Prescreen Phenotyping Measures
Values obtained during the prescreen phenotyping from the novel object task, circulating corticosterone collections, open field locomotion, and elevated plus maze task are listed in . All parameters measured in these rats fell within the range of expected values based on the literature and our prior experience.
Values Obtained in Prescreen Test Battery
Consumption During Ethanol-Only Phase
During the 3 days when ethanol was the only available fluid, the rats as a group exhibited a general decline in ethanol consumption [RMANOVA, F(2,70) = 49.48, p < 0.0001; see ]. There was a significant decrease in consumption from days 1 to 2 [F(1,35) = 39.47, p < 0.0001] and from day 2 to day 3 [F(1,35) = 6.81, p = 0.0132]. There was substantial inter-individual variation in consumption which increased across the 3 days (see ).
Fig. 1 Time course of ethanol consumption. (A) Average consumption (g/kg) across the time course of the experiment. Error bars, standard error, *, significant difference between choice day 10 and relapse day, RMANOVA, p < 0.05. (B) Grams of ethanol consumed (more ...)
Consumption During Ethanol Versus Water Choice Phase
During the choice consumption phase, rats initially avoided the ethanol-containing solution and consumed mostly water. Over the course of 10 days, however, the group exhibited an increase in ethanol consumption [RMANOVA, F(9,315) = 13.005, p < 0.0001; see ]. Consumption on day 10 was significantly greater than consumption on day 1 of the choice phase [RMANOVA, F(1,35) = 19.567, p < 0.0001]. There was also an increase in inter-individual variation in consumption (see ). This result suggests that individual patterns of consumption emerged gradually over the 10 days of the 2-choice period. On average, the rats exhibited an increase in consumption following the deprivation period. Ethanol intake (g/kg) was significantly greater on the rebound day than on day 10 of the choice phase [RMANOVA, F(1, 35) = 8.677, p = 0.0057; ].
Preference for ethanol (percent of total fluid intake) also increased over the course of 10 days of choice consumption in a pattern similar to g/kg consumption [RMANOVA, F(9,315) = 7.156, p < 0.0001; ]. There was a significant increase in preference for ethanol from choice day 1 to choice day 10 [F(1,35) = 7.417, p = 0.0100, RMANOVA], and from choice day 10 to the relapse day [F(1, 35) = 11.201, p=0.002, RMANOVA, ]. Alcohol preference was significantly correlated with total alcohol (g/kg) intake on choice days 8 to 10 (R2 = 0.95 p < 0.0001) and on the relapse day (R2 = 0.90, p < 0.0001), but not on choice day 1, when little ethanol was consumed by all of the rats.
Fig. 2 Time course of ethanol preference. Ethanol fraction of total fluid consumption during 8% ethanol versus water (choice) days 1 to 10, and relapse day. Closed circles, rats that were in the highest tertile of consumption on forced day 3, open circles, middle (more ...)
Individual Differences in Consumption
Individual variation in ethanol consumption patterns began to emerge during the forced consumption phase, and was maintained throughout the course of the experiment. As shown in , there is variance in gram EtOH consumed per kilogram during the 3 days of ethanol-only consumption. This variance largely disappears when water is offered as an option (choice days 1 to 2), but re-emerges over 10 days of choice consumption. Individual levels of ethanol consumption which were established by day 3 of the ethanol-only phase reemerged over the 10 days of the choice consumption period. There is a significant positive correlation between ethanol consumption on forced day 3 and ethanol consumption on choice days 8 to 10 (R2 = 0.13, p=0.029; ) as well as a significant correlation between forced day 3 and relapse day (R2 = 0.28, p = 0.0009; ). However, the level of early consumption does not fully account for the variation in consumption in the later days measured. Consumption on forced day 3 accounts for only 13% of the variance (R2) in consumption on days 8 to 10, and 28% of the variance in consumption on the relapse day. Thus, other factors, as yet unidentified, determine the level of postdeprivation consumption.
Fig. 3 Scatterplot showing the relationship between early and later ethanol consumption. (A) g/kg consumed on forced day 3 is significantly correlated with g/kg consumed on choice days 8 to 10. (B) g/kg consumed on choice days 8 to 10 is significantly correlated (more ...)
By days 8 to 10 of choice consumption, patterns are strongly established. There is a significant correlation between consumption on days 8 to 10 of the choice phase and consumption on the relapse day (p = 0.0002; ). The variance in consumption on days 8 to 10 accounts for 35% (R2 = 0.35) of the variance in relapse consumption.
These correlations are probably not the result of consistent patterns of thirst. There was no correlation between consumption of water during the first night in the lickometer cages and consumption of ethanol on ethanol-only nights, nor was there a significant correlation between water consumption on the first night and ethanol consumption on days 8 to 10 of the choice phase, or postdeprivation ethanol consumption (data not shown).
Patterns of alcohol preference were similarly established early in the experiment and maintained through the postdeprivation consumption test. As shown in , consumption (g/kg) on Forced day 3 was significantly correlated with postdeprivation preference (R2 = 0.26, p = 0.0015). In addition, preference on days 8 to 10 was significantly correlated to preference on relapse day (R2 = 0.22, p = 0.0039). As mentioned above, preference was highly correlated to consumption within days 8 to 10 and relapse day (i.e., preference on days 8 to 10 was correlated with g/kg consumption on days 8 to 10, etc.). These data suggest that preference, along with consumption pattern, is established early in the experiment.
Ability of Prescreen to Predict Alcoholism-Like Behaviors
We hypothesized that our prescreen measures would be useful for predicting relapse-like behavior. To test these predictions, we assessed correlations between postdeprivation ethanol consumption and each of the prescreen measures individually. In this analysis, only the number of entries into the open arms of the EPM was significantly correlated with postdeprivation ethanol consumption (R2 = 0.17, p = 0.0138). None of the prescreen measures were correlated with drinking on days 8 to 10 of the choice phase or consumption on forced day 3.
We also tested whether the prescreen measures could account for additional variance in relapse drinking when combined with consumption on forced day 3. In a multiple linear regression model, a significant beta coefficient was obtained for open arm entries in the elevated plus maze when combined with consumption on forced day 3. However, the model containing both forced day 3 and open arm entries together (R2 = 0.39; p = 0.0003) did not account for significantly more variance in rebound drinking than forced day 3 alone (R2 = 0.28; p = 0009) (F-ratio, 2 factor model vs. 1 factor model = 1.12; p > 0.05). Similarly, a model containing choice days 8 to 10 and open arm entries did not predict significantly more variance in relapse drinking than a model containing choice days 8 to 10 alone. Overall, our prescreen was not useful for predicting postdeprivation consumption.
Test of Predictive Value
As we performed so many correlational analyses in the original data set, it was possible that we obtained false-positive results. Therefore, to confirm our conclusions we tested an additional 12 rats to see if the same relationships emerged. As shown in , consumption on forced day 3 was again correlated with consumption on days 8 to 10 of the choice phase (R2 = 0.46, p = 0.0153) and there was a trend toward a correlation between forced day 3 and rebound consumption (R2 = 0.25, p = 0.099). Postdeprivation consumption was also significantly correlated to ethanol consumption on choice days 8 to 10, as in the original model (R2 = 0.89, p < 0.0001).
Our findings relating to alcohol preference were nearly but not exactly replicated in the confirmatory group. Preference on days 8 to 10 was significantly correlated to consumption on forced day 3, which was not observed in the original group (R2 = 0.39, p = 0.029, ). Preference on the relapse day was nearly significantly correlated to consumption on forced day 3 (R2 = 0.31, p = 0.06, ). The relationship between entries into the open arms of the elevated plus maze and relapse consumption was not replicated, suggesting that the weak correlation in the original group may have been a false positive result. All other relationships were replicated in the confirmatory group as indicated in .
Based on the results from these 2 separate groups of rats, we conclude that relapse-like ethanol consumption and preference in adolescent male rats is partially predicted by early ethanol consummatory behaviors. Individual consumption and preference levels continue to emerge across the choice phase, but are partially established early in the experiment.