Except as otherwise noted, this info comes from an article on p396 of the
March 1994 _Diabetes_ by researchers at Eli Lilly.

Insulin is a protein. Proteins consist of sequences of amino acids. Human
insulin has the amino acid lysine at position B28 and proline at position
B29.

Insulin molecules naturally pair off (like people) and combine into dimers.
The dimers interact with small amounts of zinc and combine into hexamers, the
form sold as "regular" insulin.

   From another source, now forgotten: the time required to disassociate the
   hexamer into the dimer, and then the dimer into the monomer so that it
   can be absorbed, is the main reason for the delay in the action of regular
   insulin and the reason for injecting it 30 to 45 minutes before meals.

Switching the B28 and B29 positions on the protein has no effect on the
normal activity of the insulin but inhibits the formation of the dimer and
the hexamer. Thus the insulin is in monomeric form when injected and can be
absorbed immediately.

The name LysPro comes from the names of the amino acids, lysine and proline,
that occupy the swapped positions. According to an article in the August 1996
Diabetes Forecast, the spelling 'lispro' is now preferred.

Challenges in the development include the biochemical process for swapping the
amino acids, and making the result reasonably stable in the monomeric form.

 From another source, now forgotten: US FDA approval was not automatic, since
the insulin molecule has been modified. In fact, several other amino acid
exchanges have been tried and met with unacceptable side effects.

Some points from the article in the August 1996 Diabetes Forecast:

   Patients with gastroparesis, or taking acarbose, should be careful with
   lispro. Gastroparesis is a condition caused by neuropathy which causes
   the stomach to empty slowly and erratically. (See the section on
   gastroparesis later in this section.) Acarbose is a drug which slows
   the absorption of carbohydrates from the intestine. Either may result
   in lispro insulin acting too quickly.

   Response to lispro is variable. Some patients love it, others hate it.
   On the average, it does not change bg control either for better or for
   worse, but some patients definitely find it one or the other. Eli Lilly
   is promoting lispro for convenience, not for better control.

   Doctors and patients are still experimenting with the best regimens for
   using lispro insulin. "Best" clearly varies from one patient to another.
   Typically lispro insulin is injected very close to mealtime.

An obvious concern is that hypoglycemic reactions might be more common with a
faster acting insulin. A paper presented at the 1996 ADA Scientific Papers
conference studied this possibility:

   Reducing the Incidence of Hypoglycemia with a Novel Insulin Formulation
     J. Anderson, R. Brunelle, A Pfeutzner et al.
     Indianapoils, IN and Bad Homberg, Germany

In fact, they found the rate of hypoglycemic incidents slightly lower among
those using lispro insulin. They found no difference on most other measures,
including especially HbA1c. I've only seen the abstract of the paper, so I
know nothing about their methodology. (They also state the lispro forms
hexamers just like regular insulin but that the hexamers dissociate much more
quickly. I don't know who to believe, but from a practical point of view it
doesn't matter.)