ICP-MS because of the number of species that you can measure and the lower limits of measurement. If you are doing such work please look at the ISO/IEC quality requirements. Without satisfying hem the work may not pass peer review.
In general, I agree ICP-MS would be the "better" option, due to the criteria Paul Milham suggested.
However, you may find that MP-AES is sufficient for your purposes if (1) the concentrations of metals are high enough and (2) the cost of running the ICP-MS exceeds the number of samples you have. In this case the MP-AES is the "better" option, even if detection limits are higher and element choices are more limited. You may have to compromise somewhere.
I agree with above mentioned . In our study we successfully used ICP MS for estimation of heavy metals in pharmaceuticals, without sample preparation. Article Comparative Toxicity of Interferon Beta-1a Impurities of Hea...