As others have explained in the comments, you haven't quite defined what a "twist" is. However, perhaps what follow will still be helpful and matches what your intuitive picture of "twist" means.
Given a (path connected, semilocally simply connected) space $X$, you can construct its orientation cover $p : \tilde{X} \to X$, which is a double cover of $X$. If $X$ is orientable, then $\tilde{X} = X \times {\pm 1}$ is a trivial double cover (and disconnected). There's no "twist" and no "torsion" here.
If $X$ is not orientable, then the cover is nontrivial. The group of deck transformations of $p$ is a subgroup $\Gamma$ of index $2$ of $\pi_1(X)$. The quotient is $\mathbb{Z}/2\mathbb{Z}$ and contains a torsion element of order $2$. You can think of this element as measuring "twists": if you have a loop $\gamma$ that lifts to a path which isn't a loop (i.e., it's a nonzero element in $\pi_1(X)/\Gamma$) then there is some kind of "twist" along this loop. But if you run twice along the loop, you'll always get zero, so the "twist" is always going to be "2-torsion" (in the group-theoretic sense).
tl;dr: When you say "twist", you're probably thinking of non-orientability, and this corresponds to 2-torsion in a certain quotient of the fundamental group.
As far as I can tell, it's unrelated to (2-)torsion in a topological group. Lie groups are always orientable.