Can Anyone Work Out What Is Going On Here?

ALEKHINE'S Defence (1.e4 Nf6) seems to be making something of a comeback in club chess, assuming it ever went away.
At least that has been my recent experience: I have faced Alekhine's Defence three times in the past year, which is as many times as in the previous ten years put together.
It has made me think I may need to come up with something more forcing than my usual reply of 2.Nc3.
The obvious place to start is ChessBase's 2019 Mega database. It features more than 61,000 games with 2.e5, with White scoring 54%. There are more than 22,000 games with 2.Nc3, when White's success rate drops to 48%.
Lots of other second moves have been given an outing, but only one has given White a plus score: 2.c4!?
White scores 55% - one percentage point more than with 2.e5 - from 30 games. That is a small sample size, but it is not as if 2.c4!? has only been played by rabbits. Among those with White have been Ildar Ibragimov (2637), Gennadi Kuzmin (2546), Peter Lukacs (2460) and Ta Lin (2410).
Now comes the really bizarre bit - only two games featured the reply 2...Nxe4, and both those were games between unrateds.
Why do players refuse the gift? What is it they fear?
Taking the four top players to have tried 2.c4!? in chronological order, we get the following:

Peter Lukacs (2460) - Nelson Pinal (2405)
Capablanca Memorial Group B 1983
1.e4 Nf6 2.c4!? g6!? 3.Nc3 Bg7 4.d4 d6
The game has transposed to a mainline King's Indian, which White won in 22 moves.

Ta Lin (2410) - Ian Rogers (2515)
Kuala Lumpur Zonal 1990
1.e4 Nf6 2.c4!? e6 3.Nc3 Bb4 4.f3!? d5
The game has transposed to a one-off position from the Flohr-Mikenas System in the English. It was apparently drawn in 17 moves, although the whole score in Mega19 looks suspicious (the final position is winning for White, according to Stockfish10 and Komodo9).

Gennadi Kuzmin (2546) - Andrei Kovalev (2459)
Ukraine October GM 1999
1.e4 Nf6 2.c4!? g6!? 3.Nc3 Bg7 4.d4 0-0 5.Be2 d6
The game has transposed to a mainline King's Indian. It is recorded as a win for Black in 22 moves, but in the final position White has a mate-in-eight and is anyway massively ahead on material.

Ildar Ibragimov (2637) - Elliott Liu (2162)
Las Vegas National 2006
1.e4 Nf6 2.c4!? g6!? 3.Nc3 Bg7 4.d4 d6
Once again the game has transposed to a mainline King's Indian. White won in 73 moves.

So what can be concluded? Sadly, I have not stumbled across a challenge to the Alekhine, let alone a refutation.
The fact is, despite the evidence from Mega19, c4 on move two against the Alekhine should be annotated 2.c4?! or, more probably, 2.c4?
Secondly, and more worryingly, there is some very dodgy data in Mega19, which can seriously skew statistics when sample sizes are small.