CRYSTMAS study adds to concerns about renal safety and increased mortality in sepsis patients

Th e CRYSTMAS trial compared resuscitation fl uids 6% HES 130/0.4 and 0.9% NaCl in patients with severe sepsis to fulfi l a postmarketing study commitment issued by the US Food and Drug Administration (FDA). However, published trial outcomes were selective. When the more complete data set is assessed and compared against previous data, it rather suggests that use of HES 130/0.4 does not lead to clinically relevant volume savings, and suggests similar negative eff ects on kidney function as were seen with an older HES solution and an increase in 90-day mortality by 6.8%. Th e CRYSTMAS trial by Guidet and colleagues [1] lacked power to address renal safety [2]. However, observed outcomes add to existing safety concerns [3]. Th e number of subjects undergoing renal replacement therapy (RRT) doubled in the hydroxyethylstarch (HES 130/0.4) arm [4] (Table 1); mean duration of RRT was 9.1 days versus 4.3 days [4]; and Kaplan-Meier curves for time to RRT showed a trend against HES (P = 0.06; Figure 1) [4]. Serious adverse events and serious adverse events leading to death numbered 53 versus 44 and 38 versus 32, respectively (HES versus saline) [4]. Mortality rates were all higher in the HES arm, and 90-day mortality rates were remarkably similar compared to a previous sepsis trial with 10% HES 200/0.5 [3] (Figure 2). Th ese study data were not published in the journal article. Withholding data from original publications - socalled publication bias - is a major problem [5]. Th ere was also other reporting bias. Th e primary outcome required fl uid volumes - was only reported for a post hoc subgroup of patients who reached hemodynamic stabilisation (174 of all 196 patients), misleadingly called the ‘full analysis set’. Th e numbers of patients available for safety outcomes was inconsistent. Publication and report ing bias can lead to overestimation of the treatment eff ect and underestimation of safety risks [5]. Our reading of the extended CRYSTMAS study data therefore leads to somewhat changed key messages: use of HES 130/0.4 does not lead to clinically relevant volume savings; and negative eff ects on kidney function and mortality after HES 130/0.4 were similar to those seen after an older HES solution.

Th e CRYSTMAS trial compared resuscitation fl uids 6% HES 130/0.4 and 0.9% NaCl in patients with severe sepsis to fulfi l a postmarketing study commitment issued by the US Food and Drug Administration (FDA). However, published trial outcomes were selective. When the more complete data set is assessed and compared against previous data, it rather suggests that use of HES 130/0.4 does not lead to clinically relevant volume savings, and suggests similar negative eff ects on kidney function as were seen with an older HES solution and an increase in 90-day mortality by 6.8%.
Th e CRYSTMAS trial by Guidet and colleagues [1] lacked power to address renal safety [2]. However, observed outcomes add to existing safety concerns [3]. . Mortality rates were all higher in the HES arm, and 90-day mortality rates were remarkably similar compared to a previous sepsis trial with 10% HES 200/0.5 [3] (Figure 2). Th ese study data were not published in the journal article. Withholding data from original publications -socalled publication bias -is a major problem [5]. Th ere was also other reporting bias. Th e primary outcomerequired fl uid volumes -was only reported for a post hoc subgroup of patients who reached hemodynamic stabilisation (174 of all 196 patients), misleadingly called the 'full analysis set' . Th e numbers of patients available for safety outcomes was inconsistent. Publication and report ing bias can lead to overestimation of the treatment eff ect and underestimation of safety risks [5].
Our reading of the extended CRYSTMAS study data therefore leads to somewhat changed key messages: use of HES 130/0.4 does not lead to clinically relevant volume savings; and negative eff ects on kidney function and mortality after HES 130/0.4 were similar to those seen after an older HES solution.  ; total number of patients with available data is not given. b Data according to [1,6]. c Data according to [1]. n = number of patients with condition; N = total number of patients with available data. ARF, acute renal failure; RRT, renal replacement therapy.

Authors' response
Bertrand Guidet and Hugo van Aken  We appreciate the opportunity to clarify the inappro priate statements by Hartog and Reinhart. Th e manu script was based on the results of the Clinical Study Report, which refl ected defi ned statistical analysis plan (SAP) analyses prospect ively agreed with the FDA, and was accepted for publication on 20 March 2012. Th us, the primary endpoint still holds true and is not a post hoc analysis. Concomitantly and independently, the FDA requested additional, unplanned post hoc analyses by the sponsor Fresenius Kabi, which were fi nally agreed upon on 7 May 2012. At this stage, the manuscript had already been accepted for publication.
Th erefore, the Kaplan-Meier for time to RRT was newly created and confi rms that the study was too small to investigate this parameter. Notably, internationally accepted scores of kidney injury are all presented in the paper, and the use of RRT is refl ected by the Acute Kidney Injury Network (AKIN) criteria.