All of the Romance languages were Latin, but they slowly evolved over time. One of those evolutions tended to be a defaulting to the accusative followed by the dropping of the final consonant or syllable (depending). So, goodbye to -um, -am, -em, etc.

English was of course never Latin. Words with Latin roots were (rarely) imported from Latin during the Anglo-Saxon period, and more commonly taken from a Old French after 1066. These words thus also lack latinate case endings. During the early modern period, however, the language of science (and of all scholarship) was Latin. When people named things, they named them in Latin, and most of the -ium words in English seem to me to be sciency (have a link!). These are thus words that were pulled directly from Latin and thus have not undergone the usual mutations.

The OED only has the suffix -ium for chemical elements and chemical compounds. Although I'm not a French speaker it appears to be shared in French for words like Aluminium and Uranium.

Do you have an example of what you mean other than this one as it appears to be quite a specific case and not necessarily supporting your assertion.

I think you're actually looking at this the wrong way around—it's entirely logical that a non-Romance language would borrow classical Latin more intact than a Romance language. Romance languages over time change, and the Latin noun endings in particular were often systematically weakened till the final consonant was eliminated. Though Latin vocabulary was still borrowed—I'm thinking here of the slightly aberrant phrase grosso modo in French —for the most part, Romance languages already contained vocabulary that needed to be borrowed, because they're descended from Latin in the first place!