Glucose Homeostasis and Energy Balance in Children With Pseudohypoparathyroidism

Abstract

Context

Pseudohypoparathyroidism (PHP) is a rare genetic disorder characterized by early-onset obesity and multihormone resistance. To treat abnormal weight gain and prevent complications such as diabetes, we must understand energy balance and glucose homeostasis in PHP types 1A and 1B.

Objective

The aim of this study was to evaluate food intake, energy expenditure, and glucose homeostasis in children with PHP.

Design

Assessments included resting energy expenditure (REE), physical activity, food intake, sucrose preference, questionnaires, endocrine status, and auxological status. All patients underwent an oral glucose tolerance test (OGTT).

Setting

Vanderbilt University Medical Center.

Patients

We assessed 16 children with PHP1A, three with PHP1B, and 15 healthy controls.

Main Outcome Measures

Food intake during an ad lib buffet meal and glucose at five time points during OGTT.

Results

PHP1A and control groups were well matched. Participants with PHP1A had significantly lower REE without concomitant change in food intake or physical activity. At baseline, participants with PHP1A had significantly lower fasting glucose and insulin resistance. During OGTT, participants with PHP1A had significantly delayed peak glucose and a slower rate of glucose decline despite similar oral glucose insulin sensitivity. Participants with PHP1A had 0.46% lower HbA1c levels than controls from a clinic database after adjustment for OGTT 2-hour glucose. The PHP1B group was similar to the PHP1A group.

Conclusions

In contrast to other monogenic obesity syndromes, our results support reduced energy expenditure, not severe hyperphagia, as the primary cause of abnormal weight gain in PHP. Patients with PHP are at high risk for dysglycemia without reduced insulin sensitivity.

Patients with PHP1A have reduced energy expenditure without concomitant change in food intake or physical activity, resulting in early-onset obesity. They may also have increased risk of dysglycemia.

Pseudohypoparathyroidism (PHP) type 1A is a rare genetic disorder caused by impaired stimulatory G-protein (Gsα) signaling due to heterozygous mutations in the gene GNAS. Numerous G protein–coupled receptors (GPCRs) are known to signal through Gsα. In certain tissues (e.g., hypothalamus, thyroid, kidney), GNAS is paternally imprinted (1). PHP type 1A (PHP1A) occurs when a GNAS mutation is inherited on the preferentially expressed maternal allele, resulting in a further decrease in Gsα expression in imprinted tissues. A less severe form of the disease, pseudopseudohypoparathyroidism (PPHP), occurs when a GNAS mutation is inherited on the paternal allele. Clinically, PHP1A is characterized by Albright hereditary osteodystrophy (AHO; adult short stature, brachydactyly, subcutaneous ossifications, round face), multihormone resistance, cognitive impairment, and early-onset obesity (2–5). In contrast, PPHP is characterized by AHO without multihormone resistance (6). More recently, imprinting defects at the GNAS locus (PHP1B) were found to account for ~25% of patients with the PHP1A phenotype (7, 8).

An increasingly recognized feature of PHP1A and PHP1B, but not PPHP, is early-onset obesity (4, 9). In one study, >89% of the children with PHP1A were classified as obese, compared with 17% of children in the general population (4). Children with PHP1A have a reduction in resting energy expenditure (REE), which contributes to the rapid weight gain seen in early childhood (10, 11). This defect in energy balance is thought to be caused by abnormal signaling through the melanocortin-4 receptor (MC4R), a hypothalamic GPCR and critical regulator of energy homeostasis. There is imprinting of Gnas in the mouse hypothalamus, and a brain-specific Gsα exon 1 knockout mouse model recapitulates the PHP1A obesity phenotype (12). Abnormal function of the MC4R is known to cause early-onset obesity in humans and the animal model (13, 14). In humans, MC4R haploinsufficiency is associated with hyperphagia (13); however, patients with PHP1A are not clearly hyperphagic (11, 15).

In part because of early-onset obesity, PHP1A is also associated with a higher risk of type 2 diabetes (T2D). In a murine model, heterozygous Gnas mutations are associated with obesity and glucose intolerance (16). Adults with PHP1A have higher rates of T2D and lower insulin sensitivity compared with similarly obese controls (17). However, children with PHP1A do not have higher insulin resistance, based on fasting glucose and insulin measurements, and they trend toward a lower HbA1c compared with obese controls (10). In this prospective, cross-sectional study, we investigated energy balance in PHP1A by objectively measuring food intake and activity levels, along with glucose homeostasis, in children with PHP1A. We hypothesized that along with lower REE, children with PHP1A have mild hyperphagia and increased sucrose preference, leading to early-onset obesity and reduced insulin sensitivity.

Materials and Methods

Participants

Patients were recruited from the Monroe Carell Jr. Children’s Hospital at Vanderbilt and throughout the United States and Canada via online advertisements. Enrollment was open to patients 6 to 18 years old in one of the following groups: children with a diagnosis of PHP and multihormone resistance and healthy controls. Controls were matched on sex, race, age (±2 years), and body mass index (BMI) (±2 kg/m² at last clinic visit). A diagnosis of PHP1A or PHP1B was confirmed based on clinical phenotype and genetic testing, as previously reported (5). Exclusion criteria included initiation of a new weight loss program in the past 3 months, treatment with an appetite-altering drug, or diagnosis of diabetes mellitus. Additional exclusion criteria for the sibling and control groups included obesity due to a genetic syndrome, untreated endocrinopathy, parent-reported diagnosis of autism or learning disorder, or other medical conditions.

All study visits were conducted at the Vanderbilt Clinical Research Center (CRC, Nashville, Tennessee) between December 2014 and March 2017. The study was approved by the Vanderbilt Institutional Review Board, and parental consent or age-appropriate assent was obtained before study enrollment.

In a post hoc analysis, we compared patients with PHP1A to a deidentified, institutional review board–approved dataset from the Vanderbilt Pediatric Prediabetes Clinic. Patients in this clinic have a referral HbA1c ≥5.7% and clinical concern for prediabetes or T2D. As part of clinical evaluation, all patients in this data set had a repeat HbA1c and 75-g oral glucose tolerance test (OGTT).

Experimental procedures

Participants were asked to maintain their usual diet and abstain from caffeine and strenuous exercise in the 3 days before testing. They arrived at the CRC after ≥8 hours of fasting. A board-certified pediatric endocrinologist performed a physical exam, including Tanner stages. Patients were defined as pubertal if they had reached Tanner stage ≥2 for breast development (girls) or testicle size (boys).

Genetic testing

DNA samples were collected from participants with PHP if they did not have a known genetic diagnosis. GNAS exons 2 to 13 (NM001077488.3) were sequenced by GENEWIZ (South Plainfield, NJ). If a pathogenic mutation was not detected, we next evaluated the DNA samples for mutations in GNAS exon 1, the 3-kb STX16 deletion that often causes autosomal dominant PHP1B, for other deletions within STX16 or GNAS, and for GNAS methylation changes (18).

Vital signs

Heart rate and blood pressure were measured on the second day with an electronic sphygmomanometer and an appropriately sized cuff while the patient was resting in bed.

Glucose metabolism

Fasting blood samples were obtained, and then participants were given 1.75 g/kg (maximum 75 g) glucose to drink in <5 minutes. Glucose and insulin levels were measured at 0 minutes, 30 minutes, 60 minutes, 90 minutes, and 120 minutes. For baseline modeling we used the homeostatic model assessment of insulin resistance (HOMA-IR) and β-cell function (HOMA-β) (19). We calculated the homeostatic model assessment of adiponectin (HOMA-AD), another model of insulin resistance (20). We also calculated the oral glucose insulin sensitivity (OGIS) (21) and early insulin response (ΔInsulin_0–30min/ΔGlucose_0–30min) (22).

Endocrine analyses

A fasting blood sample was collected for lipid profile, TSH, free T4, HbA1c, β-human chorionic gonadotropin (in pubertal girls only), calcium, and PTH by standard commercial assays in a Clinical Laboratory Improvement Amendments certified clinical laboratory. The Vanderbilt Hormone Assay Core measured leptin by radioimmunoassay and adiponectin and resistin by Luminex assay.

REE

Participants spent the night in the CRC and were asked to be in bed by 2200 hours. The participant was awoken between 0600 and 0700 hours by study personnel. REE was measured by indirect calorimetry with a metabolic cart while the patient was awake but resting in a recumbent position. The first 10 minutes of data were discarded, and ~30 minutes of data were used to determine REE with the Weir equation (23).

Food preference and nutritional intake

At the completion of the OGTT, participants received a standard lunch meal. They were allowed to drink water ad lib. Between 5 and 6 hours after lunch, participants were offered a 4000-calorie dinner buffet prepared in the CRC Metabolic Kitchen Core. They were instructed to eat until comfortably full, and parents were asked to leave the room. The buffet was removed after 60 minutes. A small, standardized bedtime snack was provided. In the morning, the participants were offered a 4000-calorie breakfast buffet under the same protocol. Likert scales were administered before and after the buffet meals to assess hunger and satiety. Calorie and macronutrient consumption were determined by entering the weigh-back meal data into Nutrition Data System for Research software (Minneapolis, MN).

Parents completed several questionnaires. The Hyperphagia Questionnaire contains 11 questions that assess symptoms of hyperphagia in one of three categories (hyperphagic drive, behavior, and severity), as rated on a 5-point scale (1 = not a problem to 5 = severe or frequent problem) (24). The total hyperphagia score has a minimum of 11 points and a maximum of 55. The 35-item Child Eating Behaviour Questionnaire assesses positive and negative eating behaviors, rated on a scale from 1 (never) to 5 (always) (25). The 24-item Child Feeding Questionnaire assesses parental beliefs, attitudes, and practices regarding child feeding in seven areas (minimum score of 1 point and maximum score of 5 points) (26). Because of the age range of participants, “perceived child weight” was assessed through second grade, and questions 12 (your child third through fifth grade) and 13 (your child from sixth through eighth grade) were eliminated.

Participants completed the 152 question Youth/Adolescent Food Frequency Questionnaire with the assistance of their parents (27). The questionnaires were coded and scored by the Harvard T.H. Chan School of Public Health Nutrition Department (Boston, MA). Participants also reported their preference for different beverages on 5-point Likert scales (water, milk, chocolate milk, juice, fruit-flavored drinks, sweet tea or soda, diet soda, and sports drinks).

Sucrose preference testing was conducted 2 hours after the breakfast buffet. Five solutions of varying sucrose concentration were presented (3%, 6%, 12%, 24%, and 36%) in pairs, beginning with the 6% and 24% solutions (28). Participants were asked to taste without swallowing and report which they preferred. Each subsequent pair used the previously preferred concentration and an adjacent concentration. This continued until the subject chose the same concentration two consecutive times. Next, participants were given three samples (3%, 12%, and 36%) in random order and asked to rank them in order of sweetness.

Physical activity

After discharge from the CRC, physical activity level was measured over a 2-week period with the ActiGraph GT3X+ (ActiGraph, LLC, Pensacola, FL), a small triaxial accelerometer (19 g, 46 mm × 33 mm × 15 mm) worn on the waist. Data were collected continuously at a sample rate of 40 Hz and analyzed in 15-second epochs with the manufacturer-provided software (ActiLife version 6.4 software, Pensacola, FL). The minimum valid wear time criterion was one weekend and three weekdays of ≥6 hours of activity. Validated threshold values were used to determine time spent in sedentary, light, moderate, and vigorous activity (29).

Body composition

Weight was measured in the fasting state with a digital scale, and height was measured with a wall-mounted stadiometer, with the patient lightly clothed and without shoes. Height, weight, and BMI z scores were calculated with sex- and age-specific Centers for Disease Control and Prevention growth charts. Dual-energy x-ray absorptiometry (DXA) was performed with pediatric software (Lunar Prodigy, GE Medical Systems, Madison, WI) to determine estimates of fat mass, fat-free mass (FFM), percentage body fat, and bone mineral density. Waist circumference was measured at the level of the umbilicus with the tape measure parallel to the floor. Three measurements were taken and the average recorded.

Data collection

Study data were collected and managed with REDCap electronic data capture tools hosted at Vanderbilt University (30). REDCap is a secure, Web-based application designed to support data capture for research studies.

Statistics

Continuous variables were summarized as mean ± SD, and categorical variables were summarized as percentages. We tested for unadjusted differences between PHP1A and control subjects via either the nonparametric Mann-Whitney U test (continuous variables) or Fisher’s exact test (categorical variables). We present the results of all unadjusted tests performed, regardless of significance, and thus do not make any formal adjustment for multiple comparisons. Repeated-measures ANOVA was used to test for group differences in the response to the OGTT over time. We used multiple linear regression to determine whether the association between HbA1c and 2-hour glucose in patients with PHP1A differed from the association between HbA1c and glucose in controls from the Vanderbilt Pediatric Clinic. The regression model included an intercept, glucose slope, indicator for PHP1A status, and interaction between the indicator and glucose slope to test for difference in the intercept or slope between control and PHP1A subjects. Analysis was conducted with SPSS version 25 and R version 3.4.1.

Results

We enrolled 19 participants with PHP with multihormone resistance (8 to 18 years old) and 15 controls (8 to 16 years old). The genetic testing results from the PHP group have been published previously (5). In brief, the 16-patient PHP1A group includes 14 patients with an identified GNAS inactivating mutation and two patients with multihormone resistance, brachydactyly and subcutaneous ossifications but without a genetic diagnosis (one did not provide a DNA sample for analysis, one did not have an identifiable mutation, deletion, or methylation changes in GNAS). Three patients (all siblings) were found to have PHP1B on genetic testing. The single kindred PHP1B group is presented separately at the end of the result section.

PHP1A group vs control group

Baseline characteristics

All enrolled patients were white. Baseline characteristics by group are presented in Table 1. There were no differences between the two groups except that height was significantly lower in the PHP1A group, as expected (z score −0.50 ± 1.17 vs 0.50 ± 1.35, P = 0.03). In pubertal patients, the height difference was more pronounced (z score −0.79 ± 1.10 vs 0.49 ± 1.2, P < 0.01). Only one patient in the PHP1A group was postpubertal. The resting heart rate was lower in the PHP1A group than controls, but this difference did not reach statistical significance (72 ± 13 vs 81 ± 9, P = 0.09).

Table 1.

Baseline Characteristics of Participants

	PHP1A (n = 16)	Controls (n = 15)	P	PHP1B (n = 3)
Age, y	12.6 ± 2.6	12.5 ± 2.3	0.97	13.0 ± 3.0
Sex, female	63	73	0.70	33
Height z score	−0.50 ± 1.17	0.50 ± 1.35	0.03	1.13 ± 0.95
Weight z score	2.07 ± 0.70	2.24 ± 0.70	0.45	1.64 ± 1.21
BMI z score	2.27 ± 0.47	2.22 ± 0.35	0.55	1.38 ± 1.27
BMI, kg/m2	31.9 ± 8.5	31.7 ± 6.5	0.80	25.9 ± 9.0
Waist/height ratio	0.67 ± 0.09	0.66 ± 0.07	0.98	0.54 ± 0.10
Body fat, %	45.2 ± 5.9	46.7 ± 4.5	0.50	41.9 ± 9.5
Bone density z score	1.31 ± 1.03	1.41 ± 0.95	0.70	0.77 ± 1.07
Heart rate, beats/min	72 ± 13	81 ± 9	0.09	99 ± 11
Systolic blood pressure, mm Hg	118 ± 9	120 ± 17	0.83	117 ± 14
Diastolic blood pressure, mm Hg	65 ± 7	64 ± 8	0.47	76 ± 11
Pubertal, %	63	80	0.43	67

Continuous variables presented as mean ± SD, P value for patients with PHP1A and controls by Mann-Whitney U test. Dichotomous variables presented as percentages, P value by Fisher’s exact test. Boldface indicates significant values. Patients at Tanner stage ≥2 categorized as pubertal. Body fat percentage and bone density z score by DXA.

One control patient was adopted, and family history was not available. In the PHP1A group, 31% had a history of diabetes in the mother vs 14% of the control group (P = 0.40). There was a history of diabetes in the father in 13% of the PHP1A group and 14% of the control group (P = 1.0). The majority of the mothers were obese (69% of PHP1A vs 57% of controls, P = 0.70). Five mothers had PPHP; in this group 4 out of 5 (80%) were obese and 2 out of 5 (40%) had diabetes.

Endocrine analysis

We were unable to obtain venous access in two control patients. Baseline laboratory data are presented in Table 2. PTH levels were elevated in the PHP1A group (321 ± 357 pg/mL, range 39 to 1358 pg/mL), and seven patients (44%) had levels more than two times the upper limit of normal. All patients with PHP1A had normal calcium levels (range 8.5 to 9.9 mg/dL). All patients had normal TSH levels.

Table 2.

Baseline Laboratory Characteristics of Participants

	PHP1A (n = 16)	Controls (n = 13)	P	PHP1B (n = 3)
PTH, pg/mL	321 ± 357	45 ± 15 (n = 11)	<0.001	85 ± 65
Calcium, mg/dL	9.3 ± 0.4	9.5 ± 0.4 (n = 12)	0.42	9.6 ± 0.3
TSH, µU/mL	2.897 ± 1.139	2.602 ± 0.838 (n = 12)	0.42	3.168 ± 1.263
Free T4, ng/dL	1.10 ± 0.23	0.97 ± 0.10 (n = 12)	0.10	0.96 ± 0.25
Total cholesterol, mg/dL	172 ± 38	147 ± 32	0.06	176 ± 19
Low-density lipoprotein, mg/dL	112 ± 33	93 ± 3	0.17	109 ± 22
High-density lipoprotein, mg/dL	38 ± 8	38 ± 6	0.62	48 ± 7
Triglycerides, mg/dL	111 ± 47	104 ± 50	0.45	94 ± 21
Fasting glucose, mg/dL	78 ± 5	87 ± 8	0.001	79 ± 7
Fasting insulin, µU/mL	13.8 ± 5.6	17.8 ± 7.1	0.17	15.9 ± 14.3
HbA1c, %	5.17 ± 0.37	5.29 ± 0.16	0.09	5.23 ± 0.21
Leptin, ng/mL	43.86 ± 25.56	53.90 ± 23.49	0.18	70.9 ± 63.9
Adiponectin, µg/mL	14.53 ± 17.62	15.52 ± 8.91	0.42	41.57 ± 27.92
Resistin, ng/mL	28.65 ± 8.72	30.75 ± 7.66	0.11	34.18 ± 6.40
HOMA-IR	2.68 ± 1.10	3.86 ± 1.58	0.05	3.08 ± 2.74
HOMA-β	4.26 ± 4.63	2.79 ± 1.17	0.50	3.91 ± 3.38
HOMA-AD	129 ± 104	128 ± 78	0.95	87 ± 132

Results presented as mean ± SD, P for patients with PHP1A and controls by Mann-Whitney U test. Boldface indicates significant values.

Glucose homeostasis

The PHP1A group had lower fasting glucose levels compared with controls (78 ± 5 mg/dL vs 87 ± 8 mg/dL, P < 0.01). All patients with PHP1A had normal fasting glucose (range 66 to 86 mg/dL), and one patient in the control group had impaired fasting glucose (range 76 to 100 mg/dL). The PHP1A group had lower baseline insulin resistance by HOMA-IR (2.68 ± 1.10 vs 3.86 ± 1.58, P = 0.05). There was no difference in HOMA-AD or HOMA-β (Table 2).

We were unable to obtain 90-minute blood samples in two patients with PHP1A because of difficulty maintaining venous access. There was no evidence that OGIS was significantly different between patients with PHP1A (427 ± 99 mL/min/m², n = 14) and controls (377 ± 59 mL/min/m², n = 13, P = 0.17). The early insulin response was lower in the PHP1A group, although it did not reach statistical significance (2.3 ± 1.5 vs 3.4 ± 1.6, P = 0.09). The PHP1A group had prolonged hyperglycemia after an oral glucose load compared with controls (Fig. 1). From 30 minutes to 120 minutes, glucose levels in the PHP1A group decreased at a slower rate than in controls (slope 11.1 mg/dL per 30 minutes lower in the PHP1A group than controls, 95% CI −17.7 to −4.8 mg/dL per 30 minutes). The PHP1A group glucose peaked at 60 minutes (135 ± 35 mg/dL) and remained elevated at 120 minutes (127 ± 37 mg/dL). Four patients in the PHP1A group had glucose intolerance (2-hour glucose >140 mg/dL), and one had a 2-hour glucose >200 mg/dL vs none in the control group. The four patients with PHP1A with impaired glucose tolerance were more obese (BMI 35.4 ± 10.7 kg/m² vs 30.7 ± 7.9 kg/m², P = 0.21) with a higher body fat percentage (48.4% ± 5.5% vs 44.1% ± 5.8%, P = 0.17) and higher fasting insulin levels than the 12 patients with PHP1A with normal glucose tolerance (19.2 ± 5.8 µU/mL vs 12.0 ± 4.4 µU/mL, P = 0.06), although none of these differences reached statistical significance. The impaired glucose tolerance and normal glucose tolerance groups were similar in age (13.5 ± 2.1 years vs 12.5 ± 2.5 years, P = 0.52), puberty status (75% vs 58%, P = 1.0), and fasting glucose (78.8 ± 2.2 mg/dL vs 78.3 ± 6.0 mg/dL, P = 0.86).

Figure 1.

Results of a 75-g OGTT in children 8 to 18 y old. Circles and solid line: PHP1A. Squares and dashed line: matched controls. Results presented as mean ± SD. *P < 0.05 by repeated-measures ANOVA.

Despite higher postprandial glucose levels, the PHP1A group had a lower mean HbA1c than controls, although this difference did not reach statistical significance (5.17% ± 0.37% vs 5.29% ± 0.16%, P = 0.09). The three patients with PHP1A with a 2-hour glucose 140 to 199 mg/dL had HbA1c levels of 5.2% to 5.3%. The patient with PHP1A with a 2-hour glucose >200 mg/dL had an HbA1c of 6.1%. In a post hoc analysis, we compared the 2-hour glucose and HbA1c levels in the PHP1A group with a deidentified database of 189 patients who underwent OGTT as part of clinical care in the Vanderbilt Prediabetes Clinic. These clinic patients were similar in age (12.5 ± 2.6 years), sex (55.6% female), and BMI (33.9 ± 8.6 kg/m²). The clinic patients were more diverse, with 53% white and 32% black (remaining 15% other or unknown). The PHP1A group had an HbA1c 0.46% lower than that of controls after adjustment for the 2-hour blood glucose on OGTT (95% CI −0.28 to −0.65, P < 0.001, Fig. 2).

Figure 2.

Comparison of HbA1c and 2-h blood glucose after a 75-g OGTT. Circles and solid line: PHP1A. Squares and dashed line: controls from the Vanderbilt Prediabetes Clinic. Controlling for the 2-h blood glucose, on average the PHP1A group had an HbA1c 0.46 points lower than controls (95% CI −0.28 to −0.65, P < 0.001), and there was no difference in the slopes of the regression lines (P = 0.91).

Energy balance

As previously reported, the PHP1A group had lower REE than controls (1435 ± 288 kcal/day vs 1808 ± 391 kcal/day, P < 0.001). This finding remained true after adjustment for FFM (−191 kcal/day, 95% CI −304 to −77, P < 0.01 by linear regression). Three patients with PHP1A had FFM and REE measured ~6 years earlier in a previous study (10). Their longitudinal data are presented in Fig. 3. All 16 patients with PHP1A and 12 obese controls returned the ActiGraphs with adequate wear time. There was no difference in activity level between the two groups (data not shown). On average, both groups met the Centers for Disease Control and Prevention recommendation of ≥60 minutes of moderate to vigorous intensity exercise daily (PHP1A 83 ±32 min/day vs controls 82 ± 24 min/day, P = 0.73).

Figure 3.

REE is lower in patients with PHP than in controls and is persistent over time. Open squares, triangles, and diamonds each represent a patient with PHP1A measured at two time points [during this study and 6 y earlier (10)]. Solid black circles and line represent the other 12 patients with PHP1A in this study. Open circles represent three patients with PHP1B. Solid gray diamonds and line represent control patients from both studies.

Food intake

Results of the parent-reported questionnaires are summarized in Table 3. Parents reported that the PHP1A group had an “increased interest in food” at an earlier age than controls (4.3 ± 2.8 years vs 8.4 ± 3.8 years, P < 0.01). Parents of the PHP1A group were also more likely to report that their children were overweight at a young age on the Child Feeding Questionnaire (Table 3). The PHP1A group had higher Hyperphagia Questionnaire scores than controls, although this difference did not reach statistical significance (total score 24 ± 7 vs 20 ± 5, P = 0.06). The PHP1A group consumed more food during the buffet meals than controls (normalized to measured REE), although the difference was significant only at breakfast (breakfast buffet, 99% ± 49% vs 63% ± 23% REE, P = 0.01; dinner buffet, 93% ± 22% vs 79% ± 30% REE, P = 0.25). Total caloric intake and meal composition during the buffet meals and the food frequency questionnaire are detailed in Table 4.

Table 3.

Eating Questionnaires, Completed by Participant’s Parent

	PHP1A	Controls	P	PHP1B
Hyperphagia Questionnaire	(n = 16)	(n = 14)		(n = 3)
Total score	24.4 ± 7.2	19.9 ± 5.4	0.06	16.7 ± 4.9
Hyperphagic behavior	9.3 ± 2.8	8.4 ± 2.0	0.36	6.3 ± 1.2
Hyperphagic drive	10.8 ± 3.5	8.5 ± 3.0	0.06	6.3 ± 1.2
Hyperphagic severity	4.3 ± 1.7	3.1 ± 1.4	0.06	4 ± 1.7
Child Eating Behavior Questionnaire	(n = 16)	(n = 14)		(n = 3)
Positive eating behaviors	3.3 ± 0.8 (n = 15)	3.3 ± 0.6	1.0	2.6 ± 1.0 (n = 2)
Enjoyment of food	4.1 ± 0.7	3.9 ± 0.8	0.53	3.9 ± 1.4
Desire to drink	3.1 ± 1.3	3.2 ± 1.0	0.89	2.6 ± 0.7
Emotional overeating	2.7 ± 0.7 (n = 14)	2.8 ± 0.8	0.95	2.0 ± 0.4 (n = 2)
Food responsiveness	3.4 ± 1.1	3.2 ± 0.7	0.45	3.2 ± 1.4
Negative eating behaviors	2.8 ± 0.4	2.6 ± 0.4	0.26	2.6 ± 0.7
Emotional undereating	2.9 ± 0.5	2.9 ± 0.7	0.79	2.4 ± 0.1
Satiety responsiveness	2.2 ± 0.6	2.4 ± 0.7	0.50	2.5 ± 1.0
Fussiness	3.2 ± 0.8	2.6 ± 0.7	0.05	2.8 ± 1.3
Child Feeding Questionnaire	(n = 16)	(n = 15)		(n = 3)
Perceived responsibility	3.9 ± 0.8	3.8 ± 0.8 (n = 14)	0.38	2.6 ± 0.7
Perceived parent weight	3.5 ± 0.6 (n = 15)	3.4 ± 0.6 (n = 13)	0.65	3.3 ± 0
Perceived child weight	4.4 ± 0.6	3.1 ± 0.5 (n = 14)	<0.001	3.6 ± 0.1
First year of life	4.5 ± 0.7	2.6 ± 0.9	<0.001	4.3 ± 0.6
Toddler	4.5 ± 0.7	3.1 ± 0.7	<0.001	4.0 ± 0.0
Preschooler	4.4 ± 0.6	3.4 ± 0.9	0.004	3.0 ± 0.0
Kindergarten–2nd grade	4.3 ± 0.7	3.7 ± 0.7	0.07	3.0 ± 0.0
Concern about child weight	4.0 ± 1.0	3.8 ± 1.4	0.55	3.0 ± 1.8
Restriction	3.7 ± 0.8	3.5 ± 1.3	0.92	3.4 ± 1.6
Pressure to eat	1.7 ± 0.8	2.0 ± 0.9	0.23	2.3 ± 0.8
Monitoring	4.5 ± 0.5	3.6 ± 0.8	<0.001	2.8 ± 0.7

Variables presented as mean ± SD, P by Mann-Whitney U test. Subscales are in italics. Boldface indicates significant values.

Table 4.

Food Intake as Measured by Ad Lib Buffet Meals and Questionnaires

	PHP1A	Controls	P	PHP1B
Breakfast buffet meal	(n = 13)	(n = 14)		(n = 3)
Total intake, kcal	1412 ± 919	1103 ± 442	0.55	1039 ± 360
Intake, kcal/kg FFM	45.4 ± 29.3	31.8 ± 17.2	0.19	30.2 ± 13.2
Fat, %	35.5 ± 5.1	34.7 ± 6.3	0.91	35.5 ± 5.1
Carbohydrate, %	49.0 ± 8.3	50.8 ± 9.5	0.76	55.9 ± 11.9
Protein, %	16.2 ± 3.2	15.2 ± 3.1	0.40	14.7 ± 5.4
REE, %	99 ± 49	63 ± 23	0.01	70.9 ± 25.7
Dinner buffet meal	(n = 15)	(n = 14)
Total intake, kcal	1344 ± 437	1376 ± 594	0.98	1370 ± 257
Intake, kcal/kg FFM	41.0 ± 16.9	39.7 ± 18.9	0.89	40.2 ± 12.8
Fat, %	32.8 ± 4.6	30.5 ± 3.3	0.19	30.7 ± 1.8
Carbohydrate, %	48.1 ± 6.4	53.4 ± 5.5	0.02	56.4 ± 1.0
Protein, %	19.5 ± 2.9	17.3 ± 2.6	0.08	14.5 ± 2.1
REE, %	93 ± 22	79 ± 30	0.25	71 ± 26
Youth/Adolescent Food Frequency Questionnaire	(n = 15)	(n = 14)		(n = 3)
Daily intake, kcal	1749 ± 772	2134 ± 773	0.17	1891 ± 227
Fat, %	34.3 ± 5.0	33.3 ± 4.4	0.65	32.0 ± 0.7
Carbohydrate, %	50.6 ± 6.7	52.3 ± 8.6	0.95	57.7 ± 4.1
Protein, %	16.9 ± 3.2	16.6 ± 4.3	0.81	14.4 ± 1.8
REE, %	132 ± 74	127 ± 56	0.88	131 ± 227
Daily calcium, mg	1230 ± 711	1036 ± 445	0.51	1271 ± 208

Variables presented as mean ± SD, P by Mann-Whitney U test. REE was measured by indirect calorimetry. FFM was measured by DXA. Boldface indicates significant values.

Similar to our previous findings (15), patients with PHP1A clustered in two distinct groups; one with normal hunger and one with marked hyperphagia [Hyperphagia Questionnaire total score 18.9 ± 3.8 (n = 9) vs 31.4 ± 2.5 (n = 7), P < 0.001]. In a post hoc analysis, age, sex, and BMI were similar between these two PHP1A groups. There was no statistically significant difference in food intake during the buffet meals, possibly because of the small sample size [breakfast buffet, 110% ± 69% (n = 6) vs 89% ± 27% REE (n = 7), P = 1.00; dinner buffet, 97% ± 26% (n = 7) vs 90% ± 20% REE (n = 8), P = 0.61].

Sucrose preference

The PHP1A group preferred higher concentrations of sucrose than controls (14.4% ± 9.2% vs 7.7% ± 8.9%, P = 0.04, both groups n = 14). The PHP1A group also had difficulty ranking solutions by order of sweetness (71% correct in PHP1A group vs 100% correct in controls, P = 0.10). There was no difference in patient-reported preference for sugar-sweetened beverages on Likert scales.

PHP1B group

The baseline characteristics of the three siblings with PHP1B are presented in Table 1. The siblings ranged from 10 to 16 years old, and 2 of 3 had obesity. All patients had PTH and TSH resistance; baseline laboratory data are presented in Table 2. Similar to the PHP1A group, the PHP1B group had normal fasting glucose (72 to 86 mg/dL) but a delayed glucose peak at 60 minutes (138 ± 32 mg/dL) that remained elevated at 120 minutes (122 ± 28 mg/dL). One patient with PHP1B had glucose intolerance (33%). This patient was male, pubertal (Tanner stage 3), and not obese (BMI z score −0.08).

The PHP1B group also had a decreased REE (1471 ± 208 kcal/day), and the relationship between FFM and REE was similar to that of the PHP1A group (Fig. 3). These patients also met the recommended benchmark of ≥60 minutes moderate to vigorous intensity exercise daily (94 ± 40 min/day). Food intake during the buffet meals was lower in the patients with PHP1B (Table 4), and they did not show clear evidence of hyperphagia as evaluated by questionnaires (Table 3). All three patients correctly ranked the sucrose solutions by order of sweetness, and they preferred the higher-sucrose solutions (26.0% ± 17.3%).

Discussion

PHP1A is a rare genetic obesity disorder. In contrast to other monogenic obesity syndromes, in PHP1A a reduction in energy expenditure is a major driver of obesity, presumably caused by decreased signaling through MC4R in the hypothalamus (10, 11). We also present evidence that REE may be lower in patients with PHP1B and multihormone resistance. We again demonstrated a lower REE level in this larger cohort of patients with PHP1A but also furthered our knowledge of energy balance by objectively measuring food intake and activity levels. Activity level was similar between groups, but normalized to REE, the PHP1A group consumed more calories during both buffet meals and on the food frequency questionnaire, although this difference reached statistical significance only for the breakfast buffet meal. The abnormal weight gain in PHP1A appears to be driven by mild hyperphagia in the setting of reduced REE. Hyperphagia and increased food intake were not evident in the small PHP1B group. This finding is consistent with the model of reduced Gsα-coupled signaling at MC4R in the hypothalamus. In the murine model, deletion of the maternal Gnas allele from the central nervous system alone leads to an obesity phenotype through reduced REE without a concomitant decrease in food intake (12). This phenotype is recapitulated with either dorsomedial hypothalamus–specific MC4R deficiency or maternally inherited Gsα-deficiency (31). Accordingly, patients with PHP1A and patients with MC4R deficiency consumed similar calories (~40 kcal/kg FFM) during an ad lib buffet meal (13). Although this amount is higher than that consumed by normal weight controls, it is less than in syndromes with severe hyperphagia, such as leptin deficiency (32). However, humans with MC4R deficiency have not been consistently shown to have decreased REE (13, 33). The MC4R is now recognized to regulate neuronal firing through additional pathways outside Gsα, including G(q/11)α (34) and the inward rectifying potassium channel Kir7.1 (35). These non-Gsα pathways appear to regulate hyperphagia but not energy expenditure and may explain the profound energy expenditure phenotype of Gsα deficiency and the more variable energy expenditure phenotype in patients with mutations in MC4R. Interestingly, symptoms of hyperphagia are stronger in a subset of patients with PHP1A, and additional research is needed to elucidate this dichotomy.

As we hypothesized, patients with PHP1A preferred higher sucrose concentrations. They also had more difficulty ranking the sucrose solutions in order of sweetness, which could be related to cognitive impairment or decreased taste and olfaction. A mild impairment in olfaction and taste has been previously reported in PHP1A (36–39). Research suggests that impaired taste increases desire for more intense flavors and typically higher caloric content (33). This knowledge could influence dietary interventions in PHP1A.

This study was the first to evaluate glucose homeostasis in children with PHP1A. We found that children with PHP1A have glucose intolerance despite lower fasting glucose and HOMA-IR compared with controls well matched on sex, race, age, and BMI. When challenged with an oral glucose load, children with PHP1A had persistent hyperglycemia, and 25% met criteria for impaired glucose tolerance despite similar OGIS. There were no clear differences between the patients with PHP1A with and without impaired glucose tolerance, other than a trend toward higher fasting insulin. This finding is in contrast to the adult literature, where PHP1A is associated with a reduction in insulin sensitivity, particularly after an oral glucose load (17). It is possible that this is an age-related difference, although it is difficult to compare our pediatric cohort with the adult data because they included only three adult patients who were nondiabetic with PHP1A. A similar pattern of prolonged hyperglycemia was seen in the PHP1B kindred despite the group’s lower BMI. This pattern of post–glucose load hyperglycemia without fasting hyperglycemia could be explained by greater peripheral insulin resistance, impaired incretin response, or hyperglucagonemia.

The incretin hormones glucagonlike peptide-1 and gastric inhibitory polypeptide signal through G_sα-coupled GPCRs in the pancreas. Resistance to glucagonlike peptide-1 and gastric inhibitory polypeptide could decrease early insulin response, increase β-cell apoptosis, and impair β-cell proliferation, but it was not measured in our study (40, 41). Failure of hyperglycemia to suppress glucagon could also cause persistent hyperglycemia after a glucose load, and further research is needed.

Surprisingly, we found that HbA1c levels were lower in patients with PHP1A, regardless of dysglycemia. There was no difference in the slope of the regression line between the PHP1A group and controls; therefore, we hypothesize that patients with PHP1A have shorter red blood cell half-life. Red blood cells express GNAS, and haploinsufficiency could shorten red blood cell half-life, thus falsely lowering HbA1c levels. Understanding this distinction is clinically important because HbA1c is a common screening test for diabetes, and falsely low levels may contribute to delayed diagnosis in patients with PHP1A. Fructosamine levels, a measure of glycemia over 2 to 3 weeks, are commonly used instead of HbA1c in patients with hemoglobinopathies and other disorders with decreased red blood cell survival (42). We recommend measurement of fructosamine in future studies in PHP1A.

Strengths of this study include a larger cohort of patients with PHP1A than previously studied and a well-matched, contemporaneous control group. Two of the patients did not have genetic confirmation of PHP1A, but both had subcutaneous ossifications along with the AHO phenotype and multihormone resistance, making a related disorder such as acrodysostosis unlikely. The majority of patients with PHP1A were pubertal (63%), only one patient was postpubertal (Tanner stage 5), and both pubertal and prepubertal patients demonstrated glucose intolerance. The PHP1B group was limited to a single kindred, and so generalizability is limited. Measurement of food intake is difficult, and therefore we relied on observed ad lib buffet meals to precisely quantify food intake in a controlled setting, bolstering the self-reported data on hunger and dietary intake. We have improved on our previous studies of fasting glucose and insulin with dynamic testing, but more in-depth studies are needed to evaluate the incretin response, β-cell function, and insulin resistance in PHP1A. It is clinically important to understand the pathophysiology of dysglycemia in this disorder because it will inform first-line therapy. Additional studies in PHP1A, PHP1B, and PPHP can help us to better understand the role of Gsα in glucose homeostasis.

Glossary

Abbreviations:

AHO

Albright hereditary osteodystrophy

BMI

body mass index

CRC

Clinical Research Center

DXA

dual-energy x-ray absorptiometry

FFM

fat-free mass

GPCR

G protein–coupled receptor

Gsα

stimulatory G protein

HOMA-AD

homeostatic model assessment of adiponectin

HOMA-IR

homeostatic model assessment of insulin resistance

HOMA-β

homeostatic model assessment of β-cell function

MC4R

melanocortin-4 receptor

OGIS

oral glucose insulin sensitivity

OGTT

oral glucose tolerance test

PHP

pseudohypoparathyroidism

PPHP

pseudopseudohypoparathyroidism

REE

resting energy expenditure

T2D

type 2 diabetes