A recent paper was published by Dr. Elliott Vichinsky this month in the New England Journal of Medicine about the outcomes of a Phase 3 RCT of Voxelotor in sickle cell disease. This post is my way of interrogating the paper and generating some possible follow-up questions after a deep-reading of the study in question. I hope this provides a little bit of help for those who are analyzing the trial.

  1. What is the difference between intention-to-treat analysis and per-protocol analysis, and is it relevant/important to the results generated in this study? ITT analysis is generally more inclusive compared to PP analysis (as it does not exclude patients who have strictly violated protocol). In the case of this study, the differences between ITT and PP appear negligible, and the results suggest that the mean delta Hgb in each treatment arm are comparable regardless of analysis type.

  2. The prevailing theory that hydroxyurea improves sickle cell disease clinical outcomes through induction of HbF is true, but probably incomplete. It should be noted that hydroxyurea also causes WBC suppression, and this neutropenia may in fact also be correlated with a decrease in pain crises in patients with SCD. Therefore, it may be worthwhile to stratify the patients based on baseline WBC count as well, which this study did not do. As it stands, the study only stratifies patients based on whether or not they take hydroxyurea as a binary variable, which may not fully capture the relevant independent variables.

  3. 9% of the placebo group had a hemoglobin response of >1.0 g/dL from baseline at week 24. Is this expected? How might these individuals have achieved a hemoglobin response, if these patients had not initiated hydroxyurea after randomization and before week 24? Hgb has some expected variability, but in a population which is anemic to begin with, a 1.0 increase in Hgb with no additional treatment in almost a tenth of the population is pretty unexpected. One possibility is that hydroxyurea dosage was adjusted during the course of the study, though it is unclear if this is the case.

  4. Although several measured blood parameters are clearly improved in a dose-dependent manner (i.e. Hgb, LDH, bilirubin, retic %), this study has not shown any significant changes to clinical outcomes of interest (VOE/pain crises, perceived QOL from improved anemia, etc.).

  5. It is not clear from this paper what the precise differences were between ITT and PP analyses. Did any of the patients in any group receive blood transfusions during the course of 24 weeks? How many people required or received hydroxyurea dose adjustments during the study?

  6. Voxelotor works by increasing hemoglobin-oxygen affinity. Theoretically, this could cause impairments in oxygen delivery, although it is encouraging that adverse outcomes are not more common in the maximal dosage group compared to placebo, and the erythropoietin levels were not increased.