The printer’s manuscript doesn’t have the preposition of at the front of the relative clause, but the 1830 edition does. The original manuscript is not extant here. If the of was in 𝓞, it could have been accidentally dropped by Oliver Cowdery when he copied the text from 𝓞 into 𝓟. (There is evidence that Oliver tended to omit the of at the beginning of relative clauses. For some discussion and examples, see under Alma 40:19.) Or if the of wasn’t there in 𝓞, the 1830 typesetter might have added it because it seems necessary. Another possibility is that Joseph Smith dictated an of at the end of the relative clause (“which hath been spoken of ”) but that Oliver accidentally missed it when writing down 𝓞. Under this interpretation, the of at the beginning of the relative clause in the printed editions would be due to an emendation on the part of the 1830 typesetter. It is doubtful that 𝓞 had an of at the end of the relative clause since this would mean that Oliver dropped the of when copying to the printer’s manuscript and that the 1830 typesetter moved the of to the front of the relative clause. In other words, it doesn’t seem very plausible that there was an error here for both firsthand copies of the original manuscript (although that isn’t completely impossible).
When we consider all other instances in the text of the relative clause “which have been spoken” (where have can take variant forms such has hath, has, and had ), we find that of never occurs when the reference is to things actually spoken. There are 37 examples of this relative clause in the text; in 19 of these examples the antecedent for the relative pronoun which is words. On the other hand, when this relative clause “which have been spoken” refers to a topic, then the of always occurs (15 times, not counting this case in Helaman 16:16); sometimes the of occurs at the beginning of the relative clause, sometimes after spoken (each of the latter is marked with an asterisk):
What is interesting here in Helaman 16:16 is that Oliver Cowdery started to write words rather than the correct works in 𝓟. He immediately caught his error and corrected words to works. But apparently his initial words prompted him to omit the preposition of since that would have been appropriate if words had been the actual reading (that is, if the text had read “all these great and marvelous words cannot come to pass which hath been spoken”). Here in Helaman 16:16, 𝓞 undoubtedly read works since the 1830 edition and the corrected reading in 𝓟 agree in reading works. And this firm reading implies that the relative clause “which hath been spoken” should take the preposition of since works is a topic. The critical text will therefore maintain the 1830 reading for this passage.
Summary: Maintain in Helaman 16:16 the preposition of at the beginning of the relative clause “of which hath been spoken” (the 1830 reading) since the referent is works, a topic; Oliver Cowdery seems to have omitted the of in 𝓟 since he initially thought that works was words.