Cost-effectiveness projections of oxaliplatin and infusional fluorouracil versus irinotecan and bolus fluorouracil in first-line therapy for metastatic colorectal carcinoma

Authors

  • Bruce E. Hillner M.D.,

    Corresponding author
    1. Department of Internal Medicine and Massey Cancer Center, Virginia Commonwealth University, Richmond, Virginia
    • Department of Internal Medicine and Massey Cancer Center, Virginia Commonwealth University, 1101 East Marshall Street, Room 7013, Richmond, VA 23298
    Search for more papers by this author
    • Fax: (804) 828-3741

    • Daniel Sargent is a consultant to Sanofi Aventis. Richard Goldberg receives research support and honoraria from Pfizer and Sanofi Aventis. Bruce Hillner received funds for this project from a grant from Sanofi Aventis.

  • Deborah Schrag M.D.,

    1. Health Outcomes Research Group, Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, New York, New York
    Search for more papers by this author
  • Daniel J. Sargent Ph.D.,

    1. Division of Biostatistics, Mayo Clinic, Rochester, Minnesota
    Search for more papers by this author
    • Daniel Sargent is a consultant to Sanofi Aventis. Richard Goldberg receives research support and honoraria from Pfizer and Sanofi Aventis. Bruce Hillner received funds for this project from a grant from Sanofi Aventis.

  • Charles S. Fuchs M.D., M.P.H.,

    1. Dana-Farber Cancer Institute, Boston, Massachusetts
    Search for more papers by this author
  • Richard M. Goldberg M.D.

    1. Division of Hematology and Oncology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina
    Search for more papers by this author
    • Daniel Sargent is a consultant to Sanofi Aventis. Richard Goldberg receives research support and honoraria from Pfizer and Sanofi Aventis. Bruce Hillner received funds for this project from a grant from Sanofi Aventis.


Abstract

BACKGROUND

The results of a randomized comparison study (N9741) showed that oxaliplatin and infusional fluorouracil (FU) (FOLFOX) was superior to the previous standard of care in the United States, irinotecan and bolus FU (IFL), as first-line therapy for patients with metastatic colon carcinoma. The trade-offs between costs and survival for these two regimens have not been explored.

METHODS

A post-hoc, incremental cost-effectiveness (ICE) projection using simulated cohorts of patients starting FOLFOX or IFL was tracked for major clinical events, toxicities, and survival. Recurrence and survival risks were based on clinical trial data. Resource use was projected using observed dose intensity, duration of therapy, delays in therapy, and toxicities Grade > 2 in N9741. The frequency, costs, and consequences of second-line therapy were examined. The time frame was 5 years, and the perspective was that of Medicare as a third-party payer.

RESULTS

Initial treatment with FOLFOX versus IFL had an average incremental cost of $29,523, a survival benefit of 4.4 months, and an ICE of $80,410 per life year (LY), $111,890 per quality-adjusted LY, and $89,080 per progression-free year. By using the 95% confidence interval for the time to progression observed in N9741, the ICE associated with FOLFOX ranged from $121,220 to $59,250 per LY. In the clinical trial, dose delays and skipped doses were frequent. If progression-free patients were treated without delay for the first year or lifetime, then the ICE for FOLFOX increased to $117,910 and $222,200 per LY, respectively. The ICE increased to $84,780 per LY when the model incorporated a revised IFL schedule with lower early toxicity and similar rates of treatment with second-line regimens.

CONCLUSIONS

FOLFOX provided substantial benefits that incurred substantial additional costs. The ICE for FOLFOX fell into the upper range of commonly accepted oncology interventions in the context of the United States healthcare system. Cancer 2005. © 2005 American Cancer Society.

Although the survival rate for patients with metastatic colorectal carcinoma remained stagnant for several decades, recently, there has been a burst of progress.1, 2 It was found that irinotecan and oxaliplatin had activity in clinical trials, and those drugs were approved by the United States Food and Drug Administration (FDA) in 1996 and 2002, respectively. Initial treatment with irinotecan and either bolus (North American preference)3 or infused (European preference) fluorouracil (FU) with leucovorin (LV) (FU/LV)4 improved outcomes significantly compared with FU/LV. The bolus regimen of irinotecan plus FU/LV (IFL) was approved as first-line therapy for patients with advanced colorectal carcinoma in 2000. At that time, the Oncologic Drug Advisory Committee recommended to the FDA the consideration of this combination as a regulatory standard.

Intergroup study N9741 compared combinations of FU/LV, irinotecan, and oxaliplatin in patients with advanced colorectal carcinoma. That trial, which was reported first in 2002 and was published in 2004, demonstrated that patients who were treated with oxaliplatin and infusional FU/LV (FOLFOX) had improved response rates, a longer time to disease progression, and better overall survival compared with patients who received the control regimen of IFL.5 The improvement in efficacy, coupled with a favorable toxicity profile,6 suggests that FOLFOX should be considered a first-line standard of care.

Although FOLFOX has been embraced rapidly as a standard of care, the financial implications of its use are substantial. Both countries and healthcare financing structures are concerned increasingly about the cost consequences of new treatment approaches in all specialty areas. Although the economic implications of these treatments are considered formally in the product-approval process in many large democracies,7, 8 the largest payer in the United States, the Medicare program, does not consider costs in the approval process.9, 10 In contrast, cost implications are considered by non-Medicare insurers, whose decisions may then influence the treatment's utilization in routine clinical practice.9, 11

A strong case for a cost-effectiveness analysis can be made based on the price difference between oxaliplatin and irinotecan and the need to change from bolus to infusional FU/LV. Our objective was to estimate the incremental cost-effectiveness (ICE) of FOLFOX compared with IFL based on the outcomes observed in N9741. By using the perspective of an American third-party payer, the objective was to inform healthcare purchasers of the economic consequences of the changing standard of care.

MATERIALS AND METHODS

Study Design

The cost-effectiveness analysis model simulated the observations and consequences of care for patients with metastatic colorectal carcinoma who were eligible for first-line chemotherapy with FOLFOX or IFL. A Markov process model tracked the natural history of patients with newly diagnosed, metastatic colorectal carcinoma over a 5-year period after initiating therapy with either IFL or FOLFOX.12–14 Figure 1 shows a schematic diagram of the transitions between model states. The probabilities of events were derived from those observed in N9741, including 3-year survival data. The projections reflected the actual dosages and toxicity rates of IFL, rates of second-line chemotherapy, and survival after second disease recurrence. The North Central Cancer Treatment Group (NCCTG) granted access to both data collected in N9741 and previously reported data.5 Information regarding individual toxicity types and severity and cycle of onset was provided. Updated survival data from N9741 (3 yrs from randomization) also were obtained and compared with the model results.

Figure 1.

In this Markov model, all patients begin in the “Start first-line irinotecan and bolus fluorouracil (IFL) or oxaliplatin and infusional fluorouracil (FOLFOX)” oval. On a daily basis, patients move to alternative health states until death. Treatment complications leading to death or hospitalization were considered in the first 60 days and 6 months, respectively. For nonfatal treatment complications, treatment was not stopped. An asterisk indicates that every 6 weeks for IFL and every 2 weeks for FOLFOX, patients with stable or responding disease could be treated with first-line therapy. The rate of treatment decreased with time. Patients with progression of their metastatic colorectal carcinoma (CRC) could or could not receive second-line chemotherapy. The transient line from “treatment-associated hospitalization within 6 months” to “progression” has been deleted for clarity.

Endpoints and Perspective

The primary endpoint of this study was the incremental cost per added survival ($ per life yr [LY]) with FOLFOX versus IFL. A 5-year time frame was used for overall survival based on the disease's natural history. Costs and benefits were discounted at 3% per year.15 Because the majority of patients with colorectal carcinoma are diagnosed after age 65 years, we conducted the analysis from the perspective of Medicare as a third-party payer.

A cost-utility analysis was a secondary endpoint, because no differences were found in quality-of-life (QOL) scores between patients during first-line therapy by treatment arm in N9741. The QOL assessment used 3 different measures: a single-item scale, a 12-item symptom distress scale, and a symptom-specific scale. These QOL assessments were performed every 12 weeks and had a 75% completion rate.16 For the cost-utility analyses, the average single-item QOL rating scale (RS) values for patients during first-line therapy were transformed into a utility estimate (if RS < 0.95, then utility = 1.07 × RS) based on prior work showing that a QOL rating scale tends to have lower values than formally elicited utilities.17 This value (0.82) was similar to the values recorded in patients who were undergoing adjuvant FU therapy for colorectal carcinoma.18 The utility estimate for disease progressing after first-line therapy was based on those used in other cost-utility assessments.19

N9741 Trial

The details of the trial's design, patient characteristics, its novel tracking of initial toxicities, and the results have been described previously.5, 6 Five National Cancer Institute cooperative groups conducted the multicenter, Phase III trial. The control arm was IFL, and the investigative arms were FOLFOX and irinotecan and oxaliplatin (IROX). Because IROX was neither the reference standard nor the superior arm in N9741, this regimen was not evaluated in the current model.

Data Elements and Assumptions

Table 1 lists the model's assumptions. These were categorized as general assumptions, i.e., assumptions related to defining benefit, initial therapy, second-line therapy, and adverse events. Tables 2–4 list the baseline probabilities, relative risks, and costs of clinical events.

Table 1. Specific Assumptions
General
  1. FOLFOX: oxaliplatin and infusional fluorouracil; IFL: irinotecan and bolus fluorouracil; DRG: diagnostic related group; VCUHS: Virginia Commonwealth University Health System.

 Benefits and risks were expressed as relative risks that were constant over time.
 No quality-of-life adjustments (for clinical response or neuropathy) were initially included.
 Costs for palliative care after second-line chemotherapy were an assigned cost. Excluding these costs was assessed in sensitivity analysis.
 Costs for restaging during therapy were not included.
 Direct nonmedical and indirect (extended family) costs were not included.
 All costs and benefits were discounted 3% per year.
Benefits
 The survival of each cohort differed based on the relative risk of 3 events: first 60-day mortality, disease progression during first-line therapy, and progression to death.
 The daily probability of progression during first-line therapy was constant and was estimated from the observed median progression-free survival.
 The greater benefit in overall survival compared with progression-free survival in the FOLFOX cohort was reflected in a relative risk ≤1.0 for the transition from disease progression to death.
First-line therapy
 All first-line and second line therapies are given on an outpatient basis.
 The relative dose intensity per cycle per arm was the mean observed in the trial.
 The two groups differed in the need for a permanent, implanted venous-access device (e.g., Medi-port). All FOLFOX patients incurred these costs. It was estimated that 30% of IFL patients would need such a device or would incur this cost later if second-line chemotherapy included continuous infusion.
 Treatment delays were estimated based on data reported from the trial.
 Switching to second-line chemotherapy prior to disease progression was not allowed..
Second-line therapy
 The probability and type of second-line chemotherapy differed between cohorts. In sensitivity analysis, an equal probability of second-line therapy was assessed.
 The duration of second-line therapy was inferred from what was given in recent clinical trials.
 The relative dose intensity for second-line therapy was the same as for first-line therapy.
Adverse events
 Treatment-induced toxicity requiring hospitalizations were limited to the first 6 months of treatment.
 Treatment-associated deaths were limited to deaths within 60 days.
 Costs were projected for Grade 3, 4, and 5 toxicities associated with hospitalization. Per patient, per cycle toxicities were matched to their respective DRG classifications. Costs were inferred from 2004 VCUHS hospital Medicare payments and an estimate of associated physician services.
 All treatment-related deaths occurred in the first 60 days, occurred in hospital, and were assigned the cost incurred as due to septicemia.
 Only one hospitalization per treatment cycle was allowed.
 Costs of treatment-associated neuropathy were not included.
Table 2. Probability, Relative Risks, and Treatment Delays
VariableFOLFOXIFLRangeSourcea
  • FOLFOX: oxaliplatin and infusional fluorouracil; IFL: irinotecan and bolus fluorouracil.

  • a

    Estimates from N9741 were “inferred” if they were derived from data reported in Kaplan–Meier graphs, whereas “summary” estimates were taken directly from the primary reports. (see Goldberg et al., 20045 and Delaunoit et al., 200528) Drs. Goldberg and Schrag provided the expert opinions.

Probability of clinical events: IFL    
Daily well to progression 3.1 × 10−3 Inferred N9741
Daily progression to death 4.1 × 10−3 Inferred N9741
Daily toxic death in first 60 days 8.5 × 10−4 Inferred N9741
Relative risk reduction    
Initial therapy to progression0.741.000.61–0.89Summary N9741
Progression to death0.831.000.83–1.0Inferred N9741
Toxic deaths in first 60 days0.551.000.55–1.0Summary N9741
Delays or deferral in first-line treatment if progression free    
Cycle 2–6 mos (%)778577–100Summary N9741
6–12 mos (%)505050–100Summary N9741
12.0–17.9 mos (%)15250–100Expert
> 18 mos (%)110–100Expert
Semipermanent venous access prior to treatment (%)100300–100Expert
Table 3. Chemotherapy Costs
FactorFOLFOXIFLDifference (%)
  • FOLFOX: oxaliplatin and infusional fluorouracil; IFL: irinotecan and bolus fluorouracil; 95% CI: 95% confidence interval; 5-FU: 5-flurouracil.

  • a

    Assumes a body surface area of 1.8 m2. Drug costs include the dose administered and any necessary drug wastage associated with the use of single-dose vials.

  • b

    The 2004 Medicare reimbursement was defined as 85% of the average wholesale price.

  • c

    For expanded details of the components included, see Table 5. If second-line FOLFOX was given, then venous access line costs potentially were incurred.

Oxaliplatin or irinotecana   
Intended dosage (mg/m2)85125 
Average delivered dosage in N9741 (95% CI)82% (80.3–84.3)73% (70.4–75.2) 
Average delivered dose (mg)a129169 
Average dispensed dose with vial wastage (mg)150180 
Unit cost ($ per mg)b$16.90$6.51 
Dispensed cost per treatment$2535$1172 
Costs per uncomplicated 6-week intervalc   
 Evaluation and laboratory$389$35011
 Oxaliplatin or irinotecana$7605$468762
 5-FU, leucovorin, and antiemeticsa$588$5487
 Chemotherapy delivery$3855$1680229
 Office-based administration($2178)($1680) 
 Infusion pump rental and management($1677)($0) 
Total$12,437$726571
Table 4. Miscellaneous Costs and Actions
VariableFOLFOX (%)IFL (%)SourceCost ($)Source
  • FOLFOX: oxaliplatin and infusional fluorouracil; IFL: irinotecan and bolus fluorouracil; VCUHS: Virginia Commonwealth University Health System; DVT: deep vein thrombosis; FU: fluorouracil; 5-FU: 5-fluorouracil.

  • a

    It was assumed that all deaths (Grade 5 toxicity) within 60 days were due to sepsis.

  • b

    Primary data per trial participant were analyzed and aggregated per cycle by toxicity type and severity (National Cancer Institute Common Toxicity Criteria, version 2.0). These were mapped to their best-match diagnosis-related group (DRG). It was assumed that hospitalization was necessary for Grade 3 and 4 nonhematologic toxicities. ‘Projected VCUHS’ means the average 2004 Medicare reimbursement to the VCUHS for the respective DRG plus an additional 25% for physician services of all types.

  • c

    The treatment cycle length was every 6 weeks for IFL and FOLFOX every 2 weeks for FOLFOX., and toxicity events in the first 6 months.

  • d

    Dosage, administration units, and costs were the same as those for first-line therapy.

  • e

    Values indicate patient care costs for monitoring laboratory studies and physician examinations.

  • f

    Estimated from Medicare payments from audit of 10 same-day surgery procedures at VCUHS.

Grade 5 (fatal) toxicity % and costsab2.64.6N9741 (patient level)14,100Projected VCUHS
Treatment cycles with Grade 3 or 4 toxicity (%) and costsbc     
Diarrhea10.517.8N9741 (patient level)7640Projected VCUHS
Volume depletion7.010.8N9741 (patient level)7640Projected VCUHS
Nausea and emesis5.18.9N9741 (patient level)4500Projected VCUHS
Febrile neutropenia3.15.4N9741 (patient level)10,000Projected VCUHS
Pneumonia8.61.5N9741 (patient level)5400Projected VCUHS
Pulmonary embolism/DVT5.54.6N9741 (patient level)11,300Projected VCUHS
Other8.66.6N9741 (patient level)7500Projected VCUHS
Second-line chemotherapyc     
Cross-over therapy60% IFL24% FOLFOXN9741 (summary) Same as first-lined
FU alone15%42%N9741 (summary) Same as first-lined
Mean no. of cycles given  Rothenberg et al., 200323; Tournigand et al., 200424 Same as first-lined
 IFL or FOLFOX26   
 5-FU3a3a   
Other costs ($)     
Semipermanent venous access   1830Projectedf
Palliative care per day   100Ramsey et al., 200229 and Brown et al., 200230
Dose delaye   130Expert
Costs for neuropathy   0Expert
Grade 1 or 2 diarrhea or volume depletion   0Expert
Table 5. First-Line Therapy Cost Accounting Details
FOLFOX4CPT codeUnit cost or estimateaIFLCPT codeUnit cost or estimatea
  • FOLFOX: oxaliplatin and infusional fluorouracil; CPT: Current Procedural Terminology; IFL: irinotecan and bolus fluorouracil; CBC: complete blood count; 95% CI: 95% confidence interval; 5-FU: 5-flurouracil.

  • a

    The number of units used is not always shown.

  • b

    Costs were rounded to the nearest dollar.

Patient size (m2) 1.8Patient size (m2) 1.8
Relative dose intensity 84%Relative dose intensity 75%
Intended dose (mg) 85Indented dose (mg) 125
Oxaliplatin Medicare payment per mgJ926316.9Irinotecan Medicare payment per mgJ92066.52
Average oxaliplatin delivered dose (mg) 129Average irinotecan delivered dose (mg)b 169
Average irinotecan dispensed dose with vial wastage (mg) 150Average irinotecan dispensed dose with vial wastage (mg) 180
Physician office visit (every 2 wks)99214$58Physician office visit (Wks 1 and 3)99214$58
   Physician office (Wks 2 and 4)99213$35
CBC (every wk)85025$11CBC (every wk)85025$11
Protime (every wk)85610$5Protime (every wk)85610$5
Chemistries (every 2 wks)80053$15Chemistries (every 6 wks)80053$15
AntiemeticsQ0179$130AntiemeticsQ0179$130
Day 1, oxaliplatinJ9263$2535IrinotecanJ9206$1172
Day 1, infusion, first hour oxaliplatin96410$165LeucovorinJ0640$3
Day 1, office infusion, second hour oxaliplatin96412$375-FUJ9190$6
Day 1, leucovorin (high dose)J0640$3Infusion, first hour irinotecan96410$165
Day 1, office infusion, first hour leucovorin96410$165Office infusion, second hour96412$37
Day 1, office infusion, second hour96412$37Office chemotherapy, push 5-FU96408$117
Day 1, office chemotherapy, push 5-FU bolus96408$117Office chemotherapy, push leucovorin96408$117
Day 1, 5-FUJ9190$12   
Day 1, initial filling of external infusion pump96414$204   
Ambulatory reusable infusion pump rentalE0779$17   
Infusion pump supplies per weekA4221, A4222$67   
Day 2, leucovorin (high dose)J0640$3   
Day 2, office infusion, first hour leucovorin96412$165   
Day 2, office infusion, second hour leucovorin96412$37   
Day 2, office chemotherapy, push 5-FU bolus96408$117   
Day 2, refilling of infusion pump96520$156   
Day 2, 5 FUJ9190$12   

The benefits and risks were expressed as relative risks: the risk of death within 60 days, of disease progression, and of disease progression and death. The probabilities were derived from the observed rates using a standard conversion (rates to probabilities) and were assumed to be constant for the life of the patient.20 Patients treated with IFL receive weekly or FOLFOX biweekly chemotherapy. Therefore, patients undergo frequent evaluations. Consequently, we chose a Markov transition interval that reflected the finding that patients may move between states over short periods of time (daily). Ongoing follow-up of N9741 includes a median follow-up of 3 years, and > 85% of patients have died. The model-based results using a constant, daily probability of disease recurrence were highly concordant with the observed survival at 6-month intervals up to 3 years.

The probability of early treatment death was taken directly from N9741. Sensitivity analysis included a lower early mortality rate, because the rate observed after protocol amendment dictated a lower starting IFL dosage.21 Differences in the time to progression were reflected in the different relative risks of progression. Rates observed in the IFL cohort were used as the default in sensitivity analysis.

Because of the limited availability of oxaliplatin during the initial phase of N9741, there was an imbalance in the use of second-line chemotherapy between cohorts. To maintain consistency with the primary trial's results (a greater difference in median survival than in time to progression between the two arms), an inferred hazard rate from progression to death in favor of FOLFOX was used. The sensitivity analysis explored an equal and greater use of second-line chemotherapy with no difference in the late hazard ratio.

Costs

Detailed information on planned versus administered doses and adverse events were collected during the trial. Second-line chemotherapy either before or after progression, but not its duration, also was recorded. This information was used to make a post-hoc projection of anticipated costs for delivering each treatment. The 2004 Medicare national average reimbursement rates for physician and treatment administration actions were used to estimate costs for the resources used.22 In addition, we reviewed the 2003–2004 billing records (stripped of individual identifiers) of 30 patients in the Virginia Commonwealth University Health System (VCUHS) who had received either IFL or FOLFOX for colorectal carcinoma.

The largest single cost category in the model was drug acquisition (in N9741, oxaliplatin and irinotecan were supplied by their manufacturers). Costs were based on actual doses delivered, allowed for single vial wastage, and were 85% of the 2004 United States Redbook price according to Medicare regulations (Table 3). All chemotherapy was administered in a physician-owned, outpatient setting. Table 5 provides further details regarding the costs of therapy. The FOLFOX strategy used FOLFOX4, which requires a Day 2 office visit for infusional LV and bolus FU.

Detailed data on the per-cycle adverse events and their severity were collected in N9741. Data were not collected on the occurrence and number of unscheduled office, hospital, or emergency room visits. Because reported toxicity data were cumulative and were not cycle-specific, permission was obtained from the NCCTG to assess the primary data on per-patient toxicity. Table 4 shows the results of the most common toxicities aggregated by type and severity. If multiple toxicities occurred in a cycle, then the most severe result was used. Individual toxicities were mapped to their best-match diagnostic-related group (DRG). Hospitalization costs were projected for Grade 3 and 4 nonhematologic toxicities from 2004 Medicare reimbursement per DRG code. The average Virginia Commonwealth University Medical Center reimbursement for the respective DRG plus an additional 25% for physician services of all types were combined. It was assumed that all deaths within 60 days would be because of sepsis.

Dose Duration, Delays, and Skipped Doses

A modeling challenge arose because of the observed differences between the time to disease progression, the duration of treatment, and the number of treatment cycles given in each arm (Table 6). After the first 60 days, FOLFOX patients had more delayed and skipped doses, primarily because of neuropathy symptoms, and were more likely to stop treatment before disease progression.

Table 6. Dose Delay Data from N9741
VariableFOLFOXIFL
  1. FOLFOX: oxaliplatin and infusional fluorouracil; IFL: irinotecan and bolus fluorouracil; 95% CI: 95% confidence interval; FU: fluorouracil.

Progression free survival (wks)37.830.0
Cycles of treatment given2-wk cycle6-wk cycle
 Mean no. of cycles11.44.4
 Median no. of cycles10.04.0
Duration of therapy (wks)  
 Mean25.826.6
 95% CI23.5–28.124.1–29.1
 Median23.923.6
Patients with ≥1 dose delay (%)7945/
Cycles that were delayed excluding Cycle 1 (%)22.715.1
Second-line chemotherapy (%)77.066.0
Percentage of patients who started before evidence of disease progression (95% CI)42.0 (36.1–47.9)43.0 (36.9–49.1)
 Oxaliplatin5.020.0
 Irinotecan28.09.0
 FU15.020.0
Percentage of patients who started only after disease progression34.823.1
 Oxaliplatin5.26.8
 Irinotecan30.015.9
 FU28.122.4

The optimal duration of therapy in patients who respond to chemotherapy is uncertain. Practice styles range from continuous to intermittent therapy. At least two studies have found no benefit in continuing therapy indefinitely until disease progression in patients with chemosensitive disease.23, 24 Table 2 lists the declining frequency of treatments per cycle. These relative frequencies were estimated for the first 12 months from the observed data25 and beyond 12 months based on expert opinion.

After First-Line Therapy

In N9741, patients could receive second-line chemotherapy both before and after disease progression because of treatment toxicity or, more generally, at the physician's discretion. In N9741, the use and timing, but not the duration, of second-line therapy was recorded. Similar proportions of patients switched to a second-line therapy before progression: 26% in the FOLFOX cohort and 32% in the IFL cohort. The relative rates of second-line therapy after disease progression were imbalanced, as noted previously. Because data were not collected on treatment duration, response, or toxicity, these factors were inferred from data observed in randomized trials of the respective agents as second-line treatment.26, 27

After disease progression, patients were treated at their physician's discretion. Because the FOLFOX cohort had a longer survival postprogression, they may incur additional costs. Palliative care costs may be viewed in aggregate or per day. A constant daily cost, estimated from studies of the lifetime costs of colon carcinoma,29, 30 was used in the base case.

Role of the Funding Sources

The funding sources had no role in the collection, analysis, or interpretation of the data or in the decision to submit the report for publication.

RESULTS

Table 7 compares the model's simulated results with the results observed in N9741. The model slightly underestimated progression-free survival at 6 months and overestimated it at 1 year. Figure 2 presents the simulated versus observed survival curves for overall survival. The differences in survival at 6 months, 1 year, 2 years, and 3 years were < 2%. At 5 years, it was projected that 4.1% of patients in the FOLFOX cohort and 1.2% of patients in the IFL cohort would be alive.

Table 7. Comparison of Model Projections with N9741 Results
VariableFOLFOXIFL
N9741ModelN9741Model
  1. FOLFOX: oxaliplatin and infusional fluorouracil; IFL: irinotecan and bolus fluorouracil.

Death at 60 days (%)2.62.64.54.5
Progression-free survival (%)    
 Median survival (mos)9.39.56.96.9
 Progression free at 6 mos7163.95454.2
 Progression free at 1 yr3242.02030.7
Survival at specific intervals (%)    
 Median survival (mos)19.519.515.014.9
 Survival at 6 mos90.089.382.083.5
 Survival at 1 yr72.071.759.060.5
 Survival at 2 yrs40.239.223.525.9
 Survival at 3 yrs21.919.49.69.8
 Survival at 5 yrs4.11.2
Progression to death (mos)    
 Mean 9.6 8.0
 Median 6.7 5.5
No. of treatment cycles given (2-wk or 6-wk cycles)11.413.04.44.4
Figure 2.

This chart compares the survival curve from the N9741 randomized comparison with that of the model used in the current study. The lead survival point was 3 years in N9741 and was projected at 5 years with the model. FOLFOX: oxaliplatin and infusional fluorouracil; IFL: irinotecan and bolus fluorouracil.

Table 8 lists the primary cost-effectiveness results. The incremental benefit of FOLFOX was an increased mean survival of 5.1 months. The 3%-discounted, current value of the benefit was 4.4 months. FOLFOX had greater total costs. Considering both first-line and second-line treatment, FOLFOX cost $67,664, and IFL cost $42,247, for a net increase of $25,417 per patient. Palliative costs also were greater with FOLFOX at $4106 per patient. The total cost difference between cohorts was $29,523 per patient. Given the 4.4 month (0.37 yr) survival difference, the ICE was $80,407 per LY.

Table 8. Baseline Efficacy and Cost-Effectiveness Projections
Endpoint (at 5 yrs)FOLFOXIFLBenefit or differencea
  • FOLFOX: oxaliplatin and infusional fluorouracil; IFL: irinotecan and bolus fluorouracil.

  • a

    Benefits and costs were discounted at 3% per year.

Effectiveness   
Survival (mos)19.815.44.4
Costs   
 Initial semipermanent venous access$1863$559$1304
 First-line treatment$53,223$29,902$23,321
 First-line toxic deaths$358$630$272
 Nonfatal toxicities$3778$4876$1098
 Second-line treatment$8442$6326$2162
 Palliative care for advanced disease$27,029$22,923$4106
 Total costs$94,693$66,231$29,523
 Incremental $ per life yr  $80,407

The FOLFOX strategy was more expensive primarily because of the costs of oxaliplatin and treatment administration (Table 3). For patients with no complications, the projected treatment cost per 6-week interval was 71% higher with FOLFOX. The cost per 6-week interval included an additional $3015 for chemotherapy delivery, including infusion pump expenses, and $2918 per patient for oxaliplatin instead of irinotecan. The chemotherapy delivery costs of FOLFOX reflect an estimated 229% increase compared with the bolus administrations of IFL.

Sensitivity Analysis

Table 9 shows sensitivity analyses of key elements. The most important analyses were to explore the uncertainty around the relative efficacy of FOLFOX. By using the limits of the 95% confidence interval for time to progression (relative risk = 0.61–0.89) from N9741, the projected survival benefit ranged from 0.54 years to 0.22 years with an ICE of $59,250–121,220 per LY.

Table 9. Sensitivity Analyses
 Benefit (yrs)ICE ($ per life yr)Change in ICE ($ per life yr)
  • ICE: incremental cost effectiveness; FOLFOX: oxaliplatin and infusional fluorouracil; PFS: progression-free survival; IFL: irinotecan and bolus fluorouracil.

  • a

    Quality-adjusted life year projection used a utility value of 0.82 for time receiving first-line treatment and 0.5 for second-line treatment or palliative care. No additional quality-of-life decrement for neuropathy was included.

  • b

    A daily risk of noncolon carcinoma death in older patients was equal to a 10% annual morality with no change in FOLFOX relative efficacy.

  • c

    If oxaliplatin and irinotecan costs are calculated per mg dispensed versus per vial opened.

Base case0.3780,410 
Relative efficacy of FOLFOX in PFS   
 0.610.5459,250−21,160
 0.890.22121,22040,810
Revised IFL strategy: Dose reduced to 100 mg/m2, lowered 60-day death to 2.7%, and equal (65%) frequency of second-line therapy and survival from progression to death0.2584,7804370
Impact of utility scoresa0.26111,89031,480
Cost per PFS†0.2789,0808670
Palliative care   
 $0 per day0.3769,220−11,190
 $200 per day0.3791,590+11,180
Older patientsb0.3192,65012,240
Shorten time frame to 3 yrs0.2894,30013,890
Semipermanent venous access for IFL   
 0%0.3781,5801170
 100%0.3777,660−2750
Treatment delivered if PFS   
 100% treated if PFS for 12 mos0.37117,91037,500
 100% treated if PFS indefinitely0.37222,200141,790
No chemotherapy vial wastagec0.3771,630−8780
Drug cost thresholds   
 Oxaliplatin cost at $20,000 per life yr ($ per mg)4.02 −76%
 Oxaliplatin cost at $50,000 per life yr ($ per mg)10.44 −38%
 Oxaliplatin cost at $100,000 per life yr ($ per mg)21.18 +25%
 Irinotecan cost at $50,000 per life yr ($ per mg)11.73 +80%

Modified IFL

A series of changes made reflected posttrial IFL dosage and patterns of second-line chemotherapy. First, the IFL dose was reduced from 125 mg/m2 to 100 mg/m2; second, increased use of second-line chemotherapy after IFL leading to equivalent postprogression survival was allowed; and third, if IFL patients were treated with second-line oxaliplatin, then it was assumed that they would receive FOLFOX and, thus, would incur the costs of a semipermanent venous access catheter (if one had not already been inserted). These combined changes (with no change in the efficacy of FOLFOX) reduced the overall survival benefit of FOLFOX to 0.25 years but only increased the ICE to $84,780 per LY.

Palliative Care

Although daily palliative care costs are unlikely to vary by initial treatment, it was assumed initially that each extra day of survival would incur added costs. If palliative care costs are an aggregate lifetime cost, then these can be ignored. Excluding palliative care costs, the ICE decreased $12,200 per LY.

Venous Access

The ICE ratio was only $4000 per LY lower if none of the IFL patients needed an initial semipermanent venous access procedure compared with 100% of IFL patients.

FOLFOX6 versus FOLFOX4

FOLFOX6 is the more frequently used oxaliplatin regimen in the United States. In the current sensitivity analysis, it was assumed that the efficacy of FOLFOX4 and FOLFOX6 is equivalent but that the costs are reduced. In FOLFOX6, the oxaliplatin dose is increased from 85 mg per m2 to 100 mg per m2, leucovorin is given only once, and there are no second-day infusions of leucovorin and 5FU. The projected net cost reduction in the FOLFOX therapy is $320–460 (6–11%) per 2-week treatment, depending on the amount of oxaliplatin vial wastage. By using FOLFOX6, the base-case ICE was $66,000 per LY.

QOL and Patient Age

The base-case model only considered survival and did not give any QOL benefit for response rate or penalty for progression. By using utility scores of 0.82 for progression-free states and 0.5 for postprogression states, the net benefit declined to 0.26 quality-adjusted years, and the ICE ratio increased to $111,890 per LY. Upon consideration of an older age cohort with a higher risk of noncancer death (10% per yr) but no difference in relative responsiveness to FOLFOX or IFL, the results changed minimally.

Cost per Progression-Free Survival

A simplified model limited the assessment to time before disease progression. If costs and survival after progression were ignored, then the benefit of FOLFOX declined to about 3 progression-free months, and the ICE ratio rose slightly to $89,080 per progression-free year.

Three-Year Time Frame

If the time frame was limited to 3 years, corresponding to the current N9741 median follow-up, then the survival benefit decreased by approximately 25%, and the ICE increased by approximately $13,000 per LY.

Treatment Duration

More patients on FOLFOX than patients on IFL stop therapy before disease progression because of paresthesias or neutropenia. If FOLFOX patients continue some chemotherapy outside a trial, then treatment-associated costs will be higher with FOLFOX, and the ICE will increase. To assess the impact of this, two scenarios were explored. First, if all FOLFOX or IFL progression-free patients are treated in the first 12 months, then the ICE increases to $117,910 per LY. Second, if 100% of progression-free patients are treated indefinitely, then the ICE markedly increases to $222,200 per LY. In each of these scenarios, the relative dose-intensity and hospitalization rates are not changed.

Drug Costs

The model was quite sensitive to oxaliplatin and irinotecan's acquisition costs. By using common willingness-to-pay thresholds, the requisite changes in the cost of oxaliplatin were projected. At a threshold of $20,000 and $50,000 per LY, oxaliplatin's price would have to decrease 76% and 38%, respectively, to cross below these thresholds. If a $100,000 per LY threshold were acceptable, then the oxaliplatin's acquisition cost could increase 25%. If the relative price ratio of oxaliplatin and irinotecan is maintained but an organization could acquire both drugs at a 30% discount without a change in administrative costs, then the ICE would decrease to $68,010 per LY. If chemotherapy vial wastage could be avoided, then the ICE would decrease by 12% to $71,630 per LY, because there was more opened but unused oxaliplatin than irinotecan. If a FOLFOX6 schedule (with no second-day visit for FU/LV) was used and was assumed to be equally effective, then the ICE ratio would decrease to $70,100 per LY.

DISCUSSION

The FOLFOX combination, compared with IFL, in the first-line treatment of metastatic colorectal carcinoma extends progression-free survival and overall survival. The absolute benefit is approximately 2 months for progression-free survival and approximately 4–5 months for overall survival. In the current report, the ICE of FOLFOX versus IFL has been estimated using all relevant, anticipated, direct medical costs. These costs include the drug and delivery costs for first-line and second-line chemotherapy, major toxicities, and late palliative care. The baseline results indicate that, over a 5-year period, FOLFOX would provide a survival benefit of 0.37 years or 4.4 months at an ICE of $80,410 per LY.

Sensitivity or ‘what if’ analysis explored numerous factors, with three considerations emerging as the most crucial. The first was an uncertainty regarding the size of the benefit of FOLFOX. By using the 95% confidence interval range observed in N9741, the survival benefits ranged from 0.22 years to 0.54 years, with corresponding ICE ranging from $121,220 to $59,250 per LY.

The second consideration was that, in N9741, the number of cycles of initial therapy delivered was substantially lower than the observed progression-free survival. This was due primarily to treatment-associated toxicity, especially FOLFOX-associated neuropathy. The median duration of FOLFOX therapy in N9741 (23.9 weeks) was almost identical to that observed in another first-line trial using FOLFOX (de Gramont et al., 2000,31 24.0 weeks). In N9741, the median duration of IFL was similar to those of other IFL randomized trials (N9741, 23.6 weeks; Saltz et al., 2000,3 24.0 weeks; Hurwitz et al., 2004,32 27.6 weeks). Because first-line treatment costs are higher with FOLFOX, if a higher percentage of patients who are progression-free are treated, then the ICE will rise. For example, if, in the first 12 months after starting either approach, all progression-free patients continue to be treated, then the ICE would rise substantially to $117,910 per LY.

The third consideration, possibly because of the difference in the relative use of second-line therapy in N9741, was that FOLFOX patients had greater gains in survival than actually were projected by the difference in progression-free survival. In this model, the survival difference was projected at 1.6 months. Although no data were collected on palliative care costs in N9741, the economic assessment explored their impact. If these costs were ignored or doubled, then the ICE ratio fell or rose by $10,000–12,200 per LY.

In the later stage of N9741 and in current practice, patients in the United States who are treated with initial IFL or FOLFOX are likely equally to cross over to the alternative as the second-line therapy. By using a lower irinotecan dose with a concurrent lower acute death risk with IFL, and using a realistic 65% rate of second-line chemotherapy leading to a similar progression to death, projected survival benefit and ICE ratios were revised. In this scenario, the benefit of FOLFOX decreased to 3 months, but the ICE only changed to $84,780 per LY.

Cost-effectiveness analysis is a tool for improving the efficiency of healthcare guide by shifting utilization away from procedures that produce little benefit at high cost. Operationally, health or government insurance programs can promote or restrict coverage to “cost-effective” medical services by covering only services with a cost-effectiveness ratio that is equal to or less than a cut-off (threshold) value. There is no consensus regarding the appropriate threshold value of additional costs per unit of benefit that society is willing to pay for a medical treatment. A range that is used commonly as a benchmark for ICE is $50,000–100,000 per LY, which is based on historic precedent that was set by dialysis in patients with end-stage renal disease.15, 33–35 However, these thresholds have not increased in > 2 decades. Some suggest that the current United States decision makers use a threshold approaching $200,000 per LY.36 In Britain, a threshold of £30,000 (approximately $55,600) per unit of health outcome has been the norm in recent National Institute of Clinical Excellence decisions.37

Nonetheless, ranking various interventions in a “league table” can be a helpful means of comparison; but this, too, has been criticized because of methodological differences across studies, especially when the unit of benefit is quality-adjusted LYs.19 Although the rigid application of cost-effectiveness cut-off values would be nearly impossible, cost-effectiveness analyses play a major role in guiding public policy decisions in countries with national heath insurance.19 The current analysis suggests that, as first-line therapy, FOLFOX provides a benefit to patients with ICE in the acceptable range in the base-case analysis and in most sensitivity analyses. Until the relative benefit of FOLFOX falls below 17% (a hazard > 0.83), the ICE threshold of $100,000 per LY is not crossed.

The need to estimate both the anticipated costs for treatment toxicity and the duration of second-line chemotherapy was an unavoidable limitation of this analysis. These limitations did not affect the analysis significantly, because the relative impact of variations in these numbers should be minimal across both strategies.

For the current analysis, we used overall survival, not QOL-adjusted survival, as the unit of benefit. In support of our approach, QOL data prospectively collected in N9741 identified no differences between strategies in either a single-item or a 12-item symptom distress QOL scale. The greater rate of neuropathy with FOLFOX appeared to be counterbalanced by the greater rate of diarrhea with IFL. To explore this concept further, utility scores were estimated from the available rating scale QOL scores for progression-free survival and were assigned for disease progression. The ICE ratio increased by almost 40% to $111,890 per quality-adjusted LY, as expected. In addition, the current analysis did not include any added costs for patients developing neuropathy or diarrhea who ultimately did not require hospitalization for these side effects. This approach seemed reasonable, because no effective treatment for the neuropathy is known, and the treatment that is used commonly, adding calcium and magnesium to the infusions, adds minimal costs.

The method used required modeling and assumptions about noncollected, relevant data points. We believe that the current report clearly follows the current best practices for performing and reporting economic analyses.38, 39 Although the exclusion of nondirect medical care costs incurred by the patient or their family underestimates total societal costs, there is little reason to suspect that this exclusion would alter the incremental cost between strategies. Each combination is given on an outpatient basis, per 6-week interval, and the number of days on which a patient would be treated or would require monitoring of laboratory studies is approximately the same.

Ideally, costs of new therapies for metastatic colorectal carcinoma should be compared sequentially40 as a menu or ladder of interventions ranging from best supportive care, to single-agent FU/LV or capecitabine, to doublets or triplets of cytotoxic regimens like FOLFOX and IFL with or without monoclonal antibodies, such as bevacizumab and cetuximab. However, few economic analyses have been published, and even fewer have used a United States perspective.41–43

A 2001 British analysis that was prepared for their National Institute of Clinical Excellence found a base-case, estimated ICE of £27,763 for IFL compared with FU/LV and £23,047 for a FOLFOX-type regimen for each quality-adjusted LY compared with FU/LV.43 In a controversial position that was based primarily on consideration of the ICE ratio, only FU/LV was recommended as first-line therapy.44

For the current report, we used data from the definitive trial supporting FOLFOX as a United States standard of care as first-line treatment for patients with metastatic colon carcinoma. Currently, the National Comprehensive Cancer Network, a prominent source of United States practice guidelines, offers several options for such patients45: FOLFOX, FOLFIRI, and FOLFOX/FOLFIRI with bevacizumab. The current results support the relative cost-effectiveness of FOLFOX compared with IFL. Currently, there are no published cost-effectiveness reports addressing the alternative strategies. In conclusion, the cost-effectiveness model used in the current report estimated that the ICE of FOLFOX, compared with IFL, falls into the upper range of commonly accepted oncology interventions in the context of the United States healthcare system.46

Acknowledgements

The authors acknowledge and thank the North Central Cancer Treatment Group (NCCTG) for allowing use of the N9741 data to verify model assumptions and the NCCTG and the cooperative groups that participated in N9741, with cochairs from each group as follows: Roscoe F. Morton, NCCTG; Ramesh K. Ramanathan, Eastern Cooperative Oncology Group; Stephen K. Williamson, South West Oncology Group; and Brian P, Findlay, National Cancer Institute of Canada-Clinical Trials Group.

Ancillary