Pseudo-time lapse experiment: repeated measures ANOVA??

I am measuring disease severity, using 5 different types of measurements, in animals treated with drug X, over multiple timepoints. The measurements can only be made when the animal is dissected. We have 30 animals, all given the same drug X on day 0, and 5 animals are taken down every 2 days.
To minimize numbers of animals, we do not have a concurrently running untreated control group.
We find that with 3 of the measures of disease severity, there is a significant (using Dunnett's test) increase on days 4, 6, and reduction on day 8, compared to day 0.
For 2 of the measures there is no increase until day 10 and 12.

1: Is the Dunnett's test correct for identifying changes in disease severity in days compared to day 0.
2: I want to understand the correlations between different measurements of disease at different timepoints. I have biological reasons to suspect that measures of gene products in the same pathway should have similar time dependent expression patterns. How do I test whether the measurements showing visually similar patterns are:
A different from the others.
B similar to each other.

To look for similarity between measurements do I just do a linear regression analysis between different measurement methods for all animals, and measure Pearson's correlation coefficient? That seems to totally ignore the time-course nature of the experiment, so I am worried.

The set-up sounds a bit like a heirarchical clustering type problem, but I have only 5 measurements...

I am confused whether this experimental setup means the data should be treated like a time-lapse experiment (repeated measures ANOVA?) or analyzed independently.

Please help me....