Good question. My vague understanding is that antibodies react to the external shape of the virus, the spike and so on. Whereas T-cells react to the little chunks of rna produced during multiplication of the virus, kindof the byprodyucts of a virus-factory-cell, and as such are not restricted to the 'outside' of the virus but can potentially learn about any part of the virus' genome. Hence T-cells are less likely to be thrown off by spike mutations. But my grasp on this is very vague.