Portal de Periódicos da CAPES

Pharmacoeconomic profile of paclitaxel as a first-line treatment for patients with advanced ovarian carcinoma

1997; Wiley; Volume: 79; Issue: 11 Linguagem: Inglês

10.1002/(sici)1097-0142(19970701)79

1097-0142

Andrea Messori, Michele Cecchi, Paola Becagli, Sabrina Trippoli,

Health Systems, Economic Evaluations, Quality of Life

In the recent study on the cost-effectiveness of paclitaxel as first-line chemotherapy for patients with advanced ovarian carcinoma,1 a novel, lifetime approach was utilized to analyze the survival data of the two patient groups (paclitaxel group and controls) and to extrapolate the two survival curves to infinity using the Gompertz function. As pointed out by the authors, one limitation of this approach is that no assessment of quality of life could be incorporated in this long term model and therefore no cost-utility evaluation could be performed. The Q-TWIST method2-5 is a technique that has frequently been employed for evaluating survival in terms of both duration and quality of life (particularly in oncology2, 4, 5). This method quantifies overall survival (OS) in a cohort of patients and then partitions the total number of patient years of the cohort into: 1) a first component of survival without symptoms and without toxicity (denoted as TWIST, i.e., Time spent WIthout Symptoms and Toxicity); 2) a second component of survival with treatment-induced toxicity (denoted as TOX, i.e., time spent with toxicity); and 3) a third component of survival after relapse (REL). The TWIST component is characterized by a normal level of quality of life (or normal "utility" with a score equal to 1), whereas both TOX and REL correspond to a compromised quality of life level (and are generally assigned a utility score of approximately 0.5). In the context of these utility coefficients, the value of 1 represents a "normal" healthy state's quality of life and the value of 0 represents a state as bad as death. The main advantage of the Q-TWIST model is that it appropriately accounts for quality of life. Its main disadvantage is that the model is not suitable for lifetime analysis because the Q-TWIST approach is not designed to manage cost data, does not include estimates of life expectancy, and does not handle discounted survival values (which instead are typically required by cost-utility and cost-effectiveness studies). Conversely, the lifetime Gompertz model utilized in the recent pharmacoeconomic study on paclitaxel1 is able to discount survival according to current standards, but does not handle quality of life. A combined application of the Q-TWIST and the Gompertz models (extrapolated Q-TWIST method)6 can borrow the advantages of the two techniques while excluding most of their disadvantages. The main feature of this combined approach is that the Gompertz extrapolation to infinity is introduced into the Q-TWIST approach. The following analysis, based on the data of cost,1 survival,7 and freedom from recurrence7 previously reported for paclitaxel, shows how this combined model can work. In the previous pharmacoeconomic study,1 the application of the Gompertz model allowed estimation of a lifetime survival of 274.4 patient years for the paclitaxel group and 230.8 patient years for the no paclitaxel group (discounted values normalized to 100 patients and extrapolated to infinity). In the combined approach described herein, two further Gompertz' least-squares fits are performed on the disease free survival (DFS) curves of the two patient groups; in this way, lifetime DFS is estimated as 206.5 patient years for the paclitaxel group and 141.4 patient years for the controls (discounted values normalized to 100 patients and extrapolated to infinity). Figure 1 shows the areas under the two curves of OS and DFS for the controls. This figure shows the time course of the Gompertz function for overall survival (solid line with circles) and disease free survival (dashed line with triangles) with extrapolation to infinity of both curves. These two curves refer to the control patients in the study by McGuire et al.,7 who received chemotherapy with cyclophosphamide plus cisplatin. In analyzing the data of the control group, a preliminary calculation is needed to estimate the value of TOX for these patients. The data of McGuire et al.7 suggest that approximately 60% of these patients experienced toxicity and that the average time spent with toxicity was approximately 6 months for each case. Hence, TOX (normalized to 100 patients) is 30 patient years every 100 patients. The Q-TWIST method partitions the OS (230.8 years every 100 subjects) as OS = TWIST + REL + TOX and estimates the value of REL as REL = OS - DFS (i.e., REL = 230.8 - 141.4 = 89.4 patient years). Because OS = 230.8, REL = 89.4, and TOX = 30, TWIST can be calculated as 230.8 - 89.4 - 30 = 111.4 years. Using a utility value of 1 for TWIST and 0.5 for REL and TOX, the lifetime calculation of quality-adjusted life years (QALYs) for the control group proceeds as follows: QALYs = 111.4 × 1 + 89.4 × 0.5 + 30 × 0.5 = 171.1. The same calculation for the paclitaxel group starts with the estimation of TOX (TOX = 40 years every 100 patients assuming that approximately 80% of treated patients experienced toxicity7 and that the average time spent with toxicity was 6 months for each case). The value of REL is given by REL = OS - DFS (i.e., 274.4 - 206.5 = 67.9 patient years). The value of TWIST is 274.4 - 67.9 - 40 = 166.5 patient years. Using the utility values mentioned earlier, the lifetime calculation of quality-adjusted survival for the paclitaxel group is: QALYs = 166.5 × 1 + 67.9 × 0.5 + 40 × 0.5 = 220.5. These data show that the incremental lifetime utility derived from the paclitaxel chemotherapy was 220.5 - 171.1 = 49.4 QALYs every 100 patients. This latter value has a methodologic advantage because it accounts for quality of life, refers to a lifetime calculation, and incorporates 5% annual discounting. Messori et al.1 estimated that, in these patients, the incremental cost of paclitaxel versus standard treatment was approximately $900,000 every 100 subjects. Using this cost value, the ratio of incremental cost and incremental utility ($900,000/49.4 QALYs) permits the conclusion that the first-line chemotherapy with plus cisplatin implies a cost of approximately $18,200 per QALY gained in comparison with the standard treatment with cisplatin plus cyclophosphamide. This example shows that this combined application of the Q-TWIST and the Gompertz models can be useful in performing a lifetime cost-utility analysis in cancer patients. It should be stressed that this combined approach maintains some limitations of the Gompertz method, and so can be applied only when measured survival (e.g., from time 0 to 48 months in Figure 1) covers an area under the curve much greater than the extrapolated "right tail" (i.e., the area from 48 months to infinity in Figure 1). Another disadvantage derives from the approximate nature of the utility coefficients that have traditionally been used in previous Q-TWIST studies2-5 and that have been transferred without modifications into the combined approach described herein. Andrea Messori Ph.D., Michele Cecchi M.D., Paola Becagli Pharm. D., Sabrina Trippoli Pharm. D.