I thought that routine erythropoietin (EPO) treatment of preterm infants had been thrown out of the window. It is well known that EPO levels are low in preterms after birth, which is part of the multifactorial pathophysiology of anaemia of prematurity. Previous studies, and systematic review have shown that you can reduce the volume of blood transfused, if you routinely supplement with EPO, but have no significant effect on other clinical outcomes, or on donor exposure. There is some indication, from the Cochrane review, that retinopathy, and severe retinopathy is increased by early use of routine EPO, when it is started in the first week of life. After the first week of life there is no apparent effect on RoP. Direct comparisons of early versus late EPO have been performed which also seem to show more RoP with early treatment and no substantial benefit.

This is one issue that the neonatal community have investigated rather well, I thought. The RCTs include about 5000 preterm babies, many of them are high quality, and we have focused on clinically important outcomes. Done and dusted.

Or maybe not.

Ohls RK, Christensen RD, Kamath-Rayne BD, Rosenberg A, Wiedmeier SE, Roohi M, Lacy CB, Lambert DK, Burnett JJ, Pruckler B et al: A Randomized, Masked, Placebo-Controlled Study of Darbepoetin Alfa in Preterm Infants. Pediatrics 2013, 132(1):e119-e127. Darbepoetin is a long acting EPO analogue. In this rather small trial, 102 infants of 500 to 1250 g birth weight were randomized into 3 different groups. Darbepoetin weekly, EPO 3 times a week or sham treatment. The infants receiving either EPO or Darbepoetin had fewer transfusions and were exposed to fewer donors (mean 0.7 or 0.8 versus 1.2, of course the means aren’t very meaningful, the median in the sham group was 1, interquartile range 0 to 2). There were no apparent differences in RoP, but I don’t quite understand the numbers, there appear to be more babies in the various subcategories of RoP than the total number in each group (example: 32 babies were treated with darbepoietin : 22 did not develop RoP, 10 with ≤ stage 2, and 2 > stage 2 who required intervention, which makes 34).  That is a minor point, much more importantly the study is grossly underpowered for RoP outcomes. The authors quite rightly point this out themselves, and note that further (and I would emphasize, much larger) studies are required. 

One could question what is the clinical importance of reducing donor exposure? Current rates of viral infection from transfusion are tiny, is avoiding transfusion altogether or reducing the total number of donors all that important?

More interesting are the possible neuroprotective effects of EPO (also apparently shared by Darbepoetin). The babies in this study were eligible up to 48 hours of age if they had not been transfused, and actually received the first dose at a mean of 51 hours. For brain protection that is probably a little too late.

The blood transfusion practices in this study were quite restrictive, but there were a lot of transfusion protocol violations (0.8 transfusions out of protocol per patient in the Darbepoetin group, compared to 1.2 total transfusions per patient, so 2/3 of the transfusions were outside of protocol). What should be the appropriate thresholds for transfusion requires further study. Maybe the transfusion threshold studies should come before we do further studies of darbepoetin, so we would know what the baseline should be.