It seems a bit of an arbitrary choice. For example, Polish has digraphs, but Czech has diacritics, and Icelandic has a couple of additional letters that aren't modified Latin letters.
Old English had four letters that are not in today's US-ASCII, two of which are borrowed from a runic alphabet rather than created by modifying Latin letters.
It's also a bit arbitrary whether a modified Latin letter is regarded as a new character or an existing character modified by a diacritic: take Ø, for example. And there are characters like Æ. And never forget what Turkish did: add dotless i so that ordinary i could then be regarded as dotless i plus a diacritic (though of course it isn't usually regarded that way).
Old English had four letters that are not in today's US-ASCII, two of which are borrowed from a runic alphabet rather than created by modifying Latin letters.
It's also a bit arbitrary whether a modified Latin letter is regarded as a new character or an existing character modified by a diacritic: take Ø, for example. And there are characters like Æ. And never forget what Turkish did: add dotless i so that ordinary i could then be regarded as dotless i plus a diacritic (though of course it isn't usually regarded that way).