The occurrence of hypoglycemia upon cessation of GH treatment sounds to me like some kind of rebound effect that results in a functional increase in insulin levels ("functional" meaning there isn't an actual increase in insulin but the body is acting as if there is.) I think it probably works something like this:
(1) One of insulin's many roles in the body is to regulate the amount of glucose circulating in the blood by signaling muscles and the liver to withdraw glucose from the bloodstream and store it as glycogen when glucose levels are too high (hyperglycemia). One of the roles of glucagon, otoh, is to signal the liver to make new glucose (gluconeogenesis) and release glucose into the bloodstream when glucose levels are too low (hypoglycemia). Insulin levels are typically lower than expected in those with PWS and we can infer from the Stefan study that it is possible that glucagon and gluconeogenesis are also low in PWS. The combined effect of low levels of insulin, glucagon and gluconeogenesis in PWS is likely to result in an impairment of glucose availability for "burning" for energy.
(2) GH is well known to be diabetogenic, that is, it can increase insulin resistance. Functionally, insulin resistance means that the body is acting as if insulin levels are lower than they actually are, with the result that blood glucose levels stay higher. Normally this is considered to be a bad thing because too much glucose in the bloodstream has damaging effects over the long-term, but it may be that, paradoxically, an increase in insulin resistance is actually something of a good thing in PWS, particularly in early infancy and childhood.
(3) To understand how a GH-induced increase in insulin resistance may actually be beneficial in PWS, it is necessary to refer back to the findings of the Stefan PWS mouse model study, in which perinatal insulin levels were essentially zero, but so were glucagon and gluconeogenesis, with the result that blood glucose levels were severely reduced, to the point that the mouse pups were severely hypotonic, had poor suck, etc. The Stefan pups died by day 7, but human infants with PWS survive due to such measures as NG tube feeding and it seems apparent that at some point insulin (and probably glucagon and gluconeogenesis) levels start to gradually increase in PWS infants, although typically to less than normal levels, with the result that glucose levels also rise (although again probably to less than optimal levels), resulting in the gradual resolution of the initially severe hypotonia and comatose-like lethargy characteristic of early infancy in PWS. However, residual hypotonia, lack of energy and lethargy remain, perhaps due to continued impaired glucose availability and metabolism.
(4) Somewhere in this process, GH treatment is begun, with parents reporting improvement in the hypotonia, lethargy, etc. soon after its start (typically within several days to a week). The rapid response to exogenous GH in infants and young children with PWS puzzled me at first because GH's effects are generally manifested over much longer periods of time and its neurological effects usually aren't particularly notable. However, GH's effect of reducing insulin sensitivity, coupled with a probable impairment in glucose availability (as disclosed by the Stefan study), does help explain the rapid response to GH treatment by PWS infants and children in terms of improving their hypotonia and mental and physical lethargy, because increasing insulin resistance results in an increase in blood glucose levels, thereby causing an increased availability of glucose for the production of energy for use by muscles, the brain, etc.
(5) When GH treatment is stopped and exogenous GH is no longer exerting its effect of increasing insulin resistance, the functional effect can be the same as an increase in insulin, resulting in a drop in blood glucose levels and, if the person already tended towards low glucose levels, the possible development of a hypoglycemic crisis. That is what appears to have happened to Stephen when GH treatment was stopped - functionally the effect seems to have been much the same as what happens when an insulin-dependent diabetic accidentally takes too much insulin, thereby precipitating a hypoglycemic crisis as glucose levels plummet.
Several important points can be inferred from the above:
(1) For those who have finished growth and GH treatment is ended, as happened with Stephen, it seems to me to be important that GH dosing is slowly tapered off, not abruptly stopped, in order to allow the body a chance to adapt (if it can) to what is functionally an increase in insulin levels and hence an increased susceptibility to the development of hypoglycemia.
(2) If GH treatment is temporarily stopped when infants and children have a respiratory infection, as is now apparently being recommended by Miller and other PWS "experts", it would seem to me to be critically important to closely monitor glucose levels and take immediate corrective action if hypoglycemia begins to develop due to a functional increase in insulin, because untreated hypoglycemia can have catastrophic impacts, including not just the psychotic delusions and aggressiveness experienced by Stephen, but also coma, brain damage and death. Indeed, I suspect that at least some of the cases of rapid deterioration and death reported in some children with PWS when they became sick (even when they weren't undergoing GH treatment) had to do with the development of a severe metabolic crisis (including hypoglycemia) when their food intake dropped due to the loss of appetite that often accompanies colds and flu.
(3) The above also reinforces for me the importance of maintaining stable glucose levels throughout the day for those with PWS, including the use of between meal and before bedtime snacks. Not only do repeated episodes of hypoglycemia have serious negative impacts on growth and neurological development, but I suspect at least some of the temper tantrums, irritability, aggressiveness, destructiveness and psychotic symptoms reported in some of those with PWS are also related to hypoglycemia. Note that the older approach to hypoglycemia was to recommend a high carbohydrate diet and there are some who still do so. However, newer research strongly suggests that small meals spaced every three hours throughout the day and composed of roughly equal amounts (as a percentage of calories) of protein, good quality fats and low glycemic carbs (mostly veggies) are both much more effective in stabilizing glucose levels and healthier over the long run.