Even Sir Richard Peto (second author) is seeing the light. Sir P is famous for stating (loosely remembered by me) that no individual study of the role of cholesterol in CVD is particularly convincing, but the overall weight of evidence was. I'm stuck with reading study after study and realising they are crap. A big heap of crap is no more convincing than a small heap, to me. But then I'll never get a knighthood.
Back to the Oxford abstract:
"Given usual apoB, lower LDL-C (consistent with smaller LDL particles) was associated with higher risk (P < 0.0001)."
Translation: at a given number of apoB100 particles, the lower the measured LDL the HIGHER the risk.
"The ratio apoB/apoA(1) was substantially more informative about risk (chi(1)(2) = 550) than were commonly used measures such as LDL-C/HDL-C, total/HDL cholesterol, non-HDL cholesterol, and total cholesterol"
Translation: Most of what we have measured in the past is bollocks. Our new ratio is slightly less bollocks.
Notice they didn't mention HDL/trigs. And they still believe sdLDL is out to get you. And no one has pointed out to them that sugar is a great generator of sdLDL.
I've not bothered down loading the free full text. The abstract says exactly what you would expect a real view of the world to say. That's enough to take notice of without working too hard.