Application of a decision analytic framework for adoption of clinical trial results: are the data regarding TARGIT-A IORT ready for prime time?

L. J. Esserman 0 1 2 3 M. D. Alvarado 0 1 2 3 R. J. Howe 0 1 2 3 A. J. Mohan 0 1 2 3 B. Harrison 0 1 2 3 C. Park 0 1 2 3 C. O'Donoghue 0 1 2 3 E. M. Ozanne 0 1 2 3 0 C. O'Donoghue University of Illinois , Champaign, IL, USA 1 A. J. Mohan Brown University , Providence, RI, USA 2 L. J. Esserman (&) M. D. Alvarado R. J. Howe B. Harrison C. Park University of California , 1600 Divisadero, 2nd Floor, Box1710, San Francisco, CA 94115, USA 3 E. M. Ozanne Dartmouth University , Hanover, NH, USA The results from randomized clinical trials are often adopted slowly. This practice potentially prevents many people from benefiting from more effective care. Provide a framework for analyzing clinical trial results to determine whether and when early adoption of novel interventions is appropriate. The framework includes the evaluation of three components: confidence in trial results, impact of early, and late adoption if trial results are reversed or sustained. The adverse impact of early adoption, and the opportunity cost of late adoption are determined using Markov modeling to simulate the impact of early and late adoption in terms of quality of life years and resources gained or lost. We applied the framework to the TARGIT-A randomized clinical trial comparing intraoperative radiation (IORT) to standard external beam radiation (EBRT) and considered these results in the context of trials comparing endocrine therapy with and without radiation therapy in postmenopausal women. Confidence in the TARGIT-A trial 4 year results is high because the peak hazard for local recurrence in the trial is between 2 and 3 years. This is consistent with most trials, and no second peak has been observed in similar patient populations, suggesting that the TARGIT-A trial results are stable. The interventions offer approximately equivalent life expectancy. If IORT local recurrences rate were as high as 10 % at 10 years (which is higher than expected), we would project only 0.002 fewer expected life years (less than 1 day) compared to EBRT if IORT is adopted early. However, there is a $1.7 billion opportunity cost of waiting an additional 5 years to adopt IORT in low risk, hormonereceptor-positive, postmenopausal women. EBRT costs an additional $1467 in indirect costs per patient. Applying an evaluative framework for the adoption of clinical trial results to the TARGIT-A IORT therapy trial results in the assessment that the trial results are stable, early adoption would lead to minimal adverse impact, and substantially less resource use. Both IORT and no radiation are reasonable strategies to adopt. - Clinical adoption should occur when high-level randomized data clearly show the efficacy of one treatment against another, without serious adverse effects. However, several factors other than evidence regarding these outcomes influence the adoption of new findings from clinical trials. For example, when less treatment is shown to have an equivalent outcome, change in practice is highly variable. It is typically easier to adopt or add new agents or treatments, but can be hard to abandon what is seen as standard treatment. In the United States, new business for physicians, hospitals, and manufacturers that result from a positive trial often helps drive adoption. However, when new approaches disrupt the routine of practice, or when interventions reduce the use of particular services, there may be less incentive for early adoption [1]. Reasons for slow adoption include financial disincentive, disruption of practice routines, resistance to change, fear of abandoning a standard treatment, and skepticism about trial results [1]. The question of how to approach patients with early stage, favorable risk breast cancer illustrates the complexity of forces influencing decision making with regard to the adoption of new approaches. Whole-breast external beam radiotherapy (EBRT) remains the standard of care following breast-conserving surgery. However, multi-dose partial breast radiation is increasingly offered as an alternative for eligible women. Emerging technologies have provided impetus for shifts in radiation approaches despite lack of randomized clinical trial data for these devices [2]. The frequency of brachytherapy use increased from *1 % in 2001 to 10 % in 2006, despite concerns about long-term efficacy [3]. Meanwhile, the finding from a randomized trial (CALGB 9343) that older women with hormone-receptor-positive breast cancer could be effectively treated with tamoxifen without radiation therapy have yet to be adopted into clinical practice [4]. The results demonstrated that with or without radiation, distant recurrence, breast cancer mortality, and mastectomy rates are the same and very low in the two arms [5]. Despite these results with more than 10 years of follow-up, and corroborating evidence from a similar Canadian trial in all postmenopausal women, [6] radiation is rarely omitted and post-lumpectomy radiation is considered a quality measure by the American College of Surgeons for older women [7]. Fear of omitting therapy and being less aggressive often makes both physicians and patients uncomfortable, even in the face of supporting evidence to the contrary. There is a cultural bias which dictates that more aggressive treatment for cancer is better, and this bias colludes with the fear of malpractice and financial rewards to encourage intervention. With this as prelude, it is not surprising that when the results of the international randomized trial comparing a single intraoperative radiation (IORT) using low-energy photons delivered by the TARGIT device were published in 2010, there was a great deal of criticism and arguments that it was too early to adopt the findings [8]. Indeed, early adoption of technology that turns out to be inferior to the status quo can be harmful, but late adoption of technology that turns out to be equivalent or superior to the status quo can be a missed opportunity that also harms individuals and society. Clearly, a rational approach is needed to put early trial results in perspective and facilitate adoption decisions. We propose a decision framework for the adoption of clinical trial results that includes three components. First, we ex amine the level of confidence we have in the trial results, second, we predict the impact of early adoption of trial results, and third, we predict the impact of late adoption of those trial results. Predicted impact is measured through life expectancy, quality-adjusted life years (QALYs), and cost. We apply the proposed framework to the example of the TARGIT-A intraoperative radiotherapy trial to illustrate this approach in the hopes of providing a better platform for decision making for the medical community. The adoption framework (shown in Fig. 1) relies on predictive modeling, which can aid policy decisions by projecting possible outcomes. Within this framework, sensitivity analyses are used to determine the impact of initial assu (...truncated)