The threshold of z=1.65 ist problematic because of unacceptable false positve rates if you use random field theory based significance test. If you use premutation tests (e.g. randomise or palm) this threshold should not be a problem.

HTH

Bertram

Perhaps you could mention that data as a "future study" topic in the Discussion section and post the results on ResearchGate or Academia.

Is this a second order analysis - e.g. a groupwise contrast?  If so I have consistently published with P

As has been mentioned above, the critical issue is the false discovery rate. A significance level of 0.01 would be permissible, provided you can show that the expectable rate of false positive clusters remains below 0.5 over the whole-brain image.

Do the analyses at the lower significance level deliver meaningful insights? If so, defend them whatever.

Although this discussion is closed, in general, instead of cluster-based inference low cluster-forming thresholds, I suggest cluster enhancement and then an FWER-correction, like e.g. FSL TFCE does it.

For a fast parametric solution, see also pTFCE: https://spisakt.github.io/pTFCE/