Eh, nobody should be surprised. This is how you adapt when throwing temporary help at a problem isn't a viable resolution (you can't afford it or the ramp-up takes too long, for two examples). If you lose a lynch-pin you redistribute the work however possible and reprioritize it.
The simplest example in this:
Cycle 1:
Employee A is responsible for: A1, B1, C1, D1
Employee B is responsible for: A2, A3, B2, and D2
Employee B is suddenly gone for the cycle.
Employee A is now responsible for: A1, A2, A3, and B1 while B2, C1, D1, and D2 are moved back into the backlog.
It's safe to assume that the "A" and "B" tasks / stories / projects are revenue drivers or blockers for revenue drivers. "D" rank projects are things like free give-aways. When the bottleneck hits the nice-to-haves and give-aways get tossed into the backlog first because otherwise you don't meet your revenue guidelines to the extent that you could have and you ignored the business's prioritization. The latter is a fast-track to getting fired.
The kanban giveth; the kanban taketh away.
Marty Lund