Doing it, safely and correctly, with reg exp is a lot harder. You will have to ensure that href is indeed part of an a element, not commented out etc.
With DOM, you don't need to think about these things.
One line is not sufficient wether you use reg exp or DOM, since you will need to specify what you want first, and then iterate over the results. Using DOMNode::getElementsByTagName would however require an extra if check inside the iteration to see what the href attribute starts with. What's so bad with this approach?
And if you just have to cut your code by that one line, use XPath:
//a[starts-with(@href,"/stuff")]');
