Related Experiment Videos

Anticarcinogenic responses in rodent cancer bioassays are not explained by random effects

I Linkov1, R Wilson, G M Gray

  • 1Department of Physics, Jefferson Laboratories, Harvard University, Cambridge, Massachusetts 02138, USA.

Insights

This study reveals that many anticarcinogenic findings in rodent bioassays may be false positives due to random variation. However, biological factors likely contribute to true anticarcinogenic responses, which are predictable across species.

Area of Science:

  • Toxicology
  • Carcinogenesis and Chemoprevention
  • Statistical Modeling in Biological Assays

Background:

  • Anticarcinogenicity in rodent bioassays is defined as a statistically significant decrease in tumor type following chemical exposure.
  • A high percentage of chemicals tested by the National Toxicology Program show decreased tumor rates, consistent with previous findings.
  • Tumor rate decreases can result from biological effects or random variability, decreased body weight, or reduced survival.

Purpose of the Study:

  • To evaluate the rate of false-positive anticarcinogenic findings attributed to random effects, such as tumor rate variations and multiple comparisons.
  • To assess the impact of statistical significance cutoffs on false-positive determinations.
  • To examine the cross-species predictability of anticarcinogenic responses.

Main Methods:

  • Monte Carlo simulations were used to assess the contribution of random effects to false-positive anticarcinogenic findings.
  • Statistical significance levels (p < or = 0.05, p < or = 0.01, p < or = 0.005) were analyzed.
  • P-value distributions were examined, and cross-species prediction of anticarcinogenic responses was evaluated.

Main Results:

  • Random effects were found to be a significant contributor to false-positive anticarcinogenic findings, even at a p < or = 0.05 significance level.
  • Using more stringent statistical criteria (p < or = 0.01 or p < or = 0.005) reduced the proportion of false positives.
  • The observed number of anticarcinogens exceeded predictions from simulations, and statistically significant anticarcinogenic responses occurred more frequently than expected by chance.
  • Anticarcinogenic effects demonstrated good predictability between different rodent species.

Conclusions:

  • While random variability contributes to false positives, biological factors are likely involved in true anticarcinogenic responses observed in rodent bioassays.
  • Anticarcinogenic effects identified in one rodent species are generally predictive of effects in another.
  • The findings highlight the importance of rigorous statistical evaluation in interpreting rodent bioassay results for anticarcinogenicity.

Related Concept Videos