Without the multipole-expansion, the nucleus-electron interaction would be described by a very complicated integral-term, which has no general analytic solution. If a perturbation series is to be found, this interaction term would have to be evaluated first, which might not be possible analytically at all.
I think it would be possible to evaluate the integral numerically, however.