VARCHAR is specific to your system codepage. NVARCHAR holds Unicode characters. Unicode is always going to be a superset of any legacy codepage. As such, you can expect your conversions from VARCHAR -> NVARCHAR to map properly. However, if your NVARCHAR string later contains Unicode characters that are not found in your legacy codepage, the characters will be lost when converting back to VARCHAR.
When possible, you should avoid this type of conversion. Update the table column to NVARCHAR.