The TOPPS trial (Trial of Prophylactic vs. No-Prophylactic Platelet Transfusions in Patients with Haematological Malignancies) provides a resource to explore to what degree different patient factors or clinical characteristics are important prognostic factors for bleeding. A previous analysis of the TOPPS data has looked at risk factors for bleeding on data that was collected at all study centres, this included: female sex; hematologic diagnosis; treatment for their hematologic diagnosis; and TOPPS treatment arm (prophylaxis/no-prophylaxis).

This study aims to look at additional factors that were not assessed in previous analyses of TOPPS data.


We collected additional data on potential risk factors for bleeding at two study centres (Oxford and Glasgow) on 219 patients with a total of 3454 observations, as a sub-study of the TOPPS trial (ISRCTN08758735). Bleeding information was reported for over 95% of study days. It was necessary to remove all study day 1 observations (as previous day lab values were not know), therefore 3084 observations were included in the recurrent event analysis.

We collected additional data on: Hb; WCC; albumin; bilirubin; creatinine; use of antibiotics; use of antifungals; admission to ITU; use of oxygen; CRP; temperature. This additional data was only available while patients were inpatients, however most bleeds that occurred during the TOPPS trial occurred while the patient was in hospital (94%). Only the first day of consecutive bruising or petechiae was counted as a bleed, unless bleeding on the subsequent day was worse.

Bleeding assessments were performed daily and assessed bleeding over the preceding 24 hours. The blood results performed on the day prior to the bleeding assessment were in the same 24 hour period as any bleeding occurred. Analyses were performed that took missing data into account or only included the full data set. Variables considered in the model were: female sex, hematologic diagnosis; treatment for their hematologic diagnosis; TOPPS treatment arm; age; platelet count; relapsed disease; previous HSCT; Hb; WCC; albumin; bilirubin; creatinine; use of antibiotics; use of antifungals; admission to ITU; use of oxygen; CRP; temperature.


Baseline characteristics were similar to all patients within the TOPPS study. The only differences noted between the Oxford and Glasgow groups and the overall study were: higher proportion of patients receiving an allogeneic HSCT; and a higher proportion of patients with a co-existing disorder or organ failure (13% vs. 7%). The median number of days with a WHO grade 2-4 bleed was 0 days (IQR 0 to 1).

In the univariate analysis: TOPPS treatment arm (P = 0.02); female sex (P = 0.002); previous HSCT (P = 0.03); previous day C-reactive protein (CRP) (P < 0.0001); previous day Hb (p < 0.0001); previous day WCC (P < 0.0001); previous day creatinine (P < 0.0001); treatment with therapeutic antibiotics (P < 0.0001); ITU/HDU admission (P = 0.04); oxygen treatment (P < 0.0001); high temperature (P = 0.02); and previous day's platelet count were associated with bleeding (P < 0.0001).

In the multivariate recurrent event analyses we investigated factors associated with the hazard of a grade 2 to 4 bleed on any one day. Female sex (hazard ratio (HR) 1.7 ; 95% CI 1.3 to 2.2; P < 0.0001); previous day platelet count (0 to 20 x 109/L versus > 50 x 109/L: HR 3.3; 95% CI 1.4 to 7.9; P = 0.0002), previous day CRP (100 to 150 versus 0 to 9 mg/L: HR 1.7; 95% CI 1.0 to 3.0; P = 0.03); previous day creatinine (121 to 180 versus 0 to 90 µmol/L: HR 1.9; 95% CI 1.3 to 2.8: P = 0.02); and previous day WCC (0 to 0.4 x 109/L versus > 3.4 x 109/L: HR 2.6; 95% CI 1.4 to 5.0; P = 0.005) were all found to be significantly associated with bleeding in both analyses. TOPPS treatment arm was found to be significant in the full data set analysis (no-prophylaxis versus prophylaxis: HR 1.3; 95% CI 1.0 to 1.7; P = 0.04) but not in the analysis that included missing data.


This new analysis of the TOPPS data set identified additional independent risk factors for bleeding that have not been previously identified, these were CRP and low white cell count. It also confirmed risk factors identified in previous analyses on the TOPPS data as well as other clinical trial data sets. CRP was the only significant sepsis-related risk factor. The next step is to use data from this and other analyses to develop a simple model that can stratify patients into different bleeding risk categories.




No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.