A proof-of-work blockchain cannot be destroyed. Nodes follow the best (valid most-work) chain. They won't follow a weaker chain (e.g. one that erases the ledger). And their copies of the blockchain are globally distributed so they can't be "deleted." You can hard fork to a new empty ledger, but that will still leave the old fork intact -- anyone can run a node and pick up where it left off.

In contrast, a proof-of-stake blockchain can be destroyed. All it takes is for 51% of staking validators to collude in destroying the ledger. This can be remedied with centralization by hard-coding checkpoints into future versions, but there are two problems: 1) that requires a high level of centralization and thwarts the purpose of decentralized consensus, and 2) all ledger history past the last hard-coded checkpoint is always at risk of deletion. This is one of the reasons why POS is fundamentally less secure -- or more centralized, or both -- than POW.