Recombinant GH at supraphysiologic doses (0.045-0.050 mg/kg/day, up to 0.068 mg/kg/day) is recommended in TS, typically starting around 4-6 years old and ideally, before age 13 years (2). Growth promoting treatments are utilized until growth plate fusion. Several studies have examined the influences of recombinant GH on glucose metabolism during this extended treatment period and upon its discontinuation. Overall, studies have demonstrated a worsening of insulin resistance without increased prevalence of IGT or DM in TS patients treated with recombinant GH (19, 22–24). A dose-related increase in insulin resistance was noted in patients treated with higher doses of recombinant GH (0.045 mg/kg/day vs 0.0675 mg/kg/day and 0.09 mg/kg/day), as evidenced by increased area under the curve (AUC) for insulin during OGTT, insulinogenic index, and urinary C-peptide (24). There was no apparent difference in glucose metabolism in once daily versus twice daily divided doses of recombinant GH. Additionally, fasting insulin showed a sustained increase throughout GH therapy, in contrast to glucose-stimulated insulin, which showed no progressive worsening beyond four years of treatment (19, 22). Finally, insulin resistance induced by recombinant GH appeared to be reversible, as insulin levels declined upon discontinuation of treatment, although remained above pre-treatment levels (19, 22, 23). Studies involving non-TS controls also saw increased insulin levels in 46, XX females; therefore, the elevated post-treatment insulin levels in TS individuals were felt to be positively correlated with post-pubertal status and increasing age (22, 23). Thus, while recombinant GH treatment is responsible for reversible increases in insulin resistance, it does not explain the lifetime increased risk of hyperglycemia in TS.

Studies examining the effect of hormone replacement therapy (HRT) on glucose metabolism in TS demonstrate improvements in insulin sensitivity. Limited published studies have evaluated their effect on glucose tolerance. An investigation of the impact of six months of HRT with either oral or transdermal 17-beta-estradiol plus norethisterone in adult TS patients compared to healthy controls demonstrated that HRT did not worsen insulin resistance, but the AUC glucose on OGTT significantly increased (20). These differences were comparable between the transdermal versus oral estradiol groups, demonstrating no perceived benefit of one therapy over the other with regards to carbohydrate metabolism. A subsequent comparison of transdermal versus oral estradiol replacement on metabolic parameters in women with TS similarly did not demonstrate a perceived metabolic benefit of one therapy over the other although this study only measured fasting glucose (25). Additionally, HRT ameliorated the recombinant GH-induced increase in insulin resistance in women with TS (26). Another study also demonstrated that the GH-induced increase in insulin resistance was improved by the addition of estradiol; however, estradiol had no effect on glucose tolerance (27). Thus, HRT in TS seems to improve insulin sensitivity, but the effect on glucose tolerance (either alone or in conjunction with recombinant GH or alone) is not clear but is likely minor.

Oxandrolone is a non-aromatizable androgen utilized for adjunctive growth promotion in TS (28). Data regarding oxandrolone’s effect on glucose metabolism are mixed. A randomized, placebo-controlled study comparing recombinant GH in combination with placebo or oxandrolone (either at 0.03 or 0.06 mg/kg/day), demonstrated no significant impact on insulin sensitivity (29). There were no differences between (recombinant GH + oxandrolone) versus (GH + placebo) groups with regards to developing IGT or decreased insulin sensitivity; however, fasting glucose levels and hemoglobin A1c (HbA1c) values decreased more in those treated with oxandrolone compared with placebo. In contrast, some studies have demonstrated a worsening of glucose tolerance with oxandrolone (21, 30). Specifically, AUC glucose and insulin levels during OGTT rose in patients treated with oxandrolone, either alone or in combination with recombinant GH (21). Despite these changes, HbA1c levels and fasting glucoses in treated individuals remained normal. This study used oxandrolone doses higher (0.125 mg/kg/day) than current recommendations (<0.05 mg/kg/day) (2). Similar increases in AUC insulin levels were noted during OGTT in individuals with TS on GH/oxandrolone combination therapy in a subsequent study, which also demonstrated a reversibility of glucose intolerance in TS individuals upon discontinuation of these treatments (30).

In summary, the effects of oxandrolone and estrogen on glucose metabolism are less clear than the effect of recombinant GH, but all may impact glycemic response as well as insulin resistance. Accordingly, recombinant GH, estrogen, and oxandrolone should be noted and controlled for when designing and interpreting clinical studies evaluating glucose metabolism in individuals with TS.

Early in the study of hyperglycemia of TS, excess circulating GH was documented in 4 individuals with TS (74). Thus, it was hypothesized that excess GH secretion was responsible for the hyperglycemia seen in TS, similar to acromegaly. It should be cautioned that the early study utilized the tibia line assay, which can give values which are much higher than those from radioimmune assays (14). Five additional studies followed up on the GH hypothesis (17, 33, 47, 75, 76). The studies involved between 14-30 individuals of varying ages with TS. Three studies measured fasting GH levels using various radioimmunoassay techniques (17, 47, 76), and in one study, the assay method is not described (75). No difference in fasting GH levels was noted between individuals with TS and normal glucose metabolism and those with TS and hyperglycemia (75), or between individuals with TS and non-sex-matched controls. The third study did not report fasting GH levels (76). The fourth study observed higher fasting GH levels in TS patients than controls, however this difference was attributed to the fact that over half of the TS group was undergoing estrogen replacement (47). There also were no differences in average 24-hour GH levels or area under the curve for GH, which was sampled through an indwelling needle every 30 minutes for 24h in 10 patients (17). The fifth study reported GH levels in a combined group of individuals with TS and complete gonadal dysgenesis with 46XX karyotype, noting that those with DM had higher levels of plasma GH at the 15 and 30 minute time points; however, the frequency of abnormal GH responses in the diabetic group was not higher than in the non-diabetic group (33).

Two of these studies demonstrated normal rise in GH to hypoglycemic conditions (47, 76). Another pair of studies observed excessive increases in plasma GH levels during hyperglycemic conditions. This dysregulated GH response to hyperglycemia did not correlate with glucose tolerance (47, 75). Thus, while GH secretion to hypoglycemia in TS is intact, GH secretion may be dysregulated with respect to hyperglycemic conditions in individuals with TS. This may be exacerbated by HRT, but it does not seem to be a predominant mechanism underlying hyperglycemia in TS. A review of key GH studies in TS hyperglycemia can be found in Table 2 .

A high prevalence of organ-specific autoantibodies was an early phenotypic association made in studies of TS (77, 78). Celiac disease, an autoimmune disorder that is strongly associated with the HLA-DQA1*0501 and DQB1*0201 alleles (79), was subsequently noted to be highly prevalent in TS (80), increasing the plausibility of a type 1 DM (T1D)-like mechanism underlying hyperglycemia in TS. Two early studies (81, 82) assessed human leukocyte antigen (HLA) associations with TS: one demonstrated no skewing of HLA frequencies among those with TS and their parents (82), while the second revealed that the frequencies of B38 HLA class 1 antigens were increased in the TS population (81). One additional study specifically looked for enrichment of HLA DR3 or DR4 haplotypes, two high risk T1D HLA haplotypes (83). These two haplotypes were not identified in those with TS, but they were not studied specifically in the context of TS individuals with a glycemic phenotype (84). An in-depth assessment of T1D genetic risk loci has not been performed in individuals with TS with regards to glycemia.

There is no consensus as to whether the frequency of T1D is increased in TS (85–88). However, most cases of DM in TS do not resemble classic autoimmune DM (58, 85). Longitudinal testing for islet autoantibodies has not been completed in the TS population, so it is not known if indolent autoimmunity contributes to the TS hyperglycemia phenotype. Cross-sectional measurements of islet cell autoantibody (ICA) have been performed but were not identified in any individuals with TS (3, 32, 54, 89, 90). Glutamic acid decarboxylase autoantibody levels (anti-GAD) were measured in 4 studies (32, 54, 91), with a positive assay in 2-10% of individuals with TS. A study of 113 women with TS (3) measured both ICA and anti-GAD antibodies and reported that 2/113 (1.8%) were autoantibody positive but did not differentiate which antibody(s). The time-course of insulin, glucose, and insulinogenic index in response to glycemic challenge did not differ between the antibody-positive and antibody-negative subjects with DM. A study following 134 patients with TS for a mean of 5.4 years similarly demonstrated a prevalence of T1D of 1.5% (92). In contrast to the previous data, in a German study of 24 individuals with TS and DM, 78% were positive for islet autoantibodies (85).

The supply of TS pancreas specimens available for histological study is very limited, and this is reflected in the paucity of research literature on pancreatic pathologies in TS. Reports of only two TS samples have been published, and there was no indication of insulitis or other typical histological findings of type 1 DM (47). However, neither sample was obtained from a TS individual with known DM. Given the natural history of hyperglycemia in TS, it is unlikely that an aggressive T1D-like autoimmune process is the cause of TS hyperglycemia. To evaluate this hypothesis rigorously, it will be necessary to do longitudinal studies that combine glycemic phenotyping and islet autoantibody assessments with genotyping for genetic variants linked to high risk for T1D. The next section (6.3) will discuss evidence suggesting that defects of insulin secretion do exist in TS, making such studies potentially more important to undertake. Key studies related to autoimmunity are summarized in Table 3 .

OGTT-based analyses indicate that insulin secretion is impaired in TS and that this impairment contributes to hyperglycemia. Seven studies utilizing OGTT conclude that insulin secretion is impaired in individuals with TS who have IGT (3, 14, 16, 20, 93–95); several of these studies used age- (16, 20, 94, 95), sex- (16, 20, 94, 95), and BMI-matched (94, 95) controls. Additionally, several other studies have demonstrated that there are defects in insulin secretion even in young, non-obese individuals with TS and NGT as compared to age and sex-matched controls. In one large study (n = 25 women with TS), early defects in insulin secretion were accompanied by heightened insulin sensitivity, suggesting that perhaps glucose intolerance progresses over time in TS as insulin secretory capacity falls and insulin resistance becomes apparent (59).

Abnormalities in insulin secretion are less apparent when assessed via intravenous glucose tolerance test (IVGTT) compared to OGTT. Only two of the six IVGTT studies demonstrated any impairment in early insulin secretion (20, 59), with a modest correlation of declining insulin secretion with age (59). Thus, although defects in insulin secretion are sometimes detected by intravenous glucose administration (IVGTT), they are less pronounced than the defects seen by oral glucose administration, i.e., by OGTT.

Insulin secretion in response to non-glucose stimuli has been evaluated in even fewer studies, all with small sample sizes and sometimes lacking control groups. Early phase insulin response to tolbutamide is likely decreased based on results from two studies (75, 96). Insulin secretion in response to glucagon seems to be normal in TS. The peak insulin levels were similar in TS and controls, although with a lower AUC in those with TS (96). Similarly, insulin secretion in response to arginine (17) and to arginine plus GLP-1 also do not appear to be different between individuals with TS and controls (94). The discrepancy in beta-cell capacity in response to different secretagogues is consonant with the fact these stimuli involve different mechanisms of insulin secretion [74] and may provide areas for future investigation with regards to the mechanism of impaired insulin secretion in TS.

Studies evaluating insulin resistance in TS are less consistent than those evaluating insulin secretion, suggesting that insulin resistance is not the primary driver of hyperglycemia in TS. Poor reproducibility of these studies, in part, reflects the variability with respect to methodology, age of TS participants, matching strategies, and medications utilized in the TS group. Five total euglycemic clamp studies have been published in TS (17, 94, 97–99) – two concluded that insulin resistance was increased in TS but did not use adiposity-matched controls (97, 98). Thus, it is possible that excess adiposity is an unmeasured confounder. When individuals with TS are matched on BMI, comparisons of insulin resistance as determined via euglycemic clamp are equivocal: one study did not find differences in insulin sensitivity (13 women with TS) (94), whereas the other did find lower insulin sensitivity (4 adolescents with TS) (99). The negative study was matched on fat mass % and lean body mass % as determined by dual-energy X-ray absorptiometry, although the waist-hip ratio was higher in the TS group compared to the controls.

Additional evidence suggesting that insulin resistance is not the primary driver of hyperglycemia is highlighted in three studies comparing IVGTT-derived measures of insulin sensitivity in individuals with TS to controls (20, 59, 94). In all three studies, the subjects were matched on adiposity (either BMI or fat mass % or lean body mass %), and none of the TS patients demonstrated increased insulin resistance. In fact, the largest study (49 individuals with TS) revealed that those with TS had improved insulin sensitivity compared to controls (59). Furthermore, a study utilizing insulin tolerance testing reported that insulin sensitivity was higher in TS (14). Another stream of evidence also suggests that insulin resistance is not driver of early metabolic defects in TS, in that fasting glucose and insulin levels in euglycemic individuals with TS are similar or even lower than in controls (19, 36, 47, 96, 97, 100). OGTT-derived measures of insulin sensitivity also consistently demonstrate comparable insulin sensitivity in normoglycemic, non-obese TS women and controls (32, 59, 95, 101). Finally, to confirm the central obesity seen by abdominal magnetic resonance imaging, TS individuals underwent biochemical testing. All of them had the biochemical hallmarks of central adiposity but not of hyperinsulinemia (36).

Thus, based on multiple lines of evidence, insulin secretion in response to oral glucose seems to be diminished and may decline with age in TS. The insulin secretory responses to IV stimuli, including glucose, arginine, glucagon, and GLP-1 are less clearcut and warrant additional study with larger sample sizes and better matching. Insulin resistance does not appear to contribute consistently to early hyperglycemia in TS. It is likely that insulin resistance becomes evident as central adiposity increases in TS, and that this exacerbates the pre-existing defect in insulin secretion. Key studies summarized in this section can be found in Tables 4 and 5 .

Glucagon is a key counterregulatory hormone to insulin and has been hypothesized to play a role in the diabetogenic phenotype (102). Three studies, all published prior to 1991, investigated glucagon behavior in individuals with TS (17, 99, 103). All of them noted either normal or no differences in fasting glucagon levels. The third study also noted significantly higher glucagon levels in the individuals with TS following OGTT as well as diminished levels during insulin-induced hypoglycemia. These results may indicate abnormal glucagon secretion in TS but should be interpreted with caution as the controls in this study were not sex- or adiposity-matched. Synthesizing the limited data available, glucagon secretion may be dysregulated in TS, but there is insufficient evidence to say that hyperglucagonemia plays a predominant role in the heightened risk for hyperglycemia in TS. These studies are summarized in Table 6 .

As discussed in section 6.3, the early studies of hyperglycemia in TS reported disparate responses to intravenous and oral glucose challenges (15, 75). This implicates incretin hormones as possible contributors to hyperglycemia in TS. The incretin hormones, including glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like 1 peptide (GLP-1), are secreted by enteroendocrine cells in response to ingested macronutrients. The incretins potentiate insulin secretion in a glucose-dependent manner (104). OGTT followed by oral amino acid tolerance tests conducted in 20 individuals with TS demonstrated that those individuals with IGT detected on OGTT had similar plasma glucose and insulin responses to oral amino acids compared to those with NGT (75). These results suggest that individuals with TS and abnormal glucose tolerance can have normal tolerance and normal beta-cell sensitivity to oral amino acids and thereby provides further indirect evidence that alterations in incretin secretion may underly hyperglycemia in TS. Only two studies have directly measured incretin hormones during OGTT and contrast with each other. A small, early study of 12 adolescents with TS showed abnormal GIP response during the OGTT in those with IGT, and this impairment correlated with delayed insulin secretion (93). The other published study directly evaluating incretin concentrations in TS demonstrated no significant differences in serum GIP and GLP-1 levels between 13 individuals with TS and age-, sex-, BMI-, and lean body mass matched controls during OGTT, although notably only 2 women with TS had hyperglycemia (94). No TS studies have directly quantified the incretin effect in TS using isoglycemic IV glucose infusion. This is an area that warrants future study, as GLP-1 agonists are gaining traction to treat hyperglycemia in other conditions with high DM risk (105). Key studies discussed here are summarized in Table 7 .

The Clinical practice guidelines for the care of girls and women with TS recommend annual screening for DM after the age of 10 with HbA1c, with or without fasting plasma glucose (2). However, studies including HbA1c in TS are limited and show discordance between HbA1c and OGTT in the diagnosis of IGT and DM according to American Diabetes Association criteria in adults (3, 18, 106). It has similarly been demonstrated that 23% of girls with TS had abnormal OGTTs despite normal HbA1c (54). Isolated fasting glucose intolerance is uncommon in TS (14, 32, 35, 95) and paradoxically, a reduction of fasting insulin and plasma glucose concentrations has been associated with a deterioration in glucose tolerance in TS (20). As such, serial OGTT may be a more sensitive means to screen for glucose abnormalities in TS. A 2011 study demonstrated that a significant percentage of adult patients with TS lacked medical follow up, and those without transition care or preceding specialist care had an increased risk of having undiagnosed cardiovascular risk factor (107). Thus, it is crucial that patients with TS understand the necessity of and differences in metabolic screening as they transition to adult care.

There are no TS-specific treatment guidelines for DM. However, optimization of DM treatment should be a priority given the significant contribution of diabetes to morbidity and mortality in TS. Additionally, DM is associated with lower health-related quality of life in TS (108) making this an important area of research. There are no published studies reporting the types of DM treatments used in TS and how they vary by age and in different health care settings. There are also no studies reporting the efficacy of targeted lifestyle modifications or anti-diabetic agents to mitigate the progression from IGT to DM in TS – given the natural history of hyperglycemia in TS (3), this should be an area of priority. It could be that most individuals with TS and DM are started on metformin in line with American Diabetes Association guidelines for the pharmacologic treatment of type 2 DM (109). However, given that insulin secretory defects seem to exist early on, even before the development of glucose abnormalities, and fasting hyperglycemia is atypical in TS, this drug may be a less appropriate first-line hyperglycemia treatment for individuals with TS.