On this blog, we have discussed the many possible causes of platelet activation. One factor often under scrutiny is the impact platelet age has on activation status. At LightIntegra we have collected a large database of platelet activation status from platelet units across the United States. We decided to use this data set to reinvestigate if a link exists between activation rate and platelet age.
The Impact of Bag Age – Full Data Set
To minimize possible confounding factors, we looked only at platelets stored in 100% plasma. Additionally, we only included platelets from the suppliers for which we know the collection method. Below you will see the activation rate broken down by day for all 2745 samples in the database.
At first glance, a strange phenomenon appears the rate of activation spikes on Day 3 and then drops down again on Day 5. Furthermore, we found the differences between Day 3 and the other three days to all be statistically significant, or nearly statistically significant, as shown in the table below. However, an explanation for why Day 3 would be the most activated eluded us for quite some time.
|Comparison with Day 3|
Investigating Platelet Collection Method as a Confounder
The phenomenon of platelet units becoming more activated in the middle of their shelf life and then reversing to become non-activated towards the end, is unlikely. Thus, we concluded the differences were likely caused by something other than storage lesion and further investigation was required. In a previous blog, we discovered that platelets collected with Amicus apheresis machines are more likely to be activated then platelets collected on Trima Accel. To see if this might be playing a role here we looked at the relative proportion of platelets produced by the two collection methods for each day in the data set.
The Day 3 data segment is dominated by platelets collected with Amicus, whereas the other three days are dominated by platelets collected with Trima Accel. Additionally, we found a very good correlation (R = 0.98) between the percentage of the platelets in our data set that comes from Amicus vs the average Activation Rate for each day. These findings indicate that the reason Day 3 appears to be a spike in platelet activation is a result of the platelet units tested on Day 3 were primarily collected using a method that causes higher activation rates.
Activation Rate by Day and Collection Method
Taking this information we looked at the activation rates over time of the two collection methods separately, and a more accurate picture emerged. It seems that the age of a Platelet product, at least up to day 5, does not significantly impact its activation status, compared to its own baseline.
For the Trima Platelets, the average activation rate is very consistent across the platelet age, fluctuating just 1%. For Amicus, we see substantially more variation between the days, however, the sample size for Day 5 is very small and may not accurately represent the average activation rate of Day 5 platelets from an Amicus machine. As the table below shows, our data set does not show a statistically significant difference for activation rates between Day 3 and any of the other days when separated out by apheresis collection machine.
|Comparison with Day 3|
|Difference||P value||Difference||P value|
This analysis has some major limitations. For one, platelets tend to be tested on ThomboLUX when they are first received at a hospital, as such, any activating stimuli that occur while stored in a hospital blood bank will be missed in this analysis. Additionally, we were not able to track the storage life of individual Platelet products in a longitudinal study, but rather we are looking cross-sectionally at 2745 different Platelet products.
Even with the limitations mentioned this analysis has led to two important insights. For one, it is unlikely that platelet age, within the range of Day 2 to Day 5, plays a significant role in platelet activation status. If that were the case we would expect to see decidedly higher or lower activation rates in Day 5 compared with Day 2 with this large of a sample.
The other insight from this investigation comes from Occam’s Razor; when presented with competing hypothetical answers to a problem, select the answer that makes the fewest assumptions. The initial finding of a statistically significant difference in Day 3 had our team testing far-flung hypotheses to creatively explain what was observed. Ultimately, it was the simple idea of applying the insights from our previous post on the difference between Amicus and Trima platelets that explained the phenomenon.