XPath Evaluation without DOM in .NET

I'm working on a .NET project that requires XPath evaluation over documents, but that must not load the XML DOM in memory, where cases are that the contents to analyze can be of big dimensions (in the order of gigabytes). This is closer to the SAX model that is more common in Java, but it's not very common in .NET.

Back in the days, someone at GotDotNet created a nice library (still downloadable at http://ift.tt/1IZ07Rv) named XPathReader, which mimics the SAX model, evaluating XPath expressions in a forward-only direction, that is pretty much what I'm looking for. I see that Microsoft has developed the concept also inside BizTalk (http://ift.tt/1IZ09sS), which seems logic in that context, but it hasn't make it a general purpose library anywhere.

Since the XPathReader library is unmaintained since 2004 (11 years!), I'm wondering if any other effort is known on this field, that copes with more support behind and more in sync versions.

Thanks for all your input!