Technical Comments

Response to Comment on "Chromosomal Instability and Tumors Promoted by DNA Hypomethylation" and "Induction of Tumors in Mice by Genomic Hypomethylation"

See allHide authors and affiliations

Science  14 Nov 2003:
Vol. 302, Issue 5648, pp. 1153
DOI: 10.1126/science.1089838

We have recently shown that genome-wide DNA hypomethylation induces genomic instability and tumors in animals (1, 2). Based on these results, we cautioned that long-term treatment with a demethylating agent such as 5-aza-2′deoxycytidine (5-aza-dC) may produce unwanted side effects such as tumor induction due to chromosomal alterations.

Based upon measurements of DNA methylation in colon tumor samples from untreated patients and patients treated with 5-aza-dC, Yang et al. (3) contend that our results should have no negative implications for the treatment of tumor patients with this drug. Their argument is based on the observations that (i) no substantial demethylation was detected after short-term drug treatment and (ii) that no or little DNA hypomethylation was seen in tumor samples compared to normal tissue when tested for Alu or LINE-1 elements. The latter finding is inconsistent with a large body of evidence demonstrating global hypomethylation in tumors [summarized in (4)] and suggests that DNA hypomethylation is not uniform throughout the genome or varies in different tumors.

In our model system, which used a genetic approach to inhibit DNA methyltransferase 1 (Dnmt1), the level of hypomethylation was more pronounced than in the patient samples after short-term drug treatment (the longest treatment in the Yang et al. study was not more than 2 weeks). However, although the Dnmt1 level was decreased by 10-fold in our system, the extent of genomic methylation was only moderately reduced in contrast to levels stated by Yang et al. (3). Using a similar model in a previous study (5), we showed that hypomethylation causes a significant reduction of intestinal tumor incidence in mice carrying a mutation of the APC tumor suppressor gene. Importantly, these early findings also demonstrated a beneficial prophylactic effect of the hypomethylation-inducing drug 5-aza-dC on tumor development, suggesting that a child carrying a mutation in this gene would benefit from the long-term administration of a methylation-reducing drug. Our present studies show that hypomethylation predisposes to some other forms of cancer by reducing genomic stability. These results raise the concern that any long-term prophylactic treatment with demethylating drugs aimed at protecting against cancer incidence in one tissue, may have the unwanted side effect of promoting tumors in other tissues. Another potential complication of using this drug for any long-term treatment stems from the previous observation that 5-aza-dC causes a significant increase in the rate of point mutations under conditions where it protects against intestinal tumors (6).

We fully agree with Yang et al. that such unwanted side effects may be of little relevance in relatively short-term treatment regiments for cancer patients. This is because the level of demethylation induced by drug treatment is lower than in our model. Therefore, the consequences of slightly increased chromosomal instability or increased rate of point mutations may be of little consequence for the patient and may be outweighed by beneficial therapeutic effects. Since no significant demethylation was observed in patients treated with 5-aza-dC, it should be considered that any therapeutic effect of the drug may not be due to decreased methylation levels, but to other effects such as Dnmt1-mediated drug toxicity (7). However, our study raises the concern that drug-induced hypomethylation may contribute to genomic instability as a consequence of long-term treatments, even if the effect is too small to be detected in short-term analyses.


Navigate This Article