We thank Liu and colleagues for their comments; however, it is important to clarify a number of points.
This was not a comparative study of stem cells versus differentiated cells.1 Bone marrow (BM) contains a wide range of cell types with potentially diverse actions when delivered during liver injury. Positive effects have been detected following the application of stem/progenitor cell containing BM fractions of varying purity and potency.2, 3 In our experiments, unfractionated BM delivery caused a net increase in murine liver fibrosis.1 This phenotype is likely to be the result of several (interacting and opposing) BM cell populations. BM stem cells, including hematopoietic and mesenchymal stem cells, constitute a numerically small minority of BM cells. Importantly, injected mesenchymal stem cells can give rise to profibrogenic liver cell populations,4, 5 highlighting the potential risks of uncharacterized cell populations.
The advantages of a single differentiated cell type in terms of predictability and mechanistic clarity do not preclude the potential therapeutic benefits of mixed cell fractions. However, for the field to progress, we need to understand the precise activities of individual donor cell lineages before we can make sense of the potentially diverse effects of mixed BM fractions.
BM stem cells or their derivatives may have benefit in chronic liver disease. However, to date, clinical studies have been small or uncontrolled, hence larger, randomized controlled trials are now required to determine whether there are positive or negative clinical effects. Indeed, adverse events have already been detected following autologous BM cell infusion.6 To this effect, we are part of a multicenter randomized trial testing CD133+ hematopoietic stem cells in cirrhosis that has already opened in the United Kingdom (ISRCTN: 91288089).
Regarding the possibility of BM stem cells transdifferentiating in vivo into hepatocytes, this has repeatedly been shown not to occur to any meaningful degree, if at all. Therefore, paracrine mechanisms of action are more plausible.7, 8
The chronology of human cirrhosis is measured in years, so clinically useful disease-modifying treatments may require repeat administration. Furthermore, it is possible that cell-based treatments will ultimately be tailored to parameters such as disease etiology and stage of disease, necessitating a broad and well-understood therapeutic arsenal.