Can we apply statistics on replicated data (Nested repeated measure)?

Dear Experts,
In my study I have 2 groups i.e controls (n=23) and patients (n=30). And again in patients, I have 2 sub-groups i.e subgroup1 (n=15) and subgroup2(n=15).
I have spinal cord area (SCA) at 4 different spinal cord levels (C1, C2, C3, C4) from these groups.
I would like to investigate A) if the SCA is significantly different in Controls compared to patients at all spinal cord levels (C1, C2, C3, C4);
B) if the SCA is significantly different between the controls and the subgroups i.e. controls vs subgroup1; controls vs subgroup2 and subgroup1 vs subgroup2;
Now my main question is whether I can test both #A and #B in a single model i.e. if I apply ANOVA for each spinal cord level and use controls, patients, subgroup1, subgroup2 to build ANOVA, is this the right way to do statistics, because I am afraid that I am replicating the patients data twice, as subgroup1 and subgroup2 are the subgroups patients data.
So far, I have tried the following options in stata:
option 1: one mixed model comparing Controls and Patients, and pairwise postestimation; and an additional mixed model comparing the 3 groups (controls, subgroup1, subgroup2), both cases with postestimation comparison. i.e. mixed sca || subjId: cordLevelNames ,cov(un) pwcompare, effect mcompare(bonferroni)
option 2: I found nested anova, or nested mixed models which look to be the ideal solution. but nested anova is not working in my smallStata version because of memory; whereas nested mixed models doesn't allow to check for all levels comparison: i.e.
mixed sca i.patientGrp##c.cordLevelNames || groupName: || subjId:cordLevelNames, cov(un)
// postetstimation capture drop fit predict fit lincom cordLevelNames + 1.patientGrp#cordLevelNames
pwcompare i.patientGrp#c.cordLevelNames, effect mcompare(bonferroni) pwcompare i.groupName#c.cordLevelNames, effect mcompare(bonferroni)
I would be glad if you could give some suggestions
thanks in advance for your help
Best regards