With our study, we found the following – in part surprising – results:
- In patients with AN, baseline as well as meal-induced CCK-LI release is well within the range of normal values when compared to values obtained from healthy normal-weight volunteers (see ). This is against our expectation (Hypothesis 1) and in part against current published data.
In animals (rats) elevated CCK plasma concentrations and down-regulated CCK-A receptors after prolonged food restriction of more than 20 days were found 
. After short-term fasting for only three days, CCK levels were decreased in rats in another study 
. In humans, two studies 
found higher CCK levels in anorexia nervosa (AN) patients compared to controls, whereas three others 
did not. More recent studies found a decrease of CCK in AN 
but an increase in undernourished women 
. In obese humans no changes in CCK release were found after 10 weeks of dieting 
. Binge-eating, however may cause an adaptive lowering of CCK release to large amounts of food, which can be observed in the same way in normal weight binge eaters 
. Discrepancies especially in humans may in part be explained by the fact that all these studies did not assess sulfated bioactive CCK but rather assays with a considerable cross-reaction with gastrin, and in light of the ten-fold higher plasma concentration of gastrin the evidence for a role of CCK in AN remains unclear.
Our data indicate that CCK-LI release in long-term and severe anorexia nervosa (irrespective of the subtype) is well adapted to the chronic malnutrition and responds normally when stimulated with a standard (test) meal of low volume and caloric content (250 kcal).
- A moderate (2 kg) increase in weight achieved during controlled inpatient treatment for a few days to weeks results in a significant increase of both baseline as well as meal-induced CCK-LI release that exceeds the response of healthy volunteers on average by nearly 100%. This has not been observed previously and may in part be explained the novel measure of biologically active CCK that was applied here for the first time.
According to common clinical experience, such a strong response to moderate food intake may be able to limit and counteract the therapeutic attempts to further gain weight. It may explain the increasing difficulties that anorectic patients perceive in the first weeks of weight gain to maintain a sufficient food intake. This phenomenon may be called a “hunger trap”: Once the CCK system has adapted to low food intake, it self-limits weight gain by increasing satiety even with moderate food intake. Only with further weight gain the CCK release returns to the initial and normal state (Hypothesis 2). High postprandial CCK levels were observed in girls with AN 
and were made responsible for the aggravation of the course of the disease via intensifying nausea and vomiting.
- We found a consistent but small difference in CCK-LI release between restricting and bulimic AN patients, with bulimic patients exhibiting overall lower values (see ). This supports our Hypothesis 3, and the missed significance may in part be due to small sample sizes in the restricting (n=10) and bulimic (n=13) subgroup.
Lower CCK-LI levels in bulimic as compared to restricting AN patients would explain that bulimic AN patients feel less satiety after eating and thus are vulnerable to loose control during binges. Binge-eating may cause an adaptation of CCK release to large amounts of food, which can be observed in the same way in normal weight binge eaters 
. However, CCK-LI release with a meal was closer to the normal ranges than in restricting AN patients in our study and thus questions this explanation.
Differences between restricting and bulimic food intake on CCK release have not previously been described in other experimental studies, but it needs to be kept in mind that there is however no adequate animal model for bulimic eating behavior.
As to expect, bulimic patients had a higher frequency of vomiting at admission but the difference vanished during the course of therapy and therefore cannot explain the difference between restricting and bulimic patients throughout the therapy. It has been reported 
that percent body fat is higher in bulimic AN than in restricting AN which we confirm here (see ). Bulimic AN patients also had higher body weight at admission. Whether this limits the CCK-LI release – despite similar body weight and BMI (see ) – in comparison to restricting AN patients has to remain open but warrants further studies.
- We expected the CCK-LI release before and after a meal to be associated with weight, BMI, and weight and BMI gain during therapy (Hypothesis 4). While human data are not available on this issue, animal data are conflicting: Rats lacking the CCK-A receptor gene showed disordered eating and increase in body weight , while knock-out mice lacking the same receptor grow to normal weight , .
We found a supporting but moderate association (explaining 25% of variance) (see ) between the starting weight and the increase in pre-meal CCK-LI between the initial measurement (T0) and after a moderate weight gain (T1) but no further associations of the initial CCK-LI levels to weight and BMI changes. It is thus not likely that CCK is involved in the normal process of weight regulation but further studies are needed.
5. CCK release is associated with symptoms in normal weight gastrointestinal disorders and induces motility disturbances associated with complaints 
. On the one hand, gastrointestinal symptoms are frequent in AN (see ) 
; on the other hand, patients with functional bowel disorders frequently exhibit signs of eating disorders 
. We therefore expected associations between gastrointestinal symptoms and CCK-LI release, specifically at the initial measurement point (T0) (Hypothesis 5).
Both the gastrointestinal symptom score (GIS) as well as the subjective impairment by these symptoms (GIM) were moderately associated with CCK levels at T0, and with both pre as well as post-meal levels (). With increasing body weight, normalizing CCK-LI response, and decreasing GIS and GIM scores, these associations were no longer significant at T2. This may indicate that the association between CCK response and gastrointestinal symptoms may be strong in disordered eating and/or digestion, but may not be related to each other when both return to normal functioning. It may also indicate that AN patients who do not adapt their CCK level to low food intake (see above) represent a – potentially genetically determined 
- subgroup of AN patients that are at risk to not respond to weight gain 
. This may also link CCK to further and specifically neuropsychological functions in AN 
. Further research is needed to substantiate this hypothesis.
Anorectic patients, specifically of restricting-type AN show well-adapted CCK-release pattern with stable but abnormal eating behavior and under-nutrition. With the initiation of weight gain, an initial exaggerated CCK-response occurs that - clinically - may terminate further weight gain by inducing premature satiety (“hunger trap”). Only if therapy can overcome this intrinsic handicap, further weight gain will lead to normalization of CCK levels again. Patients who do not adapt to the normalization of CCK are at risk to experience high levels of gastrointestinal symptoms that do not improve with weight therapy.