VO2 max is a bad predictor for race performance. Yes the aerobic capability must be extremely developed to run fast, but the best runners have in fact a suprisingly high variation in VO2max. Also for a group of fast runners, it seems that RE = Running Economy (O2 used per pace) is inversely correlated to VO2max. One starts to believe that increasing VO2 max might reduce the RE. But this might be false. One explanation is in fact that VO2max and RE is not really connected, they just seem to be statistically correlate. It is obvious that running at a certain pace is a piece of work meaning to get your body weight from A to B in a certain time. This piece of work needs a minimum amount of energy per time and can be calculated as VO2 minimumly required. If you are very efficient you would need less, if you are less efficient you need slightly more VO2. If your VO2max is too low, that might limit the maximum marathon pace (but not really for the best as it is impossible to run too close to VO2max for 2 hours), if it is high you might have plenty to run a fast marathon.
The runner comes in a variety of shapes and forms and so also the combo of VO2max and Running Economy. And the fact is nevertheless that Vo2 cannot be a limiting factor or else you wont run fast enough for such a long time. If the VO2max is relatively low, the RE needs to be high to utilize the VO2 efficiently.
Both Vo2max and RE are individually bad at predicting race performance. Combined they may be a little bit better, but actual running performance is obviously better at predicting, since it is not predicting performance, but is rather measuring it directly.
Increasing VO2max is also questionable. For young runners, the body while growing is increasing VO2max a lot. But developed runners tend to NOT develop VO2max at all in their prime. It does not seem to be a good indicator or something that isolated gives improvements. Running Economy is connected to how efficient (used O2) one is at a certain pace and if you run faster at the same VO2 consumption you are per def more efficient and you have per def improved the pace. So RE as an indicator relative yourself is a good and direct indicator the same way as time trialing.
Science and statistics can be used the good way and the wrong way. One can easily be fooled by numbers.
But the good trainers out there dont bother too much about the numbers, as it is so easy to measure improvement in running directly.
I guess the perfect indicator for marathon performance is running a marathon...