I read some meta-analysis studies and subgroup analysis was done in the studies because the results showed great heterogeneity (in this case was obvious, the I-squared test >90), regardless of the significance of the overall effect.

Colleagues who can share knowledge about "the cut-off value" (if any) for the I-squared test that is mandatory (or recommended) for subgroup analysis will be appreciated. Furthermore, should the subgroup analysis be done in a meta-analysis in which the number of studies is limited (e.g ten studies) when the data in it is heterogeneous?