1.1 Distinct facial and physical features

The facial and physical features seen in individuals with PWS were first described by Prader et al. in 1956.¹² These facial features include a small narrow bifrontal diameter, almond-shaped palpebral fissures, narrow nasal bridge, and thin upper lip with downturned corners of the mouth and decreased salivary secretions. Other physical features include short stature, small hands with flattened ulnar border, small feet, hypoplastic genitalia ^1-8 and hypopigmentation in relationship to first-degree relatives in those with the 15q11-q13 deletion.¹³

GH deficiency has been documented in PWS with GH treatment considered the standard of care.^{4, 6, 7, 14} Overall, GH therapy affects linear growth by increasing adult height in addition to improvement of physical activity, strength and muscle mass that can impact craniofacial features, body habitus and quality of life with improved cognition and possibly behavior in PWS.^{6, 8, 15, 16}

The aim of this study was to compare the physical and dysmorphic features in individuals with PWS in a large cohort and in relationship to genetic subtypes and GH treatment to determine if GH treatment effects are different for individuals with either deletion or UPD.

2.1 Dysmorphology evaluation

Physical and facial features, including continuous and categorical variables, were assessed. Continuous variables included physical measurements of height, weight, body mass index (BMI), head circumference (HC), craniofacial, arm span, hand, foot, and penile length. Data were collected for analysis at the initial enrollment visit per participant. For statistical purposes, the data were converted into age and gender-adjusted centiles using the WHO (World Health Organization) and CDC (Center for Disease Control) reference tables.^{18, 19} Categorical variables included esotropia, exotropia, head shape, narrow nasal bridge, flat philtrum, downturned corners of mouth, dental, genitalia, skin picking, hair texture, and pigment. The data were summarized using mean and SD for continuous variables. Participant groups were subdivided by PWS molecular genetic classes and GH use, duration, and onset, and then compared using two-group t-tests for continuous variables and chi-square tests for categorical variables. The statistical analyses were accomplished using SPSS 20 Statistics software (Armonk, NY). Statistical significance was considered at p < 0.05. This project was the focus of the Master's thesis by one of our co-authors (A.L.).²⁰

3.1 Dysmorphic and clinical features in PWS

A total of 355 PWS study participants were analyzed and comprised of 160 males (45.1%) and 195 females (59.2%). Ninety-three percent of the PWS participants were Caucasian. The average age (±SD) for the 355 PWS participants was 13 (±1) years with a range of 2 months to 62 years. The mean age at diagnosis was 3.1 ± 6.7 years with a range from birth to 48 years. Sixty-two percent of the PWS participants were diagnosed at less than 1 year of age and 26% were diagnosed greater than 3 years of age. The PWS molecular genetic classes included 217 with 15q11-q13 deletions (61%), 127 with UPD (36%) and 11 with imprinting center defects (3%). Overall, 289 participants (81.4%) had a history of GH treatment with an average age of onset of 2 (±1) years, including 137 of 160 males (85.6%) and 152 of 195 females (77.9%). In addition, 179 of 217 (82.4%) had deletions, 103 of 127 (81%) with UPD, and 7 of 11 (63.6%) individuals with imprinting center defects received GH treatment (Table 1).

In the entire PWS cohort (N = 355 participants), microcephaly (HC less than third percentile) was found in 8% of participants, flat occiput in 22%, upslanting palpebral fissures in 23%, craniosynostosis in 0.8%; strabismus in 42%; abnormal dentition in 32%; small hands in 63% and small feet in 70%, hypopigmented hair in 30%; striae in 32% and skin picking in 26%. Bilateral cryptorchidism was present in 63% and hypoplastic scrotum in 44% of males. Hypoplastic clitoris and labia minora were found in 38% and 46% of females, respectively during the initial baseline clinic visit (Table 1). Forty of the 160 males had recorded penile stretched length and 15 (38%) were considered to have a micropenis  (<5th percentile).

3.2 Comparison between the 15q11-q13 deletion and UPD 15 in the entire PWS cohort

When comparing deletion with UPD, participants with deletions were found to be heavier (mean weight percentile of 75 ± 31 vs. 62 ± 35, p = 0.002), but BMI percentile was not significantly different (mean BMI percentile of 86 ± 36 vs. 84 ± 38, p = 0.137). Those with deletions had a smaller HC with mean HC percentile of 51 ± 34 vs. 61 ± 31 (p = 0.009). However, height was not different between the two PWS molecular classes (p = 0.510). Almond-shaped eyes (68% vs. 58%; p = 0.045), a flat occiput (27 vs. 14%; p = 0.005), low-anterior hairline (26% vs. 15%; p = 0.04), and abnormal dentition (42% vs. 25%; p = 0.009) were seen at a higher incidence in the deletion group compared with UPD. Individuals with the deletion had a higher incidence of shorter fifth fingers (40% vs. 28%; p = 0.024) and nail abnormalities (26% vs. 13%; p = 0.050). The participants with the deletion were more fair-haired (40% vs. 15%; p < 0.001) than their family members which may reflect the loss of a single biallelically expressed OCA2 gene allele found in the distal 15q11-q13 region and deleted in the deletion process leading to hypopigmentation.^{13, 21} Interestingly, the UPD group had a higher incidence of strabismus (53% vs. 39%; p = 0.03), downward slanting of the fissures (13% vs. 6%; p = 0.015) and posterior angulated ears (14% vs. 6%; p = 0.019) than in the deletion group.

3.3 Comparison between GH treated and non-GH treated participants with PWS

The study participants were categorized according to their history of GH treatment (treated vs. not treated) regardless of their PWS molecular class. Table 2 tabulates the frequency of individuals taking GH in different age groups. We found that 41.7% of individuals began GH by 1 year of age and 70% were started by 4 years of age. Not surprisingly, individuals who received GH were taller (p = 0.005), had longer hands (p = 0.049), and had larger HCs (p = 0.005); however, weight was not statistically different. Mean BMI percentile was 83 ± 37 in the GH treated group vs. 85 ± 44 in the non-GH treated group, (p = 0.789). The data were standardized for gender and age, and each GH group had the same molecular class distribution (e.g., 62% deletion). We found that individuals on GH were taller (p = 0.005), had larger head size (p = 0.005), had a lower incidence of almond-shaped eyes (p = 0.005), a narrow nose (p = 0.020), abdominal pale striae (p = 0.030), skin picking of the face (p = 0.010), larger thigh circumferences (p = 0.003), longer hand length (p = 0.049), increased incidence of esotropia (p = 0.012), hypoplastic labia minora (p = 0.005) and hypoplastic clitoris (p = 0.044) (see Table 3).

Because of the wide age range of our participants, we studied the frequencies of dysmorphic features by age of initiation of GH treatment in different age groups (birth to 1 year, 1 to 4 years of age, 4 to 12 years, or 12 to 21 years), versus GH treatment initiated during adulthood (i.e., 21 years or older) (Table 4). This analysis was done to test the hypothesis that if GH treatment was initiated at a younger age, then a stronger effect may be present by ameliorating the physical and dysmorphic features associated with PWS. Individuals who had GH treatment initiated at a younger age (from birth to 1 year) in contrast to an older age group showed lower incidences of micrognathia (p = 0.039), slit-like eyes (p = 0.025), a narrow nose (p = 0.013), abdominal or central distribution of fat (p = <0.05), kyphosis (p = <0.05), and short fifth fingers (p = 0.026). Participants who started GH treatment at a younger age had fewer dysmorphic features as they received treatment for a longer duration. However, no statistically significant difference was found when the duration of treatment was compared among these age groups (one-way ANOVA; p = 0.818), but both micrognathia and slit-like eyes were statistically more common in the older age groups.

3.4 Comparison of effect of GH treatment on specific PWS molecular genetic classes

Analysis of effects of GH treatment was also undertaken for each individual molecular genetic class separately (Table 5). The duration of GH treatment was also calculated based on the reported age at initiation of GH treatment, if at first visit the participant was on GH or whether they were currently on GH, as well as at their current age, or age when discontinued. The mean age of starting GH treatment was 4 ± 0.4 years (range from birth to 49 years) with an average duration of 13 ± 0.8 years (range from birth to 53 years) with no significant differences in the two groups. Similar trends were noted in the molecular classes; individuals with the deletion versus UPD on GH treatment or non-GH treatment had a higher incidence of hypopigmented hair, or fairer hair color than their family members (p = 0.029). There were no differences in the frequency of deletion or UPD participants on GH treatment or not on GH treatment or age difference found in the two molecular class groups. Other findings that showed differences when comparing effects of GH treatment on deletion versus UPD are included in Table 5. There was a greater weight percentile (p = 0.021); hypopigmentation (p = 0.030), a higher incidence of lower anterior hair line (p = 0.046); almond-shaped eyes (p = 0.023), dental caries (p = 0.007) and kyphosis (p = 0.001) in the deletion group without GH treatment; and a higher incidence of abdominal striae (p = 0.006), hypopigmentation (p < 0.001), scoliosis (p = 0.011) and interestingly flattened occiput (p = 0.002), in the deletion group on GH treatment. Interestingly the UPD group had a higher incidence of scoliosis (p = 0.039), and broad nasal bridge (p = 0.037) in those not on GH treatment; and more downslanting fissures (p = 0.006) and posteriorly angulated ears (p = 0.002) in the group on GH treatment.