Check this glossary, the first three params described (Elapsed Time, Latency, Connect Time).
That is what JMeter is measuring on it's side - which is a client side, obviously.
While Dynatrace measures server-side processing only.
So both numbers are legit and gives you an important clue - you just have to read it properly.
Armed with that numbers, you'd probably like to deep into what happened in between, where this difference is coming from.
In a nutshell, in between lays: network; standard components of the server (OS & JBoss), including ones that serves networking stuff; same for client; and JMeter itself.
To get the first clue on that path, I suggest you to set up the sniffing proxy (e.g. Fiddler) on your JMeter host, manage all traffic to go through it, and see what it shows in its detailed time statistics for the networking operations.
This way, you'd have your JMeter excluded as a factor, so you can see how much it brings there.