Of the 86 eligible subjects, 41 (48%) consented to participate in the double blind study (). Subjects who refused participation primarily did not want to take any medication and did not differ by age, weight or antidepressant medication status. For subjects on an antidepressant (n=16) the median time on antidepressant prior to being randomized into the study was 11 weeks (range 1 week to 6 years). Only 3 patients had been on an antidepressant for less than 6 weeks when they started the study drug. 40 subjects started the study medication. Two subjects completed baseline measures but elected to discharge from the treatment program within 3 days of starting the study medication, for reasons unrelated to the study, and returned to out-of state locations.
CONSORT diagram showing the flow of participants
There was no significant difference in age, study medication dose, time on study medication, LOC, % IBW, BMI, REE or baseline rating scales (EDI-2 DT and BD, CAPT, BIS tasks, MASC) between the 2 groups (). Thirty of the subjects were under age 18. Mean age for those < 18 was 14.8 years.
Group differences at baseline
The first exploratory hypothesis was that subjects on risperidone would have a greater decrease in the drive for thinness, body dissatisfaction and body image distortion than subjects on placebo. This exploratory hypothesis was partially supported. From the mixed effect model, there was a significant difference for the EDI-2 DT subscale; R subjects had a significant decrease over the first 7 weeks (t-value for difference between treatments in change from baseline to 7 weeks = 3.25 with 54 d.f., effect size 0.88, p=0.002), but this difference was not sustained to week 11 (t-value=1.55 with 54 d.f., effect size 0.42, p=0.13). EDI-2 DT did not significantly change over time for PL subjects. There were no other significant differences between R and PL in change over time for measures of body dissatisfaction (change from baseline to 7 weeks, R vs. PL: EDI-2 BD, effect size 0.52, p=0.06, CAPT, effect size 0.40, p=0.16, BIS ADJ-desired, effect size 0.42, p=0.35), or body image distortion (BIS-PSE, effect size 0.34, p=0.29, BIS-DL, effect size 0.59, p=0.08, BIS ADJ- current, effect size 0.05, p=0.87). There were no significant changes between R and PL in change over time for anxiety as measured by the MASC (effect size 0.21, p=0.44). There was an unexpected finding on the EDI-2 subscale “Interpersonal Distrust” (EDI-2 ID); R subjects had a significantly greater decrease than PL subjects (t-value for difference in change from baseline = 2.22 with 54 d.f., effect size 0.60, p=0.03).
The second exploratory aim of the study was to determine if risperidone was an effective mediator in shortening the length of time it takes a subject with AN to reach 90% of IBW. Initially, we hypothesized that risperidone would shorten the length of time to reach target weight, and that this effect may be mediated by decreased body metabolism (measured by REE). Not only were there no significant differences between R and PL for change in %IBW and BMI over the period of the study (), there was no evidence that subjects on R had a decrease in body metabolism. In fact, at study end, subjects on R had a non-statistically significant mean increase in REE compared to subjects on PL (increase for R=73 (SD 185), n=10; vs. PL 12 (SD 110), n=13, effect size 0.41, p=0.34). There were no significant differences in proportions of patients reaching study end points (p=0.72, ), weeks in study by each end point or the time to reach End Point 1 (p=0.76) between R and PL. Thus, the second exploratory hypothesis was not supported.
Figure 2 A: Observed subjects percent of ideal body weight (%IBW) by week of visit (shown in gray) and estimated mean %IBW at each week across all subjects (shown in black). B: Observed subjects body mass index (BMI) by week of visit (shown in gray) and estimated (more ...)
Mean weeks in study by study end point
Orthostatic blood pressure and pulse, ECG, triglycerides, cholesterol, liver enzymes and glucose showed no significant differences between groups during the study (). Prolactin levels were significantly increased for R at week 7 and for subjects who were in the study through week 11 (R prolactin 37.7 ng/ml (SD 36.2) vs. PL prolactin 5.5 ng/ml (SD 1.6), p=0.001). Total reported physical complaints were similar between the 2 groups, R=146, PL=143. Physical complaints more frequently reported in the R group were fatigue (39 vs. 22) and dizziness (15 vs. 4). PL subjects reported more gastrointestinal complaints (53 vs. 35) and headache (8 vs. 5). Psychiatric symptoms reported more frequently in the R group were depression (2 vs.0) and suicidal ideation (4 vs.0), while the PL group reported more anxiety (6 vs. 1) and insomnia (4 vs. 0). There were no significant differences between groups on the SAS or AIMS rating scales, with 8 subjects in each group having a SAS score > 0, and 5 R subjects vs. 1 PL subject scoring > 0 on the AIMS (p=0.06). There were no Significant Adverse Event’s during the study.
Labs, Electrocardiogram (ECG) and pulse (Mean (SD))