Search

BackgroundWhile therapeutic bronchoscopy has been used to treat malignant central (CAO) airway obstruction for >25 years, there are no studies quantifying the impact of therapeutic bronchoscopy on long-term quality-adjusted survival.MethodsWe conducted a prospective observational study of consecutive patients undergoing therapeutic bronchoscopy for CAO. Patients had follow-up at 1 week and monthly thereafter until death. Outcomes included technical success (ie, relief of anatomic obstruction), dyspnoea, health-related quality of life (HRQOL) and quality-adjusted survival.ResultsTherapeutic bronchoscopy was performed on 102 patients with malignant CAO. Partial or complete technical success was achieved in 90% of patients. At 7 days postbronchoscopy, dyspnoea improved (mean ∆Borg-day-7=−1.8, 95% CI −2.2 to −1.3, pConclusionsTherapeutic bronchoscopy improves HRQOL as compared with baseline, resulting in approximately a 5.8% improvement in HRQOL per day of life. The risk-benefit profile in these carefully selected patients was very favourable.Trial registration numberResults; NCT03326570.

Background: Acalabrutinib (Acala) is a next-generation, highly selective, covalent BTK inhibitor (BTKi) approved for the treatment of CLL. The phase 3 ASCEND study demonstrated significant progression-free survival (PFS) benefit with Acala vs investigator's choice of chemoimmunotherapy (bendamustine + rituximab [BR]) or pi3K inhibitor (idelalisib) + rituximab (IR) in R/R CLL patients (pts) (Ghia et al. J Clin Oncol. 2020). In ELEVATE-RR, Acala demonstrated non-inferior PFS vs a first-generation BTKi (ibrutinib [Ibr]) in R/R CLL pts with high-risk genomics (Byrd et al. J Clin Oncol. 2021). Acala was well tolerated in both trials. Quality-adjusted Time Without Symptoms and Toxicity (Q-TWiST) analysis balances risks (toxicity) and benefits (prolonged survival without symptoms of disease progression or adverse events [AEs]) of oncology treatments. We conducted a Q-TWiST analysis using data from these trials to assess risk-benefit of Acala vs comparators among R/R CLL pts. Methods: Patient survival time with 16 months (ASCEND) and 41 months (ELEVATE-RR) of median follow-up data were partitioned into 3 states: time with toxicity before disease progression (TOX); time without progression or toxicity (TWiST); and time from disease progression until death or end of follow-up (relapse [REL]). Toxicity was defined as grade 3/4 AEs and, for the comparison vs Ibr, an additional definition of toxicity included AEs of any grade with frequency ≥10% (AE ≥10%). Sensitivity analyses with toxicity defined as grade 2-4 AEs will be presented. Kaplan-Meier survival curves were constructed for TOX, PFS, and overall survival (OS). For the TOX curve, event time for each pt was calculated by summing days with AEs (per toxicity definition) before disease progression, with no pt censored. Restricted means were calculated for the longest PFS time among treatment arms. Restricted mean TWiST duration was estimated using the difference between restricted means of PFS and TOX. Restricted mean REL duration was estimated using the difference between restricted means of OS and PFS. Q-TWiST was calculated by summing weighted restricted mean durations of TOX, TWiST, and REL. Utility of 1.0 for TWiST and 0.5 for TOX and REL (Gelber et al. Am Statistician. 1995) were applied. Sensitivity analysis was conducted using utility measures from trial data. Two-sided p Results: Among Acala vs IR/BR-treated pts, significantly longer duration of TWiST (mean diff, 3.58 months [95% CI, 2.42, 4.74]) and shorter duration of REL (mean diff, −2.51 months [95% CI, −3.50, −1.60]) and TOX (mean diff, −0.73 months [95% CI, −1.24, −0.24]) were found, indicating Acala vs IR/BR-treated pts had longer time without disease progression and toxicity and shorter time with toxicity. Q-TWiST was significantly different (1.96 months [95% CI, 1.13, 2.81]) between Acala (17.48 months) and IR/BR (15.52 months). In the Acala and Ibr comparison, differences in durations of TWiST and TOX varied based on the toxicity definition while the difference in REL durations remained consistent (mean diff, 0.37 months [95% CI, −1.82, 2.60]; p=0.33]). With the toxicity definition of AE ≥10%, Q-TWiST was significantly different (2.56 months [95% CI, 0.19, 4.83]) for Acala (33.02 months) vs Ibr (30.46 months). When toxicity was defined as grade 3/4 AEs, Q-TWiST had higher estimates for Acala but the difference was not statistically significant (1.28 months [95% CI, −1.34, 3.80]). In the sensitivity analysis, Q-TWiST gains were not statistically significant, most likely because the study was not designed to collect EQ-5D data post-treatment discontinuation. Conclusions: Q-TWiST analyses integrate efficacy, safety, and quality of life into a more appropriate measure useful for detailed benefit-risk assessment of cancer treatment. This analysis demonstrated that Acala treatment is significantly superior to IR/BR treatment. Quality-adjusted survival gains of Acala over Ibr varied by toxicity definition, but was numerically higher for Acala. Overall, Acala demonstrated a favorable benefit-risk profile vs different comparators in R/R CLL. Figure 1 Figure 1. Disclosures Seymour: Mei Pharma: Honoraria, Membership on an entity's Board of Directors or advisory committees; Celgene: Consultancy, Research Funding, Speakers Bureau; Gilead: Honoraria, Membership on an entity's Board of Directors or advisory committees; BMS: Honoraria, Membership on an entity's Board of Directors or advisory committees; AstraZeneca: Honoraria, Membership on an entity's Board of Directors or advisory committees; Janssen: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding; AbbVie: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau; F. Hoffmann-La Roche Ltd: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau; Morphosys: Honoraria, Membership on an entity's Board of Directors or advisory committees; Sunesis: Honoraria, Membership on an entity's Board of Directors or advisory committees; Takeda: Honoraria, Membership on an entity's Board of Directors or advisory committees. Gaitonde: AstraZeneca: Current Employment, Current holder of stock options in a privately-held company, Research Funding. Emeribe: AstraZeneca: Current Employment, Current holder of stock options in a privately-held company. Cai: Celgene Corporation: Ended employment in the past 24 months; Google: Current equity holder in publicly-traded company; AstraZeneca: Current Employment, Current equity holder in publicly-traded company. Mato: Johnson and Johnson: Consultancy, Research Funding; MSKCC: Current Employment; TG Therapeutics: Consultancy, Other: DSMB, Research Funding; Genentech: Consultancy, Research Funding; AbbVie: Consultancy, Research Funding; Pharmacyclics LLC, an AbbVie Company: Consultancy, Research Funding; Genmab: Research Funding; Acerta/AstraZeneca: Consultancy, Research Funding; LOXO: Consultancy, Research Funding; Nurix: Research Funding; BeiGene: Consultancy, Research Funding; Janssen: Consultancy, Research Funding; Sunesis: Consultancy, Research Funding; DTRM BioPharma: Consultancy, Research Funding; Adaptive Biotechnologies: Consultancy, Research Funding; AstraZeneca: Consultancy.

INTRODUCTION The impact of treatment on both the quality and the quantity of life, i.e. the ‘net clinical benefit’, should be considered to facilitate shared decision making. Two methods that combine survival and health-related quality of life (HRQoL) data: Quality Adjusted Effect Sizes (QASES) and Joint Modelling (JM) were applied to gain insight in the net clinical benefit. METHODS The net clinical benefit in one RCT (EORTC 26951 comparing radiotherapy (RT) + PCV chemotherapy versus RT alone) was calculated as a proof of concept for other trials. With the QASES method, effect sizes for differences in survival and HRQoL between treatment arms were calculated. JM allows simultaneous modeling of a longitudinal outcome (HRQoL), and a time-to event outcome (survival). HRQoL scales/items that were selected for primary analysis in the main study were also selected for this analysis: fatigue, global health, social functioning, communication deficit, seizures, physical functioning, and nausea/vomiting. RESULTS 288/386 patients completed baseline HRQoL forms and were included in the analysis. Overall survival (OS) was significantly longer with combined treatment (42.3 vs. 30.6 months). The percentage of patients who experienced a clinically relevant deterioration (≥10 points) in nausea/vomiting, fatigue, social functioning and global health up to one year after treatment compared to baseline was larger in the RT+PCV arm. QASES corresponded to a reduction in the median OS difference from 9.7 months up till 5.5 months, given equal weights to OS and HRQoL. JM analyses resulted in a theoretical loss of treatment effect in OS of 2–6% when adjusting for HRQoL. CONCLUSION Both methods showed that adjusting for the impact of treatment on a relevant HRQoL parameter reduced the survival benefit in the experimental treatment arm compared to standard treatment arm. Applying these methods may facilitate communicating the impact of treatment to patients in clinical practice.

To compare the quality-adjusted survival of nivolumab plus ipilimumab combination and nivolumab alone versus ipilimumab alone among treatment-naive patients with advanced melanoma based on a minimum 36-month follow-up from the CheckMate 067 trial. Overall survival was partitioned into time without symptoms of progression or toxicity (TWiST), time with treatment-related grade ≥ 3 adverse events after randomization but before progression (TOX), and time from progression until end of follow-up or death (REL). Mean quality-adjusted TWiST (Q-TWiST) was calculated by multiplying the mean time spent in each health state by a utility of 1.0 for TWiST and 0.5 for TOX and REL. Sensitivity analyses included varying utilities of TOX and REL; Q-TWiST gains at different follow-up times were calculated using EQ-5D-3L utilities from the trial. Relative Q-TWiST gain of ≥ 10% was considered clinically important. Compared with ipilimumab-treated patients, those who received nivolumab + ipilimumab combination had significantly longer TWiST and TOX but shorter REL; nivolumab-treated patients had significantly longer TWiST, shorter REL, and shorter but statistically nonsignificant TOX. Mean Q-TWiST was highest for nivolumab + ipilimumab (23.5 months; 95% CI 21.9–25.2), followed by nivolumab (21.8 months; 95% CI 20.2–23.4) and ipilimumab (15.3 months; 95% CI 13.9–16.6). Relative Q-TWiST gains were favorable and clinically important for nivolumab + ipilimumab combination (+ 36.81%) and nivolumab alone (+ 29.18%) versus ipilimumab alone. Relative gains increased with follow-up from 3 to 40 months for all comparisons. These gains remained consistent in magnitude and direction in the different sensitivity analyses. Nivolumab + ipilimumab combination and nivolumab alone resulted in a statistically significant and clinically important improvement in quality-adjusted survival compared with ipilimumab alone.

Breast cancer treatment recommendations will often require an appraisal of likely benefits in relation to likely side effects on survival and quality of life end points, and possibly also an evaluation of the size of the anticipated net clinical benefit against financial costs. Quality-adjusted survival (QAS) analysis methods provide a formal approach for deriving an estimate of net clinical benefit to facilitate this appraisal process. QAS analysis methods have been applied in trials with breast cancer patients of adjuvant therapies as well as treatments for advanced/metastatic disease. QAS analyses based solely on trial data may fail to capture plausible longer-term benefits; thus, methods to explore the possible outcomes of treatment beyond the limits of trial data have been developed. These modeling approaches can help researchers gain insights and identify future research priorities, but do not replace the need for long-term evidence from randomized trials.

Purpose: To incorporate quality-of-life considerations in assessing high dose therapy (HDT) for patients with Multiple Myeloma (MM). Patients and methods: A quality-adjusted survival analysis, using the quality-adjusted time without symptoms or toxicity (Q-TWiST) method, was applied to two randomized clinical trials conducted in patients with MM which compared randomized assignment to HDT vs. conventional chemotherapy (CCT) alone (MAG91) or followed by HDT (MAG90). Treatment benefit in terms of mean Q-TWiST was assessed through threshold utility analyses, i.e., sensitivity analyses of the choice of the utility coefficients over all possible values of utility weights. Results: In both trials, results slightly favored the first-line HDT group over the first-line CCT group, with an average gain in TWiST of about 5.5 months over the 58 month-median follow-up period (27.8 vs. 22.3 months, respectively) in the MAG90 trial and 5.8 months over the 56 month-median follow-up period (19.1 vs. 13.3 months, respectively) in the MAG91 trial. The utility threshold analyses revealed that the first-line HDT group had a statistically increased mean quality-of-life adjusted time compared to the other group for a broad range of utility coefficient values. Conclusion: The development of such understandable and intuitive measures of expressing the relative benefit of complex treatment strategies is expected to be used in clinical decision making in the near future.

2032 Background: Routine clinical use of bevacizumab (Bev) in recurrent glioblastoma (rGBM) is controversial, as no large RCT has shown a survival advantage. We describe treatment algorithms, survival (OS), quality-adjusted survival (QAS) as well as costs of patients (pts) with GBM treated at an academic hospital in Switzerland, where Bev is registered and reimbursed for rGBM. Methods: Pts’ and treatment characteristics from diagnosis until death (including neurological symptoms and toxicities) of all pts over a 5-year period were retrospectively retrieved from our GBM database. For each treatment period (1st-line, recurrence and “best supportive care”) time to next treatment (TNT), OS and QAS were calculated and modelled for prognostic factors (Cox regression). QAS was evaluated as previously described (Murray et al). In- and out-patient costs were calculated from time of diagnosis until death in respect of Bev treatment (+Bev vs. –Bev). Results: 82 newly diagnosed GBM pts with a median age of 66 years (range 39-85), median KPS of 90% (range 50-100%), who were treated with 1 (n = 75), 2 (n = 36) or 3 (n = 14) lines of therapy, respectively, lived for a median OS of 11.9 (SD 9.7) months (mos). QAS was 5.3 (SD 6.9) mos i.e. 44% of the numerical survival time. 40% of patients were treated with Bev at 1st or 2nd recurrence. Pts, who were selected for BEV treatment, had a longer time from diagnosis to 2ndline treatment (median 5.2 (SD 6.3) mos) as compared to pts in the -Bev group (median 2.1 (SD 1.8) mos). Younger age and Bev treatment were associated with longer OS and QAS. QAS to OS ratio was 52% (9.4 out of 18.1 mos) for BEV treated patients and 34% (2.8 out of 8.2 mos) for the –Bev group, respectively. Bev treatment increased the overall treatment costs by 1.7x. The population adjusted incremental cost-effectiveness ratio (ICER) was CHF 75,669 per life year gained. Conclusions: QAS in patients with GBM is short (one third to half of OS). In our cohort, pts selected for Bev treatment had longer OS and longer QAS - at costs below the accepted threshold of 100,000 CHF per life year gained. Whether this gain of lifetime is a direct result of Bev treatment or a consequence of a selection bias needs to be addressed prospectively.

In clinical studies, we often consider not only patients' survival time, but also their quality of life. Quality adjusted life years (QALY) is an integrated measure of medical outcome that combines a patient's quantity and quality of life. Estimation of mean QALY for a group of patients is complicated by the fact that some patients are censored. The conventional approach is to obtain the Kaplan-Meier estimate of the survival function associated with individual QALYs and then use the area under the Kaplan-Meier curve as an estimate of mean QALY. Glasziou, Simes and Gelber showed that this method is biased because censoring at the nominal time scale is informative for predicting unobserved QALYs. In this paper, we propose a methodology for consistent estimation of mean QALY. Simulation studies are conducted to investigate the relative performance of the new method and the conventional method.

One way of examining trade-offs between quantity and quality of life (QOL) is to combine them into a single measure such as quality-adjusted life year (QALY). If censoring occurs, then estimation presents some difficulties. One approach, known as Q-TWiST, is to define a series of health states, use a 'partitioned' survival analysis to calculate the average time in each state, and then weight each state according to its quality of life to calculate QALYs. Such health-state models, however, are unhelpful when the transitions between health states are unclear or if they do not adequately reflect variations in quality of life. We therefore examine an alternative analysis to be used when repeated measures of quality of life are available from individual patients in a clinical trial. The method proceeds by separating quality of life and survival, that is, dQALY/dt = S(t)Q(t), where S(t) is the survival curve, estimated from the standard Kaplan-Meier method, and Q(t) is the quality of life function, derived from individual repeated measures of quality of life. We derive single health-state (QALY) and multiple health-state (Q-TWiST) models and illustrate the approach by comparing different durations of adjuvant chemotherapy for breast cancer.

Clinical trials in radiotherapy sometimes compare changes in radiation dose distribution using different radiation techniques. The use of quality adjusted survival can, in special circumstances, reduce the requirement of large patient numbers in order to show a significant difference in overall outcome. The provisos are that marginal improvements in survival or tumour control endpoints and a reduction in toxicity scores are present. The converse findings would also be amenable to this approach. Random sampling methods are used to construct a patient population where the first set of conditions is met. Further work is necessary to refine the absolute indications for this technique.

To perform a cost-utility analysis utilizing a cooperative group protocol and constrasting the results with the published quality adjusted survival.A cost-utility analysis was performed on Radiation Therapy Oncology Group (RTOG) protocol 83-02. The quality-adjusted survival has been published previously. Pretreatment tests and chemotherapy costs are not considered, as these were similar across all treatment arms. Payor costs are calculated from Federal Register data for Medicare Region IV. Global charges are used to calculate the professional and technical charges. Costs are measured in relative value units (RVUs) and are tabulated assuming equal treatment complexity for all treatment arms.The number of RVUs calculated for each arm were 48 Gy--166.65; 54.4 Gy--182.17; 64.8 Gy--232.53; 72.0 Gy--272.19; 76.8 Gy--287.11; and 81.6 Gy--302.63. The RVU/QALY for the50-year-old group were 48 Gy--119.03; 54.4 Gy--100.65; 64.8 Gy--104.78; 72.0 Gy--90.73; 76.8 Gy--193.99; and 81.6 Gy--165.37. The RVU/QALY for the50-year-old group were 48 Gy--198.39; 54.4 Gy--276.85; 64.8 Gy--426.57; 72.0 Gy--423.71; 76.8 Gy--703.70; and 81.6 Gy--519.10. Sensitivity analysis of one treatment plan, simulation, and set of blocks for the 48 Gy and 54.4 Gy arms decreased the RVU/QALY to 105.34 and 90.05, respectively.Our analyses shows the experimental arm with the lowest RVU/QALY is also the arm with the longest quality-adjusted survival. This arm had the most efficient use of resources in this protocol. Prospective collection of all pertinent cost data is required for comparison of one treatment against another. All cooperative group protocols should have prospective quality of life and economic endpoints to allow for comparison of the most cost efficient treatment method.

SUMMARY Quality adjusted survival analysis is a new approach to therapy evaluation in clinical trials. It has received much attention recently because of its ability to take patients' quality of life into consideration. In this paper, we present a method that enables us to calculate the survival distribution of quality adjusted lifetime. Using martingale theory for counting processes, we can show that our estimator is asymptotically consistent, normally distributed, and its asymptotic variance estimate can be obtained analytically. Simulation experiments are conducted to compare our estimator with the true underlying distribution for two cases that are of practical importance.

Background In survival analysis, those who are alive are statistically coded as 1.0 while those who are dead are coded as 0.0. Since everyone who remains alive is given the same score, a person confined to bed with an irreversible coma is alive and is counted the same as someone who is active and asymtomatic. The Quality of Well-being (QWB) scale defines levels of wellness on the continuum between death and optimum function and integrates morbidity and mortality into the same number. This paper demonstrates the effect of including mortality in QWB estimates for male adults with HIV infection. Method This study involves follow-up of a cohort of 386 male adults participating in the San Diego HIV Neurobehavioral Research Center (HNRC). Patients were evaluated using the QWB at enrollment and at six month intervals. All patients were classified into three stages of HIV disease according to the Center for Disease Control (CDC) classification: CDC IV (symptomatic HIV disease), CDC II or III (asymptomati...

We develop a method for incorporating covariates as regressors in a quality adjusted survival analysis (Q-TWiST) using Cox's proportional hazards model. The standard Q-TWiST method assumes that patients progress through a series of health states which differ in quality of life. The Kaplan-Meier product limit method is used to estimate the mean duration of each state by estimating the survival curves for the health state transition times. These estimates provide the basis for quality adjusted survival analysis. In this paper, the survival curves are modelled using Cox's proportional hazards regression. Quality adjusted survival is estimated given sets of covariate values, allowing one to profile patients. The results are useful for investigating how prognostic factors affect treatment benefits in terms of quality of life. We give a brief review of the standard Q-TWiST method and illustrate the extended methodology with an example from the International Breast Cancer Study Group Trial V comparing short duration versus long duration chemotherapy in the treatment of node-positive breast cancer.

The benefit of high-dose chemotherapy (HDC) has not been clearly demonstrated. It may offer disease-free survival improvement at the expense of major toxicity and increasing cost. We evaluated the trade-offs between toxicity, relapse, and costs using a quality-adjusted time without symptoms or toxicity (Q-TWiST) analysis.The analysis was conducted in the context of a randomized trial (PEGASE 01) evaluating the benefit of HDC for 314 patients with high-risk breast cancer. A Q-TWiST analysis was first performed to compare HDC with standard chemotherapy. We then used the results of this Q-TWiST analysis to inform a cost per quality-adjusted life-year (QALY) comparison between treatments.Q-TWiST durations were in favor of HDC, whatever the weighting coefficients used for the analysis. This benefit was significant when the weighting coefficient related to the time spent after relapse was low (0.38). For quite high values of this coefficient (0.78), HDC offered no benefit. For intermediate values, the results depended on the weighting coefficient attributed to the toxicity period. The incremental cost per QALY ranged from 12,691euro/QALY to 26,439euro/QALY, according to the coefficients used to weight toxicity and relapse.The benefits of HDC outweigh the burdens of treatment for a wide range of utility coefficients. Economic impact is not a barrier to HDC diffusion in this situation. Nevertheless, no significant benefit was demonstrated for a certain range of utility values.

3543 Background: The pivotal phase III PRIME trial demonstrated that, compared with FOLFOX alone, panitumumab plus FOLFOX prolongs overall survival (OS) and progression-free survival (PFS) in patie...

537 Background: The pivotal phase III PRIME trial demonstrated that, compared with FOLFOX alone, panitumumab plus FOLFOX prolongs overall survival and progression-free survival in patients with RASWT mCRC without having a detrimental effect on overall quality of life. The objective of this analysis was to use the quality-adjusted time without symptoms of disease or toxicity of treatment (Q-TWiST) method to compare quality-adjusted survival between the treatment arms of PRIME. Methods: Patients with RASWT mCRC from each treatment arm of PRIME were included in this analysis, and the area under the survival curve (AUSC) was estimated using the nonparametric Kaplan-Meier method (48 months follow-up). The AUSC was partitioned into three health states: toxicity (TOX), time without symptoms of disease progression or toxicity (TWiST, i.e., PFS minus TOX), and relapse (REL, i.e., OS minus PFS). The durations of the health states were then adjusted using utility weights derived from patient-reported EuroQoL 5 measures. The null hypothesis of no difference between treatment arms was tested based on the normal approximation, with standard errors calculated by the bootstrap method. In the primary analysis, TOX consisted of grade 3 and 4 adverse events. A sensitivity analysis, with grade 2, 3, and 4 adverse events classified as TOX, was performed. Results: Of 1,183 patients with mCRC who were randomized, 512 patients had RAS WT tumors (panitumumab plus FOLFOX, n=259; FOLFOX alone, n=253) and were included in this analysis. Patients receiving panitumumab plus FOLFOX had a significantly longer Q-TWiST (20.5 months) than patients receiving FOLFOX alone (18.2 months) resulting in 2.3 ± 1.0 (SE) additional quality-adjusted months (p

We present methods for analyzing quality of life data through the use of weighted prevalence functions when data allow a classification of scores into discrete states of health. Individual longitudinal data is transformed from an event history model into a non-parametric setting to provide a simpler interpretation in terms of probabilities. We discuss similarities and differences with quality adjusted survival methodology and illustrate the methods with an example.

The quality of life of patients is an important component for the evaluation of therapies in clinical trials, especially in cancer. The Quality-adjusted Time Without Symptoms and Toxicity (Q-TWiST) method evaluates treatment trade-offs in clinical trials using an endpoint based on both quantity and quality of life. The methodology has been most useful for evaluating the trade-off between increased toxic side effects of treatment and improved clinical outcome. This paper provides an overview of the Q-TWiST method, brief descriptions of related statistical tools for regression analysis and meta-analysis, and examples illustrating the techniques using data from cancer clinical trials.

6630 Background: How can a clinician combine survival and toxicity data to interpet treatment benefit? Quality adjusted life years (QALYs) attempts to account for both the quality and quantity of l...

Purpose: To quantify the quality of life of malignant glioma patients treated on a randomized Phase I/II trial of twice-daily radiation therapy (RT) and carmustine, using a modified quality adjusted survival (QAS) model, and to compare the QAS among assigned treatment arms. Materials and Methods: The Radiation Therapy Oncology Group (RTOG) accrued 786 malignant glioma patients to a Phase I/II randomized dose escalation trial of twice-daily RT with carmustine from 1983 to 1989. Patients were randomized to one of four arms of hyperfractionated RT in 1.2 Gy twice daily (BID) fractions (64.8 Gy, 72.0 Gy, 76.8 Gy, or 81.6 Gy) or to either of two accelerated hyperfractionated RT arms in 1.6 Gy BID fractions (48.0 or 54.4 Gy). Although preliminary toxicity and survival data have been published, little information is available regarding the quality of these patients' lives during and following such therapy. QAS is a refinement of the methodology for assessing survival quality among breast cancer patients receiving adjuvant chemotherapy. The QAS method allows for inclusion of both improvement and decline in neurologic functional status. Patients were scored by the presence or absence of 15 neurologic signs and symptoms at on-study and at every follow-up. Within each category were gradations of severity, with the quality survival time (Q-TIME) adjusted according to any changes in these neurologic findings. The summation of all changes in signs and symptoms were weighted by 1/15th and incorporated into the QAS model as QAS=Q-TIME-TOX-RRX. TOX was the time spent with treatment-related toxicities, and RRX was the time spent in recovery from subsequent therapy. Results: Of 747 evaluable patients, the average QAS time was 18.5 months. The average QAS for the hyperfractionated arms of 64.8 Gy, 72.0 Gy, 76.8 Gy, and 81.6 Gy were 15.6, 20.8, 10.0, and 13.7 months, respectively. For the accelerated hyperfractionated RT arms of 48.0 and 54.4 Gy, the average QAS times were 13.1 and 13.4 months. The QAS time of the 72.0 Gy arm was significantly longer than that of all other groups, except the 64.8 Gy arm. As expected, the QAS times were strongly discriminated by both age and Karnofsky Performance Scores (KPS) (p

We present a technique, quality adjusted survival analysis, for the analysis of controlled trials where patients may experience several health states which differ in their quality of life. When the data are censored, a survival analysis of the quality adjusted life years achieved may involve informative censoring, and produce biased estimates. To overcome this, we partition the survival curve; the resulting areas, which represent the mean time in each state, are multiplied by utility weights to provide an unbiased estimate of (restricted) quality adjusted survival. If the appropriate weights are in doubt, the results are best presented as a threshold analysis over the utility weights, allowing individual recommendation to be read from a simple graph. The certainty of the conclusions can be presented as confidence bands on the threshold line. The techniques are illustrated with a re-analysis of a large three-arm trial of adjuvant chemoendocrine therapy for stage II breast cancer in postmenopausal women. This shows that if the value of time spent in toxicity is greater than the time spent in relapse, we can be 95 per cent confident that chemoendocrine therapy is the preferred option.

5049 Background: In a phase 3, randomized, 3-arm study of TEMSR or IFN or the combination of TEMSR + IFN in the first-line treatment of patients with advanced renal cell carcinoma (RCC) and poor-prognostic features, TEMSR improved overall survival (p=0.0069) and progression-free survival (p=0.0001) vs. IFN (Hudes et al. J Clin Oncol. 2006;24:LBA4). Quality-adjusted survival was a pre-defined endpoint and is reported. Methods: Quality-adjusted time without symptoms and toxicity (QTWiST) was estimated by partitioning overall survival into 3 distinct health states: time with serious toxicity, time with progression, and time without symptoms and toxicity (TWiST). Survival was value-weighted when patients completed quality of life questionnaires (EQ-5D) at weeks 12 and 32, when a grade 3 or 4 adverse event (AE) was reported, upon relapse or progression, or upon withdrawal from the study. Treatment group comparisons used restricted means analyses estimated from censored survival data. Restricted means were computed for duration of each health state by truncating data at median follow-up (17.9 months). Variance and covariance were estimated using bootstrap methods. Results: All 626 randomized patients in the study were included in computation of health state durations. EQ-5D questionnaires were obtained from 260/300 (87%) upon progression and 230/570 (40%) after a grade 3/4 AE. Patients receiving TEMSR alone had 38% greater TWiST than those receiving IFN alone (TEMSR=6.5 months vs. IFN=4.7 months, p=0.00048). There was no significant difference in TWiST between the combination arm and IFN alone (IFN+TEMSR=5.4 months vs. IFN=4.7 months, p=0.1288). Patients receiving TEMSR alone had 23% greater Q-TWiST than those receiving IFN alone (TEMSR=7.0 months vs. IFN=5.7 months; p=0.0015). There was no significant difference in Q-TWiST for the combination arm compared with IFN alone (IFN+TEMSR=6.1 months vs. IFN=5.7 months, p=0.3469). Conclusions: Patients with advanced RCC receiving TEMSR alone had significantly greater quality-adjusted survival than those receiving IFN alone. No significant financial relationships to disclose.

3625 Background: QoL is infrequently assessed in RCTs of adjuvant therapy for CRC. This analysis aims to assess QoL and QAS in patients receiving adjuvant 5-FU for stage II and III CRC according to treatment arm, age, gender and primary tumour location. Methods: We performed a multicentre RCT in which 801 patients were randomised to 6 months of bolus 5-FU/leucovorin (n=404) or 12 weeks of PVI 5-FU (n=397). QoL was assessed by EORTC QoL core 30 questionnaire before randomisation, during treatment and follow-up to a maximum of 5 years. QAS analysis was performed by integrating the product of the grouped QoL and the survival function. P-values

'Department of Biostatistics, Harvard School of Public Health, and Division of Biostatistics, Dana-Farber Cancer Institute, 44 Binney Street, Boston, Massachusetts 02115, U.S.A. 2Department of Biostatistics, Harvard School of Public Health, Division of Biostatistics, Dana-Farber Cancer Institute, and Department of Pediatrics, Harvard Medical School, Boston, Massachusetts 02115, U.S.A. 3Centocor, Inc., Malvern, Pennsylvania 19355, U.S.A.