The off-score is a way of judging the relationship between total red cells/Hb and new cells or retlics. Withdrawal or transfusion or EPO used alters the balance. In the old days, a score of 130 or so flagged a rider. As I noted before, we still have not seen data from LA that reaches that level.
The passport is more sophisticated, takes additional factors into account. By comparing values to a baseline, it is a little better at flagging as suspicious values that aren't quite as high. IOW, in the old system, the 130 value was based on what anyone should have. It meant that a rider's value was significant compared to the population at large. In the passport system, the value is compared to what the rider himself should have.
As an analogy, suppose everyone undergoes normal fluctuations in weight. In the old system, you wouldn't flag someone unless he gained or lost, say, 20 pounds in a month, because a study of a large population indicated that would be extremely unusual. In the new system, you might flag a specific individual if he gained ten pounds in a month, because based on his past history, this would be extremely unlikely for him, even though it might be quite plausible for someone else.
So while LA's blood values would not raise a flag based on off-score, they might raise a flag based on passport. There are complex algorithms into which you plug the numbers and allow you to see. My guess is that LA's values did not trigger a flag using these algorithms, either. If they did, the values should have been given to a panel of experts, who would have made the final decision. It may be it went to a panel and they decided he passed. A third possibility, of course, is a coverup.
Assuming it's not a coverup, though, USADA will make their case by combining these values with testimony. E.g., if they can get a teammate to indicate manipulation around the same time a large fluctuation occurred, that would strengthen their case. If they can show this on two or three different occasions, it would be extremely powerful evidence. In this case, the fact that the values did not trigger a positive in the algorithms could be overruled.