REPLYING TO H. H. H. Adams, S. A. Swanson, A. Hofman & M. A. Ikram Nature 537, http://dx.doi.org/10.1038/nature19086 (2016)

In the accompanying Comment1, Adams et al. argue that we have not excluded two alternative explanations for our finding of extensive amyloid-β (Aβ) deposition in relatively young individuals who received extracts of human pituitary glands: that this pathology may be a consequence of either the underlying diagnosis for which the treatment was given or of the treatment with growth hormone itself, irrespective of whether it was contaminated. As we made clear in our letter2, our study was observational rather than an epidemiological or experimental one. Although, by its nature, it cannot exclude these hypotheses (or other possible explanations for which there is no supportive evidence), we considered them unlikely. These patients received cadaveric pituitary-derived human growth hormone (c-hGH) for various reasons and the individuals in our cohort that developed Aβ deposition were treated for pathogenetically unrelated conditions, or because of short stature of no obvious cause, making a common mechanism unlikely. We also identified no publications that report a causal relationship between panhypopituitarism, short stature or craniopharyngioma and Alzheimer’s disease or increased Aβ deposition. We also consider the proposal that growth hormone itself (acting through the growth-hormone–IGF (insulin-like growth factor) axis) administered through adolescence is a possible trigger of Aβ deposition to be unlikely. IGF-1 production is stimulated by growth hormone, and several reports describe increased IGF levels to be associated with increased Aβ clearance5,5 and decreased risk of Alzheimer’s disease5, although the authors also cite a single report stating the opposite6.

Adams et al. refer to our hypothesis (that Aβ seeds in batches of c-hGH triggered Aβ amyloidosis in recipients) as being untested. Clearly this cannot be experimentally tested in humans but there is a substantial body of experimental data in vitro and in vivo demonstrating Aβ seeding, including, as we cited in our letter, that peripheral inoculation of laboratory mice with Alzheimer’s disease brain extracts leads to cerebral amyloid angiopathy7. Investigating the role of seeded protein aggregation (often referred to as ‘prion-like’ mechanisms) is one of the most active current areas of neurodegeneration research8. As we note, it will be important to examine archived batches of hGH for presence of Aβ seeds by animal inoculation studies and this work is planned.

To suggest that our comparison with patients with other forms of prion disease unrelated to hGH treatment was futile is inappropriate. Were we to have found that patients with sporadic CJD or other forms of prion disease had a similar frequency and severity of Aβ pathology at comparable ages, or an increased frequency compared with controls, this would have argued strongly against our hypothesis, and indeed the two alternative hypotheses suggested by Adams et al. By comparison, with a large series of controls we noted that the frequency of Aβ pathology in sporadic CJD was in keeping with chance coincidence9, as did other studies10. These analyses in our paper argue strongly against an effect of prion disease per se on Aβ pathology, as suggested by Adams et al. An important question that we posed in our letter was whether similar pathology suggestive of transmission of Aβ seeds is seen in iatrogenic CJD caused by other medical procedures (unrelated to c-hGH treatment or the underlying conditions for which it was given). The other most common cause of iatrogenic CJD is use of dura mater grafts. Importantly, following our study, a recent Austrian and Swiss series of dura mater graft recipients has now been reported, which also found frequent vascular and parenchymal Aβ pathology consistent with our hypothesis11,12.

However, we agree with Adams et al. that carefully controlled epidemiological studies are valuable and indeed we hoped that our study would stimulate such discussion and encourage precisely such work. With respect to comparing patients treated with hGH and synthetic hormone, this will be difficult as these necessarily represent different age cohorts.

The authors of this Reply represent the authors of the original paper.