I asked Exact Sciences about the risk of 'spread' versus recurrence since the OncotypeDX (the test I had) doesn't specifically predict that. The answer is interesting and the person in the science section whom I spoke with is sending me more info. But, basically, the same TaylorRX data pool from which Oncotype based much of its algorithm (not a good explanation but to keep things simple) also included 'spread' as well as recurrence statistics. Inferentially, a low risk of distant recurrence correlates with, also, a low risk for spread from cancer site assuming clean margins. Cancer-free lymph nodes lowers that risk further. If only we could get 100% perfect data, decisions would be so much easier, sigh. But at least we have better tools than women even 20 years ago.