I presume that you mean the set of transition regions that begin at bar 36.
Looking at those transition regions, I can see that they’re quantized to three bars, and that the loop region ends slightly before the third bar does. As a result, those regions are never actually being triggered, as the playback position inevitably reaches the end of the loop region and is sent back to its start, and never actually arrives at the transition regions’ quantization point at the start of bar 39. This loop region sends the playback position back to a beat before the end of bar 18; from there, as you have observed, it plays out that final beat of that bar, then hits one of the quantization points at the start of bar 19.
There are a few different ways you could change this behavior, but all revolve around ensuring the transition markers’ quantization point comes before, or at the same time as, the end of the loop region.
It might be relevant to know that if multiple transition points or loop region end points are at the same point on the timeline, their conditions are evaluated from top to bottom - such that if two or more things that could jump the playback position are at the same timeline position, only the topmost will affect the playback position at that timeline position. Thus, if you intend the loop region to affect the playback position only if none of the transition regions are active, it’s best to drag the loop region lower than the transition regions.
Personally, I’d lean towards replacing those three transition regions with transition markers, as you seem to only want the transition to occur at one specific point on the timeline instead of a range.