PATCH trial: explanatory analyses

TO THE EDITOR:

Patients using antiplatelet therapy before spontaneous intracerebral hemorrhage (ICH) have higher case fatality due to more ICH growth compared with patients not using antithrombotic drugs.^1-3 

Platelet transfusion to mitigate effects of antiplatelet therapy and reduce ICH growth⁴  was shown to increase risk of death or disability compared with standard care in the Platelet Transfusion in Cerebral Hemorrhage (PATCH) randomized controlled trial.⁵  We found a higher incidence of serious adverse events after transfusion, due to complications of ICH (brain edema, ICH growth, and intraventricular extension [IVE]) possibly causing poor outcome. We hypothesized that this could be due to the proinflammatory effects of platelets, possibly enhanced during storage.⁵  Therefore, we undertook further analyses of brain imaging and transfusions to investigate the mechanism by which platelet transfusion may be associated with poor outcome.

Detailed methods are described in supplemental Methods (available on the Blood Web site) and have been published previously.⁵  Additionally, we collected time of symptom onset or time the patient was last seen well from the medical records at participating sites if available. We calculated time from symptom onset or last seen well to baseline imaging in hours. Furthermore, shelf life in days from donation to transfusion and preparation type (buffy-coat derived vs apheresis) of platelets was recorded.

Brain imaging, either computed tomography or magnetic resonance (MR), was performed on admission and repeated after 24 ± 3 hours with the same modality. Presence of IVE was rated by blinded central reading. We measured volume using an automated planimetric method⁶ : total lesion involving ICH and perihematomal edema (PHE) and the area involving ICH alone were manually segmented on unenhanced baseline and follow-up imaging. We calculated ICH, PHE, and total lesion volume (TLV) in milliliters. For patients with MR imaging, fluid-attenuation inversion recovery and susceptibility weighted images were used for measurements. Relative PHE (rPHE) was calculated by dividing PHE by ICH volume.⁷  We calculated ICH, PHE, and TLV growth between baseline and follow-up imaging in milliliters. We defined clinically significant ICH growth as absolute growth >6 mL within 24 hours.^3,-10 

Volume at baseline was compared between platelet transfusion and control groups using descriptive statistics. We compared ICH growth >6 mL, ICH growth >6 mL or >33%, PHE growth, TLV growth, rPHE, and new IVE at 24 hours between the 2 treatment groups using χ² and Mann-Whitney U tests where appropriate. We used logistic regression to adjust risk of ICH growth >6 mL for ICH volume at baseline.³  Sensitivity analyses were performed to adjust for time of symptom onset to baseline imaging, to adjust for dual antiplatelet therapy, and to exclude patients with MR imaging, the latter to evaluate whether hematoma measurements with MR had offset the results.

Proportions of patients with poor outcome (modified Rankin scale score 4-6 inclusive) were compared between different platelet shelf lives and preparation type using the χ² test. When patients received multiple platelet units, the oldest was used for analysis. Linear and logistic regression was used to investigate the association between platelet shelf life and preparation type and occurrence of ICH growth or PHE expansion. Analyses were done in the as-randomized and as-treated populations to further investigate the possible effects of platelet transfusion on lesion volume.

The study population was previously described⁵  and is detailed in supplement 2 (Summary of population and CONSORT diagram). At baseline, TLV, ICH, and PHE volumes were significantly larger in patients who received platelet transfusion (Table 1). There was a nonsignificant excess of IVE in the control group. Clinically significant ICH growth (>6 mL) was more frequent in the transfusion group. Other imaging parameters did not reach statistical significance, but were all higher in the transfusion group except for rPHE (Table 2).

After adjusting for hematoma volume at baseline, the odds ratio (OR) for ICH growth >6 mL in patients receiving platelet transfusion compared with standard care was 2.36 (95% confidence interval [CI], 1.03-5.40) in the as-randomized population. Sensitivity analyses correcting for time from symptom onset to baseline imaging, use of dual antiplatelet therapy, and exclusion of patients with MR imaging all showed similar results (supplemental Table 1). Seventy-eight percent of patients with ICH growth >6 mL had poor outcome compared with 58% with ≤6 mL growth (OR, 2.45; 95% CI, 1.07-5.63). The proportion of patients with poor outcome was not different across shelf life or preparation type of platelets and there was no association with significant ICH or PHE growth (supplemental Figure 1; supplemental Table 2).

ICH volume and PHE volume at baseline were greater in patients who received platelet transfusion compared with patients receiving standard care. Imaging after 24 hours showed more frequent clinically significant ICH growth (>6 mL) in patients treated with platelet transfusion, which was adjusted for baseline hematoma volume. We did not find an association between shelf life or preparation type of transfused platelets and clinical outcome at 3 months or imaging parameters after 24 hours. The main analysis of PATCH was adjusted for baseline imbalances using the ICH score, which incorporates ICH volume and IVE,¹¹  and found an increase in risk of poor outcome (common adjusted OR, 2.05; 95% CI, 1.18-3.56).⁵  With these new results, it is most likely that ICH growth explains the worse clinical outcome after platelet transfusion found in PATCH.

These post hoc analyses have several limitations. Eighty-one percent of patients had follow-up imaging; although not complete, it was balanced for both treatment groups. Not all imaging was performed with the same modality, and intensity of hematoma can vary between MR and computed tomography imaging. Also, follow-up imaging was once performed at 24 hours. We cannot exclude that lesion volumes increased further after 24 hours.¹²  Furthermore, time from symptom onset, strongly associated with ICH growth, was not complete. Unfortunately, blood group (ABO) compatibility could not be investigated because most patients did not undergo blood group testing in the emergency setting. We can therefore not exclude effects of ABO incompatibility. However, whether blood group matching would be feasible in the emergency setting in which these patients are treated is questionable. Also, no serum samples had been collected, so no additional testing for effect of transfusion on hemostasis could be performed.

Other baseline imbalances were also present in PATCH. Patients receiving transfusion more often used dual antiplatelet therapy and had peripheral arterial disease. Dual antiplatelet therapy could instigate hematoma growth, although a sensitivity analysis adjusting for this showed similar results. We did not perform an analysis adjusting for peripheral arterial disease because this is not known to promote hematoma expansion. Finally, more recently, the importance of restarting antithrombotic drugs was highlighted.¹³  Unfortunately, data on restarting antithrombotics was not collected in PATCH.

It remains unclear whether the results of PATCH are explained by baseline imbalances alone or by a combination with a detrimental effect of platelet transfusion that remains to be identified. Because the findings of PATCH are at least partially explained by baseline imbalances, a definitive detrimental effect of platelets appears to be unlikely. However, it is also not likely that a positive effect on outcome in patients with ICH was overlooked. More recently, additional evidence has become available on possible adverse effects of platelet transfusion. A trial investigating transfusion threshold in neonates¹⁴  found that neonates receiving more platelet transfusions had a significantly higher rate of death and major bleeding. This, in combination with PATCH, should at least prompt caution in regular use of platelet transfusion to improve hemostasis. Future research is mainly needed in patients undergoing surgery for ICH because hemorrhage risk could be offset by operative risk in these patients.