Corrections to the semiclassical approximation in nearly forward Planckian energy collisions are here reconsidered. Starting from the one-loop superstring amplitude, we are able to disentangle the first subleading high-energy contribution at large impact parameters, and we thus directly compute the one-loop correction to the superstring eikonal. By comparing this result with previous ones by Amati, Ciafaloni and Veneziano (ACV) for pure gravity, we identify one-loop gravitino contributions which agree with previous results by Lipatov. We finally argue, on the basis of analyticity and unitarity, that gravitinos do not contribute at all to the large distance two-loop ACV correction, which thus acquires a universal "classical" interpretation.