e17552 Background: The Eisai Metastatic Breast Cancer Study Assessing Physician’s Choice Versus E7389 (EMBRACE) established clinical benefits. This study evaluates its translation into comparative economic value. Methods: Because of superior survival benefits and a non-inferior safety profile, we use a cost-effectiveness analysis. Costs include medication and administration costs, cost of toxicity management, and indirect cost. The primary endpoint is the ratio of incremental means of costs and quality adjusted life years (QALY) yielding an incremental cost effectiveness ratio (ICER). Partitioned survival analyses were evaluated using a log-logistic function for overall survival, progression free survival, response duration, and toxicity time. Health state utilities or quality weights are applied to each component and aggregated. Sensitivity analyses assessed the influence of variability in a number of parameters. Results: Overall survival based on the log-logistic extrapolation was 686 days in the eribulin group compared to 550 days in the TPC group for a difference of 136 days. Mean time without progressive disease was 214 days and 160 days, respectively, for a difference of 59 days. Time spent with Grade 3 and 4 toxicity were 3.62 and 2.25 days in the eribulin group versus 2.25 and 0.98 days in the TPC group. Response times were 15.62 and 9.64 days, 5.98 days longer in the eribulin group. After weight components of overall survival with corresponding utilities and converting to years, the QALY were 1.166 in the eribulin group compared to 0.926 in the TPC group for a mean incremental improvement of 0.24 years. Mean treatment costs were $14,302.80 AUD and $1,672.02 AUD for a difference of $12,630.78 AUD. Total incremental cost is $13,794.35 AUD. The ICER with and without discounting is $48,134.29 AUD and $45,770.97 AUD, respectively. Survival time and drug cost were the most influential variables on the ICER. Conclusions: At a threshold of $50,000 AUD per QALY, eribulin is good value at $48,134 AUD per QALY. Clinical trial information: NCT00388726.

1050^ Background: In a phase III trial comparing eribulin (E) vs. capecitabine (C) in pts with locally advanced or MBC, a trend for improved OS was observed but a statistically significant superiority was not demonstrated with E vs. C for OS or PFS. The AE profiles were consistent with known side effects. We now report QoL results from this trial. Methods: Pts received eribulin mesylate 1.4 mg/m2 on Days 1 and 8, or C 1.25 g/m2 BID orally on Days 1-14, of a 21-day cycle. Eligible pts had received prior therapy including an anthracycline and taxane, and were receiving study drug as 1st-, 2nd-, or 3rd-line therapy for advanced disease. QoL, a secondary objective, was assessed using EORTC QLQ-C30 and QLQ-BR23 questionnaires at baseline, 6 weeks, 3, 6, 12, 18, and 24 months after starting treatment (or until progressive disease or treatment change), and at unscheduled visits. Longitudinal analyses were carried out using weighted generalized estimating equations adjusted for non-random attrition due to death within 12 months. Model covariates were time (visit), region, and baseline QoL. The primary endpoint was change from baseline for Global Health Status (GHS)/overall QoL; exploratory endpoints were change from baseline for each functional domain, and signs/symptoms. Results: 1,102 pts were randomized (E 554; C 548). GHS/QoL scores were low at baseline for E (56.3) and C (54.7) on a scale of 0 (worse) to 100 (best). GHS/QoL and cognitive functioning improved significantly more in pts receiving E vs. C, (6.5 [p=0.048] and 15.3 [p<0.001], respectively). Emotional functioning improved significantly for pts receiving C vs. E (3.3; p=0.033). Pain was comparable at baseline, and was lower at subsequent visits with both treatments. Patient-reported signs/symptoms in favor of E included nausea and vomiting (E1.9; p=0.043) and diarrhea (-3.7; p=0.001); systemic side effects (5.2; p<0.001) and upset by hair loss (9.3; p=0.023) favored C. Conclusions: GHS/QoL scores improved more in pts receiving E than C. E showed advantages in terms of gastrointestinal effects while C had advantages in relation to hair loss. Clinical trial information: NCT00337103.

e17576 Background: The EORTC-QLQ30 is the core quality of life (QOL) instrument developed by the European Organization for Research and Treatment in Cancer. We assessed content validity for the signs and symptoms component in an investigational trial. Methods: We use a Phase II clinical trial where 291 women with locally advanced or MBC were required to have received prior treatment with an anthracycline, a taxane, and capecitabine. EORTC-QLQ30 was collected every second cycle, before drug administration and before patients were informed of their tumor assessment. We use the reported adverse events (AE) for those signs and symptoms to test the correlation between reported AEs and responses to the QoL questionnaire. The sum of the incidence by patient is generated as well as average scores for each sign and symptom. An endpoint for respondents positively reporting signs and symptoms without any AE is also evaluated (False-Positive). Statistically significant positive correlations between AEs and signs and symptoms as well as negative correlations for False-Positive establish content validity. Results: 283 women provided 1068 QOL assessments. Their general health status was 68.88 (P<0.0001) on a 100 point scale, improving by 7.90 (P=0.0153) when responding to eribulin. Nausea and vomiting were positively correlated with the reported incidence of nausea as AEs, 0.126 (P=0.03) and vomiting (P<0.01); false positives were negatively correlated, (-0.329; P<0.01 and -0.352; P<0.01, respectively). Constipation also satisfied the criteria for content validity, 0.189 (P<0.01) for correspondence and (-0.394; P<0.01) for lack of False-Positives. Diarrhea had a correspondence of 0.292 (P<0.01) and specificity of (-0.260; P<0.01). Dyspnea had a 0.226 (P<0.01) correspondence and (-0.349; P<0.01) specificity. Insomnia failed the criteria. Upset by hair loss was weakly correlated with alopecia but very specific (-0.479; P<0.01). Conclusions: Signs and symptoms as measured by the EORTC-QLQ-BR32 questionnaire have content validity in an investigational setting and they contribute clinically meaningful information to reported AE incidence data. Clinical trial information: NCT00246090.

1055^ Background: Key properties of QoL instruments in a clinical trial setting are reliability, the ability to detect a change, and content validity. Using data from the Study 301 phase III breast cancer trial, we report here content validity and ability to detect a change for the EORTC QLQ-C30 and breast cancer-specific QLQ-BR23 questionnaires. Methods: Patients with locally advanced or metastatic breast cancer were randomized to 21-day cycles of either eribulin mesylate 1.4 mg/m2 given on Days 1 and 8, or capecitabine 1.25 g/m2BID orally on Days 1-14. QoL questionnaires were completed at baseline and at 6 weeks, 3, 6, 12, 18, and 24 months. Objective tumor response was evaluated (complete response [CR]; partial response [PR]; stable disease [SD]; progressive disease [PD]). Univariate and multivariate longitudinal analyses using weighted generalized estimating equations were employed to assess the responsiveness of the QoL scales to objective tumor response. Results: 1,102 patients were randomized (554 eribulin, 548 capecitabine). Global health status (GHS)/QoL scores were low at baseline (55). GHS/QoL scores were highest for patients with CR (70.8), followed by those patients with PR (63.5), SD (60.5), and PD (58.1). Physical functioning followed the same pattern: CR (98.3); PR (79.1); SD (72.8); PD (71.0). Role and social functioning scores were also responsive. Pain increased, while fatigue and body image worsened, with poorer tumor responses. Using the weighted generalized estimating equations, there were improvements in physical (34.78; p<0.01), cognitive (27.29; p<0.01), and social (22.04; p<0.01) functioning, and future perspective 11.47 (p<0.01), in patients who responded (CR and PR) to treatment compared with non-responders. Pain decreased significantly by 28.62 (p<0.01) on a 0-100 scale. Patients who responded also gained appetite and had fewer breast symptoms. Conclusions: These results suggest content validity of the EORTC QLQ-30 and QLQ-BR23 questionnaires as they correlate with changes in objective tumor assessments. Clinical trial information: NCT00337103.

Abstract Withdrawn by Author. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P1-12-08.

Objective: Determine the prevalence and costs of rheumatoid arthritis (RA) during three consecutive years: 2004, 2005, and 2006. Methods: Medical Expenditure Panel Survey was used for persons with RA. Regressions estimate health care costs and income loss. Absenteeism and age-adjusted workforce participation compared means and rates. Results: The prevalence of RA was 0.40% in 2004, 0.44% in 2005, and 0.43% in 2006. Health care cost associated with RA was $4422, $2902, and $1882 (all P < 0.01) in 2004, 2005, and 2006, respectively. Rheumatoid arthritis sufferers were employed, 36.8%, 39.5%, and 44% compared with 70.5%, 69.8%, and 71%. Individuals with RA also missed more days of work, 4.86 in 2004 (P = 0.04), 1.70 in 2005 (P = 0.22), and 2.99 in 2006 (P = 0.04). Rheumatoid arthritis reduced income by $2404 (P = 0.03), $2207 (P < 0.001), and $1212 (P = 0.002). Conclusions: Costs of RA are considerable.

Abundance of zero values is commonly observed in cost data resulting in skewed distribution. This analysis measured the inpatient cost and workplace absenteeism associated with low back and neck pain and demonstrated the consequences of ignoring zeros in inpatient cost and unreported absenteeism. We used employer-based claims from the Thomson Marketscan© Research Database (2007), a database representing approximately 100 payers of insured employees containing health and productivity management (HPM) and health care utilization data. Adult insured employees with continuous eligibility in 2007 were included. The ICD-9 codes identified medical conditions including low back and neck pain without (nociceptive pain, NOCI) or with a neuropathic component (mixed pain, MIXED). Ordinary least squares (OLS) and two-stage Tobit analyses evaluated the marginal inpatient costs while OLS and Heckman's Selection Bias (HSB) were applied to absenteeism data. Estimated inpatient costs and absenteeism using OLS versus two-stage techniques were compared. A total of 2,046,332 employees (male=59.2%; mean age 40.2±11.6 years) were analyzed. Hypertension (9.8%), NOCI (9.5%), diabetes (3.7%), MIXED (3.0%) and depression (1.1%) were the most prevalent medical conditions among these employees. 1,976,952 (96.6%) employees had no inpatient episodes, thus, with no inpatient costs. Mean inpatient cost for the entire study population was $537.45 (median=$0) versus $15,851.93 (median= $8,302.20) among those with inpatient episodes. The incremental inpatient costs associated with MIXED and NOCI were $1,333.02±26.67 and $328.36±15.63 using OLS versus $2,478.97 [95%CI: 2,148.50 – 2,811.16] and $1,242.41 [95%CI: 1,020.10 – 1,469.18] using the two-stage Tobit. Unreported absenteeism occurred in 80% of the employees. Annual absenteeism associated with MIXED and NOCI using OLS were 5.25±0.21 and 4.06±0.35 compared to 45.92±1.06 and 16.33±2.01 hours using the HSB technique. Ignoring zeros in cost data and unreported absenteeism may result in substantial underestimation of inpatient cost and workplace absenteeism associated with low back and neck pain.

To demonstrate replication of the quantification of relationships between surrogates and endpoints as well as reconciliation with previous epidemiological studies; original studies for heart rate as a surrogatefor all-cause mortality, pain management and gastrointestinal adverse events, and treatment for diabetes and HbA1c and HbA1c and complications. For heart rate, three epidemiological studies from three countries using a Weibull survival analysis and Generalized Estimating Equations were used; namely, the Coronary Artery Surgery Study (CASS), the Copenhagen City Heart Study (CCHS) and the General Practitioner Research Network (GPRN). These equations reproduced a meta-regression and meta-analysis of all available placebo-controlled clinical trials with heart rate as a prognostic factor for all-cause mortality. For pain, data consisted of 2005 Health Care Utilization Project (HCUP) and Premier. Logistic regressions were used to obtain evaluate and compare odds-ratios. In diabetes, Generalized Estimating Equations (GEE) allowing serially correlated behavior with repeated HbA1c reading at variable frequencies and durations between their measurement. Heart is consistently prognostic for all-cause mortality. Moreover, its quantification is consistent, 0.00694 (P<0.001) in CASS and 0.00683 (P<0.001) in CCHS (1981-1983) and 0.00717 in CCHS (1991-1993) with the Weibull. With the GEE, the coefficient is 0.0268 (P=0.006) in GPRN, 0.0249 (P=0.008) in the meta-regression of controlled clinical trials, and 0.01595 in the GEE with CCHS data. All three equations reproduced the published clinical trials with odd-ratios within 1/100ths.Conditional odds-ratios were replicated in measure between the two datasets for fecal impaction, post-operative illeus, other bowel obstruction, vomiting and abdominal pain. The diabetic equations were replicated exactly in 3 countries, treatment and HbA1c and complications with coefficients within 1/100th in patients with newly diagnosed T2DM. These are three studies where the quantification of the relationship between a surrogate and and endpoint have beed replicated with precision and subsequently applied to clinical trials.

BACKGROUND:Several early studies demonstrated that bile acid sequestrants were useful for lowering lipid levels in patients with hypercholesterolaemia and may also be useful for lowering glucose levels in patients with type 2 diabetes mellitus (T2DM) uncontrolled on existing treatment (metformin-, insulin- or sulfonylurea-based therapies).OBJECTIVE:This study modelled efficacy and safety data from the three clinical trials to evaluate the cost effectiveness to US Managed Care Organizations of add-on treatment with colesevelam for reducing diabetes-related complications.METHODS:Three randomized controlled trials in patients with T2DM and one in hyperlipidaemia established that colesevelam lowered both glycaemic and lipid parameters in adult patients participating in the studies. The validated 'diabetic risk equation' (DRE) and the 'LIPID cardiovascular risk equation' (LCRE) were used to translate the observed clinical benefits (surrogate markers related to T2DM [glycosylated haemoglobin {HbA(1c)} and fasting plasma glucose] and cardiovascular disease [low-density lipoprotein cholesterol {LDL-C}]). Performing an appropriate economic evaluation required the use of both the DRE and the LCRE. These equations parameterize the clinical efficacy measures as continuous, facilitating their application to clinical trial results as well as the replication of other well established epidemiological data. Tobit regressions were applied to a large commercially available managed care administrative claims database (2000-6), Integrated Health Care Services (IHCS), to evaluate the incremental costs associated with each type of diabetic complication. Costs were inflated to 2010 values using the Healthcare Consumer Price Index, while second- and third-year cost savings were discounted at 5% to the current year. Bootstrap sampling with 5000 samples of 100 patients per cohort was conducted, varying the number of events avoided as well as their associated cost.RESULTS:With established metformin-, insulin- or sulfonylurea-based therapies, the addition of colesevelam significantly reduced HbA(1c) by approximately 0.5% (p < 0.001) in all three studies. In addition, colesevelam reduced placebo-adjusted LDL-C by 12.8-16.7% (p < 0.001). Using the DRE and LCRE equations, the total savings from reductions in diabetes-related and cardiovascular events were $US3543, $US4074 and $US3855 for colesevelam added to metformin-, insulin- and sulfonylurea-based regimens in patients with normal lipid levels. After subtracting the cost of colesevelam, first-year savings were $US1326, $US1852 and $US1629 in the metformin, insulin and sulfonylurea studies, respectively, for patients with raised lipid levels.CONCLUSIONS:In adult patients with T2DM, the addition of colesevelam to metformin-, insulin- or sulfonylurea-based therapies significantly improves glycaemic control while also reducing LDL-C, and these improvements could translate into substantial cost reductions due to reductions in the rates of diabetes-related and cardiovascular complications.

Objective: The aim of this study was to quantify the costs and resource utilization associated with a relapse of schizophrenia or schizoaffective disorder. Methods: The study comprised a retrospective audit of data from 200 patients diagnosed with schizophrenia or schizoaffective disorder who were admitted to hospital for a relapse of their disorder in two mental health services in Australia between 1 June 2001 and 31 May 2002. Resource use and costing data were collected for 12 months before and 12 months after the hospitalization. Results: There was an increase in contacts per month and associated outpatient costs after the index admission which persisted for the full 12 month data collection period (total of AUD $637). There was also a total increase in hospital costs but this did not persist beyond the first 2 months of the follow-up period and is likely explained by the index admission. Conclusions: Increased healthcare resource utilization and costs results from relapse in patients with schizophrenia or schizoaffective disorder. An increase in service use and costs persist for a considerable time period after an episode of relapse.

Abstract Abstract 4524 Adverse event (AE) profiles and laboratory abnormalities are different amongst TKIs used as second line therapy in chronic myeloid leukemia. These differences are apparent between nilotinib and dasatinib, as evidenced by the required monitoring parameters that are contained within the products' labeling. AIM Ancillary costs often accrue beyond drug acquisition costs; this assessment translates required monitoring as per FDA approved product labeling for AEs and laboratory abnormalities into annual ancillary costs for dasatinib and nilotinib in the treatment of CML. METHODS Prescribing Informations (PI) of dasatinib and nilotinib were carefully reviewed to tabulate the required monitoring parameters and their recommended periodic timing during the initial one-year course of treatment. For example, due to the risk of myelosuppression with these agents, a complete blood count (CBC) is required for patients treated with either dasatinib or nilotinib at 2 weeks, 4 weeks and monthly thereafter, or 13 times in year one. Nilotinib also requires an ECG at baseline, 1 week, and periodically thereafter (assessed as every other month) throughout treatment due to the risk of QT prolongation. QT intervals should also be closely monitored in patients with hepatic impairment. Also with Nilotinib treatment, the following are recommended: blood tests periodically due to the risk of serum lipase elevation; hepatic function tests due to the risk of liver function abnormalities; and screening for electrolyte abnormalities. While fluid retention is associated with dasatinib therapy, it requires only careful observation. Patients who develop symptoms suggestive of pleural effusion such as dyspnea or dry cough should be evaluated by chest X-ray. A large, commercial managed care database based on more than 4 million records was used to assess the mean costs for each required test; those costs were multiplied by the required frequencies for the first year. RESULTS The average cost and standard deviations for CBC is $30.46±73.75. The average costs for an ECG, electrolyte panel and hepatic panel are $174.40±74.71, $13.60±15.91, and $21.20±17.30. Cost data are skewed to the right inflating the standard deviations. First year annual costs associated with required monitoring for myelosuppression is $426.44 for both dasatinib and nilotinib. Additional monitoring costs for nilotinib in the first year included $81.60 for electrolyte panels, $2,092.80 for ECGs and $275.60 for hepatic panels. Total costs are $2,721.29 and $426.44 for monitoring parameters for nilotinib and dasatinib, respectively. CONCLUSION Drug acquisition cost rarely provides a complete picture of treatment cost. Nilotinib requires nearly $500 a month just for required monitoring due to greater risks for adverse events and laboratory abnormalities. Disclosures: Simons: Bristol-Myers Squibb: Consultancy. Maclean:Bristol-Myers Squibb: Employment. Cairns:Bristol-Myers Squibb: Employment.