We thank the authors for their response to our recent review. We read their hypothesis of the role of DNA in initiating a “xanthine oxidase driven free radical crisis” with interest, as it was not one of the hypotheses presented in Table 2 of our article outlining what might drive the inflammatory response in preeclampsia. However, there are problems with this idea.

There is some confusion in the model of events proposed. The authors argue that xanthine oxidase produced by the liver induces cell-free fetal DNA (cffDNA) resulting in oxidative stress that in turn causes acute atherosis and damage in the placenta. However, this does not explain why there is damage and “excessive embolization of trophoblast” releasing excess cffDNA in preeclampsia in the first place; it cannot be from oxidative damage to the placenta, as that would create a circular argument. Indeed, acute atherosis is only seen in 20-40% of cases of preeclampsia as well as 14% of normal pregnancies. It only occurs locally in the uterine arteries and not elsewhere in the vasculature, so seems unlikely to be caused by systemic oxidative stress.

The article describing oxidative stress as localized to the maternal side of the placenta has a number of weaknesses: firstly, it appears that only the basal plate is depicted, with a focus on extravillous trophoblast rather than villous syncytiotrophoblast (STB). Furthermore, a comparison is made between villous cytotrophoblast and decidual cells in vitro even though cytotrophoblasts have much higher levels of antioxidant defenses than STB. Thirdly, there is good evidence for oxidative stress in the STB of preeclamptic placentas, at least in early-onset cases, with increased release of exosomes, DNA, and other oxidative products. Thus, the assertion that oxidative stress “is more injurious to maternal tissue than the fetoplacental unit” needs much better supporting evidence.

Although both the liver and the placenta contain xanthine oxidase, the free radicals produced by this enzyme have a very short half-life and seem unlikely to be circulating mediators of stress. Some other mediators, perhaps released from the stressed liver, are more likely systemic inflammatory mediators. In line with this, the use of antioxidants and drugs such as allopurinol in the prevention of preeclampsia have often been disappointing, and future investigation of oxidative stress as a mechanism in preeclampsia must explain how these failures translate into clinical practice. Thus, further clarification of this model would certainly be welcome.

The induction of inflammation and dysregulation of hepatic purine metabolism by high levels of placental DNA are not mutually exclusive. As such, cffDNA could act both as a whip to the “horse” of oxidative stress in hepatocytes and as an activator of pattern recognition receptors (PRRs) in its own right. The relative reality and importance of these 2 potential actions is yet unclear. We agree that the evidence for a “major direct role” of cffDNA is not yet wholly convincing (more recent animal evidence makes such a role seem less rather than more likely even though preeclampsia only occurs in human beings ). There is indeed great difficulty in separating any potential role of cffDNA from that of oxidative free radicals and other potential inflammatory mediators such as soluble FMS-like tyrosine kinase (sFLT) and tumour necrosis factor alpha (TNF-α), in determining the source of these mediators (the placenta, the liver, the endothelium, or elsewhere) and in defining the order of events. However, we believe the existing level of evidence for a contribution from cffDNA presented in our review and others still warrants further investigation.