As the article says, this immunotherapy essentially works by turning off some of the immune system's "don't attack your own body" safeties. This frees the immune system to attack the cancer (provided it can identify it).

One big advantage of immunotherapy is that the therapy causes a long-term (permanent? I'm not sure anyone knows; the drugs are less than a decade old) change to the immune system - the body continues to attack the cancer even after the drug treatment has ceased.

If you put these two things together, however, you might see why the side effects can be so bizarre. Your immune system might decide that the liver looked at it funny one day, but then decide that the GI tract is actually the real troublemaker a few days later. And since the immune system change persists post-treatment, so do the side effects.

These side effects can (usually) be controlled with corticosteroids (hydrocortisone/prednisone/etc.), which have an immunosuppressant effect. Of course, long-term corticosteroid use has its own severe side effects (plus severe dependency issues). And if you're suppressing the immune system, then you're suppressing the mechanism that the immunotherapy uses to fight the cancer in the first place. The treatment and the side effects are two sides of the same coin.