LAUSR

A strength of Evolving treatment patterns and outcomes in older AML: changing everything to change nothing? is documenting the suspected increasing use of non-intensive induction therapy in persons age ≥60 years from 1% 1999 to 2006 to 23% from 2007, when azacytidine and decitabine became widely available in Spain, to 2013 [1]. Use of intensive chemotherapy decreased from 58% to 32%. Supportive care and clinical trial percentages remained constant (27–28% and 0–2%, respectively). Despite the more common use of non-intensive induction, survival was unchanged, with a median of 4.3 months, 1-year survival of 29% and 5-year survival of 7%. This explains the subtitle changing everything to change nothing? As the authors acknowledge, the data derive from a cooperative group registry, rather than a population-based system and thereby include only hospitalized persons. The typescript might have benefitted from multivariable analyses comparing survival in the two intervals after accounting for differences in potentially confounding covariates such as age distribution or cytogenetics or, similarly, comparing less versus more intense induction. Many may see limitation of analyses to 1999–2013 as problematic. Change nothing seems inappropriate given recent regulatory approvals of venetoclax with low dose cytarabine (LDAC), azacytidine or decitabine, and glasdegib with LDAC, for older persons with newly diagnosed AML. Randomized clinical trials involving these drugs report statistically significant improvements and clinically meaningful benefits [2]. Clearly, these drugs are advances. They lengthen survival, the outcome people often care most about [3–5]. They prolong event-free survival (EFS), perhaps a better measure of drug efficacy [6]. They increase response rates, although the contributions to survival or EFS of responses less than complete remission is largely unknown [7]. They increase rates of transfusion independence of indeterminant duration. Venetoclax + LDAC improve quality of life measures versus LDAC [4], but these measures were the same with venetoclax + azacytidine versus azacytidine [3]. This commentary discusses issues these new drugs present: (1) stereotyped criteria for statistical significance and power which ignore the specifics of AML, (2) conflation between statistical significance and medical benefit, and (3 uncritical adoption of the phrase meaningful clinical benefit. The venetoclax [3, 4] and glasdegib trials [5] were statistical successes with p values < 0.05 favoring the investigational drug. A p value < 0.05 is so entrenched in the clinical trial mindset that when the pre-specified comparison between LDAC ± venetoclax produced p= 0.11, further analyses, including one which resulted in p < 0.05 after an additional 6 months of follow-up, were done, despite the dubious statistical validity of this practice [8, 9]. A skeptic might wonder whether these analyses would have been done had the initial p value been < 0.05 or if another analysis is planned after a further 6 months of follow-up to see if the p value remains < 0.05. But my point is the almost invariant use of a false positive rate of at most 5% (p= 0.05) and of a false negative rate at most 20% (power= 80%) regardless of the disease in question. More concern should perhaps attach to a false positive in a trial of a new drug for hypertension, where there are already effective therapies, compared with a false positive in AML where we lack effective therapies for older persons, such that less is lost in replacing the standard of care with an ineffective treatment. Perhaps the p value of interest in hypertension should be <0.01 and in AML < 0.10. The same reasoning suggests acceptance of a lower false negative rate in AML. Granted statistical significance, the medical benefit of the new drugs is less clear. The absolute gain in median survival was largest in the trial adding venetoclax to * Elihu Estey [email protected]