Reactive oxygen species in cancer, too much or too little?

It is widely accepted that increased levels of reactive oxygen species (ROS) contribute to carcinogenesis. However, this claim has not been confirmed by experiments. On the contrary, a growing number of studies clearly demonstrate that ROS are normal cellular signals and induce cell differentiation and apoptosis, the opposite processes to cancer, which is dedifferentiated. Thus, it is hypothesized here that decreased levels of ROS may lead to cancer development, which is supported by following observations: (1) the fast-growing tumor produces ROS at a rate only one-third of the rate found with the control liver mitochondria; (2) the reduction in tumor mitochondrial content indicates low level of ROS production; (3) the low levels of manganese superoxide dismutase in tumor mitochondria also indicate decreased production of ROS, because the enzyme activity is induced by ROS; (4) lipid peroxidation capacity was decreased in human colon carcinomas and Yoshida hepatomas; (5) low levels of lipid peroxidation de-inhibit glucose-6-phosphate dehydrogenase, whose activity is always increased in a variety of cancers without exception. Clarification of real role of ROS in cancer may shed light on the understanding of how impairment of mitochondria leads to malignant transformation of normal cells, and offer new types of strategies for cancer prevention and therapy.