 585. It was worth noting that costs of drugs that were used in the outpatient setting to treat toxicities were not included in the outpatient treatment costs.

Table 5.  Outpatient costs of treating severe toxicities associated with the use of the monoclonal antibodies bevacizumab, cetuximab, and panitumumab in metastatic colorectal cancer.

Toxicity	Resource use assumptions*	Outpatient cost (2010 USD)
Anemia	2× physician visit; 2× complete blood count	207
Dyspnea	3× physician visit; 2× spirometry	343
Failure to thrive	2× physician visit; 1× complete blood count; 1× electrolytes; 1× thyroid function test	447
Febrile neutropenia	2× physician visit; 2× complete blood count; 1× electrolytes	217
Hypertension	2× physician visit	185
Hypomagnesemia	2× physician visit; 2× electrolytes	205
Infusion reaction	2× physician visit; 1× complete blood count; 1× electrocardiogram; 1× urinalysis; 1× renal function test	327
Peripheral edema	2× physician visit; 1× electrolytes; 1× medical stocking; 1× thyroid function test	481
Peripheral neuropathy	2× physician visit	185
Proteinuria	2× physician visit; 1× urinalysis; 1× basic metabolic panel	201
Skin rash	2× physician visit	185
Urinary tract infection	1× physician visit; 1× complete blood count	103
Vomiting	2× physician visit; 1× electrolytes	195
Wound-healing complication	2× physician visits; 1× tissue culture; 1× complete blood count; 1× ultrasound; 1× vacuum closure system	585

Based on clinical expert opinion, arterial thromboembolic event, hemorrhage, deep vein thrombosis, fistula, gastrointestinal perforation, interstitial lung disease, pulmonary embolism, renal failure, and reversible posterior leukoencephalopathy syndrome were all assumed to be treated on an inpatient basis only.

*Based on in-depth clinical interview.

Discussion

The monoclonal antibodies bevacizumab, cetuximab, and panitumumab have distinct and well-defined tolerability profiles in the treatment of mCRC. Most notably, bevacizumab is associated with GI perforation, non-GI fistula formation, hemorrhage, wound-healing complications, hypertension, arterial and venous thromboembolic events, and RPLS (Avastin PI 2009; Avastin SPC 2010)2,28,29,31, while cetuximab and panitumumab are associated with infusion reactions, dermatologic toxicity, and hypomagnesemia (although with different incidence rates) (Erbitux PI 2009; Vectibix PI 2008; Erbitux SPC 2010; Vectibix SPC 2010)6,19,21,26. Our analysis illustrates that the cost of treating these toxicities varied substantially depending on the event itself, and whether it would be severe enough to warrant inpatient care. Understandably, the most costly events were those that required the most significant inpatient resources: this included GI perforation, fistula, arterial thromboembolic events, and hemorrhage, events that were considered by clinicians to always require inpatient care. Severe infusion reactions and wound-healing complications would also be relatively costly to treat on an inpatient basis. Skin toxicity, hypomagnesemia and hypertension were relatively less costly events to treat, particularly when they could be managed in an outpatient setting. It was worth noting that some events, such as GI perforation, have a much lower rate compared to the less costly events, such as dermatologic toxicities.

There are a number of limitations to our analysis. The treatment costs of toxicities for the outpatient setting were determined by resources used, which were based on assumptions validated by clinicians. Although these may not always reflect clinical practice at an event level they will be representative at a population level. However, only office visits, laboratory tests, and procedures were considered for outpatient resources, and thus the outpatient costs of some toxicities frequently treated with drugs may have been underestimated. Additionally, unit costs for outpatient services were based on Medicare reimbursement rates, which are often lower than the reimbursement rates for commercial health insurance plans. This may have also led us to underestimate the outpatient costs for toxicities. For inpatient costs, however, use of the NIS database meant that our analysis captured the total costs of the hospital stay, including drugs and devices; although these costs were estimated based on the general population, not based on mCRC patients specifically. Furthermore, the most clinically significant toxicity (as identified by clinical experts) was selected as the ‘representative’ adverse event, and this was then used to estimate the direct cost of treatment; however, the costs for treating these events in the same group might differ.

In terms of the toxicities identified by our search, ensuring a comprehensive and standardized list of severe toxicities was limited by differences between studies in the approach used to define and report grade 3–4 adverse events, the varied sample sizes among studies, variation in patient characteristics, and especially uncertainty regarding the consistency of the naming of adverse events. In addition, our focus was only on identifying the more severe toxicities (grades 3 and 4). We did not capture information on grade 1–2 adverse events reported in clinical trials, although these, by definition, do not require intensive medical intervention. Finally, the evaluated treatment regimens were used in different lines (from 1st line to 3rd line) of therapy, and the treatment populations were not the same for the three targeted agents, such as cetuximab and panitumumab, are restricted to EGFR-expressing and wild-type KRAS mCRC patients.

To our knowledge, this is the first study to evaluate the costs of treating severe toxicities associated with monoclonal antibodies for mCRC. Monoclonal antibodies have different toxicity profiles and the estimated costs associated with managing these toxicities vary greatly. Knowledge of these costs can help inform evaluations of the clinical and cost-effectiveness of monoclonal antibodies for the management of patients with mCRC.

Transparency

Declaration of funding

This study was funded by Amgen Inc. All authors had full access to the data and had final responsibility for the decision to submit the manuscript.

Declaration of financial/other relationships

ZZ, SG, and BB have disclosed that they are employees of Amgen Inc. CB, JM, and KT have disclosed that they are employees of IMS Consulting Group, a company that received funding from Amgen Inc to perform the analyses reported in this manuscript.

Acknowledgments

Medical writing assistance was provided by ApotheCom ScopeMedical Ltd, funded by Amgen Inc.