Stepping through the analysis, there are large swings in favour of black on most turns where the white knight on e4 is only protected by the pawn on d3. I'm guessing that Stockfish really likes the passed pawn on the d file.

After the white knight goes to d4 on move 10, Stockfish only thinks that white is okay after its move on move 11 (after the other knight moves to d2 and becomes a second defender of the knight on e4) and on move 14 (when there's a chance to pick up black's undefended knight on e5 in exchange for the knight on e4). As long as white can keep material even without allowing a passed pawn, Stockfish thinks it's okay. On every turn where allowing a passed pawn is the only way to keep material even, Stockfish thinks it's bad for white.

I'm no expert, but that's my guess.

Stepping through the analysis, there are large swings in favour of black on most turns where the white knight on e4 is only protected by the pawn on d3. I'm guessing that Stockfish really likes the passed pawn on the d file.

After the white knight goes to d4 on move 10, Stockfish only thinks that white is okay after its move on move 11 (after the other knight moves to d2 and becomes a second defender of the knight on e4) and on move 14 (when there's a chance to pick up black's undefended knight on e5 in exchange for the knight on e4). As long as white can keep material even without allowing a passed pawn, Stockfish thinks it's okay. On every turn where allowing a passed pawn is the only way to keep material even, Stockfish thinks it's bad for white.

I'm no expert, but that's my guess.

Edit: I'm not sure how to include Stockfish scores in board notation here, so here's a picture.

Starting on move 10, Stockfish says that ...Nxe4 is the only dominating move for black, and it is a dominating move only when dxe4 (giving black a passed pawn) is the only way to for white to equalize material.