Bayesian analyses demonstrate tissue blood volume is not decreased during acute sickle cell pain episodes: A preliminary study

Abstract

Background: Pain is the most common complication of Sickle Cell Disease (SCD). Tissue oximetry properties in SCD during steady state and acute pain are not well described.

Methods: This was a cross sectional study of tissue oximetry properties in individuals with SCD during steady state, acute pain and healthy controls without SCD. A novel tissue oximetry device was used to better account for tissue pigmentation interference. We hypothesized that during acute SCD pain, blood volume to painful areas would be at least 10% less than steady state. Bayesian analyses of the data (with flat piors) were planned a priori because of the small projected sample size.

Results: The sample included 14 individuals (4 during crisis, 5 steady state, and 5 controls). In individuals with SCD, blood volume to the lower back was higher during crisis (0.18% of tissue volume vs. 0.14% ). Bayesian analyses yielded a 3% probability that our hypothesis (that blood volume would decrease by 10% ) was correct.

Conclusions: During acute SCD pain, blood volume to painful areas is not decreased. Bayesian analyses were useful for interpretation of small sample data and may have utility in early phase trials for rare diseases.

Keywords: Sickle cell disease; pain crisis; tissue oximetry.

Fig. 1

Blood Volume, Saturation and Oxygen Delivery to the lower back during crisis, steady state and in healthy controls. The three graphs depict tissue oximetry characteristics in each of the three groups (error bars indicate 95% confidence intervals). The left depicts mean tissue blood volume expressed as a percent of overall tissue volume. The middle graph depicts oxygen saturation in the tissues (values range from zero to one). The left graph depicts oxygen delivery (expressed as the product of blood volume and oxygen saturation). These results suggested a trend towards increased blood volume and oxygen delivery during sickle cell crisis which was tested with Bayesian and frequentist statistical approaches.

Fig. 2

Bayesian Analyses of Blood Volume During Steady State and SCD Crisis. The tissue oximetry device measures the fraction of tissue occupied by blood (expressed as a percent). The top graph depicts our beliefs about blood volume during steady state in individuals with SCD. The green line indicates our beliefs before we collected data (the prior distribution), the blue line depicts the data we collected (likelihood function) and the pink line depicts our updated beliefs about blood volume based on the data collected (the posterior distribution). The middle graph depicts the same three things about blood volume during SCD crisis. The bottom graph depicts our beliefs about the difference between steady state and SCD crisis. This graph is created by subtracting the difference between random samples from the two posterior distributions above (Markov Chain Monte Carlo simulation). In the bottom graph, the majority of the distribution is above zero indicating that blood volume likely increases during crisis (82% probability). The a priori hypothesis was that blood volume would drop by at least 10% during crisis. Based on the data collected, the probability that our hypothesis was correct is 3% thus we rejected our hypothesis.