Match one part of a regular expression in a Kwrite find-and-replace regex

Among the joys of KDE, Kwrite is surely noteworthy. Having regular expression support built in by default into text editor is probably one of the top 10 best ideas of all time. Since I often need to do a lot of finding and replacing, I have already begun to save a lot of time and hassle by using Kwrite. I suppose it's probably a lot like in sed, but anyway, I am matching stuff like numbersdonot8 belongattheend0 ofwords2 using the simple regex [a-z][0-9]. How do I then tell kwrite to replace the text using the first part of the regex? Using [a-z] in the 'replace' box gives things like 'numbersdonot[a-z]' as a result. But I want, numbersdonot belongattheend ofwords