On the problem of first correction in soliton perturbation theory