In this paper, we tackle the problem of link adaptation in closed-loop turbo coded multiple-input multiple-output antenna systems with non-linear turbo receivers performing iterative linear minimum mean-square error (soft) interference cancellation (LMMSE-IC) and turbo decoding. The proposed link adaptation scheme relies on a low rate feedback and operates joint spatial precoder selection (e.g., antenna selection) and modulation-coding scheme selection so as to maximize the average rate subject to a target block error rate constraint. The paper first revisits in detail the architecture of the whole turbo receiver, then presents a semi-analytical performance prediction method to analyze its evolution through the stochastic modeling of each of the components. Based on this PHY-layer abstraction, average predicted and simulated throughputs are shown for different communication scenarios.