Yes, something along the lines of:
proc glimmix data=yourdata;
class timepoint subjectid;
model response=timepoint /*Insert any distribution or link functions here as needed*/;
random timepoint/subject=subjectid type=ar(1); /* ARH(1) may also be appropriate here. Also, if the response variable has a Gaussian distribution for residuals, then the residual option should be invoked here as well */
lsmeans timepoint/diff=control;
run;
This will give a test of each timepoint versus the baseline.
Steve Denham