β-Cell Dysfunction, Insulin Sensitivity, and Glycosuria Precede Diabetes in Hepatocyte Nuclear Factor-1α Mutation Carriers

A total of 59 families with MODY due to HNF-1α mutations were identified from the Exeter Diabetes U.K. MODY collection. We were unable to contact 21 families, 11 did not have offspring of an appropriate age, and 8 declined to take part. A total of 47 subjects from 19 extended families agreed to undergo an oral glucose tolerance test (OGTT) either at their home or within our clinical investigations unit. Subjects with diabetes (n = 7), one previously not diagnosed, were excluded from analysis except as a reference group for the presence of glycosuria to allow assessment of glycosuria as a screening tool. The study was approved by a multicentered research ethics committee, and all subjects or their guardians gave written informed consent.

A detailed history was recorded from each subject. Height and body weight were measured with the subject in light clothing without shoes. Waist circumference was measured midway between the lowest rib and the iliac crest. Hip circumference was measured over the widest part of the gluteal region. BMI and waist-to-hip ratio were then calculated.

A standard (75-g) OGTT was performed after an overnight fast. For subjects <42.9 kg, a glucose load of 1.75 g/kg was given. For subjects with known diabetes, oral hypoglycemic agents were omitted for 48 h before the study, and insulin was omitted on the morning of the study. Blood was collected at −5, 0, 30, 60, 90, and 120 min for plasma glucose and serum insulin and at −5 or 0 and 120 min for total proinsulin. Blood was collected from fasting subjects for HbA_1c (A1C), lipid profile, testosterone, estradiol, luteinizing hormone, follicle-stimulating hormone, and genetic screening. Samples for glucose, insulin, proinsulin, lipid profile, and sex hormones were separated in a portable centrifuge and kept on ice until being frozen before analysis.

Urine samples were collected at 0 and 120 min during the OGTT, and the presence of glucose was ascertained by dipstix [Combur test strips, glycosuria present if >2.8 mmol/l(+)].

Patients were assigned as pre- or postpubertal according to clinical criteria and testosterone levels in men and estradiol in women (19,20).

Fasting plasma glucose (FPG) was calculated from the mean of the samples taken at −5 and 0 min. Patients were classified as having normal glucose tolerance (NGT), impaired glucose tolerance (IGT), and diabetes according to World Health Organization and American Diabetes Association criteria (21,22). Area under the curve (AUC) was calculated using the trapezoidal rule in subjects where all time points during OGTT were available. The peak glucose value was defined as the single highest plasma glucose during the OGTT.

Fasting serum insulin was calculated from the mean of the samples taken at −5 and 0 min. Incremental insulin (I₃₀) was the change in insulin concentration during the first 30 min of the OGTT (insulin 30 min − fasting insulin). β-Cell function was assessed using the insulinogenic index, which is defined as [(insulin 30 − fasting insulin)/(glucose 30 − fasting glucose)]. AUC for insulin was also calculated using the trapezoidal rule. Fasting intact serum proinsulin was calculated from the mean of samples taken at −5 and 0 min.

Insulin sensitivity was determined from the fasting glucose and specific insulin using homeostatic model assessment (HOMA). This is based on a model of the glucose-insulin feedback system in a homeostatic state (23). We used the modified version of this program (provided by Dr. J. Levy, OCDEM, Headington, Oxford, U.K.), which is adapted for specific insulin measurement, and assigned the value in control subjects as 100% (24).

Plasma glucose was measured by glucose oxidase–based methods (coefficient of variance [CV] 1.9%). A1C was measured by in-house high-performance liquid chromatography (CV 4.2%). Cholesterol, triglycerides, and HDL cholesterol were measured on Modular analyzers (Roche Diagnostics). LDL cholesterol was derived by calculation from the Friedewald formula. Estradiol, testosterone, follicle-stimulating hormone, and luteinizing hormone were measured on Immuno 1 analyzers (Bayer Healthcare). Insulin was measured using a specific immunoenzymometric assay (Medgenix, BioSource, U.K.) calibrated against International Reference Preparation 66/304 and with no detectable cross-reactivity with intact proinsulin or 32,33 split proinsulin (interassay CV <10% over range 95–1,038 pmol/l). Total proinsulin was measured by immunometric assays (antibodies 3B1/PEP001 and A6/3B1, respectively) with enzyme amplification as the detection system (Dako, Ely, U.K.). The assay was calibrated against the First International Reference Reagent coded 84/611 (interassay CV <15% over range 16–76 pmol/l).

DNA was extracted from peripheral lymphocytes using standard methods. Genomic DNA was amplified using primers previously described (25) with modifications to avoid allelic dropout (26). PCR products were purified and sequenced using BigDye Terminator chemistry (Applied Biosystems, Warrington, U.K.). Reactions were analyzed on an ABI Prism 377 DNA Sequencer (Applied Biosystems) and electropherograms compared with the index case and a normal reference sequence in order to assign the genotype.

Parametric data are presented as means ± SD and were compared by an unpaired t test. Insulin, total proinsulin, and HOMA data were log-transformed and are presented as the geometric means ± SD range. Other nonparametric data are presented as median (interquartile range) and compared by the Mann-Whitney U test. Categorical data were compared by Fisher’s exact test. OGTT data were compared by repeated measures for ANOVA. Insulin sensitivity was compared by univariate ANOVA to allow for the potential confounders of age, BMI, sex, and pubertal stage. Data were analyzed using Microsoft Excel 2000 and SPSS for Windows (Version 11). All tests were two sided, and P < 0.05 was considered statistically significant. Our study had >80% power at P < 0.05 to detect a difference of 1.5 SD in the variables measured.

A total of 47 subjects from 19 extended families agreed to undergo an OGTT. Ten subjects from these families were already known to have diabetes and a mutation in HNF-1α and did not take part. Twenty offspring were identified with a mutation (14 different mutations identified), of whom 11 had NGT, 2 had IGT (2-h plasma glucose 8.4 mmol/l and 9.4 mmol/l), and 7 had diabetes. Of the 27 nonmutation carriers (family control subjects), 26 had NGT and 1 had IGT (2-h plasma glucose 8.2 mmol/l).

The characteristics of the nondiabetic mutation carriers and family members without mutations (family control subjects) are shown in Table 1. The two groups were of similar age, pubertal stage, sex, BMI, and waist-to-hip ratio.

There was no significant difference in A1C between mutation carriers and family control subjects (5.6 ± 0.6 vs. 5.3 ± 0.3%, P = 0.10). FPG was similar in the two groups (5.1 ± 0.5 vs. 5.0 ± 0.4 mmol/l, P = 0.65); however, the mutation carriers had a significantly higher 2-h plasma glucose compared with family control subjects (6.2 ± 1.7 vs. 5.3 ± 0.9 mmol/l, P = 0.033) (Fig. 1). In subjects where all time points during OGTT were available, mutation carriers (n = 9) had a higher AUC for glucose compared with family control subjects (n = 26) (839 ± 160 vs. 740 ± 115 mmol · l⁻¹ · min⁻¹, P = 0.05). There was a significant difference during OGTT between the two groups on ANOVA for repeated measures (P < 0.001).

Mutation carriers had significantly lower fasting serum insulin (61 [35–106] vs. 88 [59–133] pmol/l, P = 0.022) and 2-h serum insulin (141 [57–350] vs. 295 [179–488] pmol/l, P = 0.003) compared with family control subjects (Fig. 1). Mutation carriers had a significantly lower early insulin response as shown by a significantly lower incremental insulin (125.2 [64.1–244.6]) vs. 428.3 [209.5–875.7] pmol/l, P < 0.001) and insulinogenic index (53.5 [31.5–90.9] vs. 226 [126.0–407.1], P < 0.001) compared with family control subjects. In subjects where all time points were available, mutation carriers (n = 6) had a significantly lower AUC for insulin compared with family control subjects (n = 21) (25,269 [17,726–36,021] vs. 53,309 [34,967–81,272] pmol · l⁻¹ · min⁻¹, P < 0.001).

There was no significant difference between mutation carriers and family control subjects in fasting total proinsulin concentrations (6.12 [2.97–12.64] vs. 40 [4.77–14.78] pmol/l, P = 0.14), 2-h total proinsulin concentrations (25.90 [12.2–54.8] vs. 32 [21.6–75.3] pmol/l, P = 0.09), or fasting total proinsulin-to-insulin ratio (0.09 [0.05–0.18] vs. 0.09 [0.06–0.16], P = 0.86).

In keeping with the similar fasting glucose but lower fasting insulin, insulin sensitivity as assessed by HOMA was significantly increased in mutation carriers compared with control subjects (144.6 [82.7–252.7] vs. 100 [66.9–149.4], P = 0.025). This remained significant when compared by univariate ANOVA, taking into account age, BMI, sex, and pubertal stage (P = 0.026). There were no significant differences in serum HDL, total-to-HDL cholesterol ratios, or triglycerides between the two groups (Table 1).

All seven diabetic subjects had glycosuria at 2 h, with a mean peak glucose of 20.4 ± 5.2 mmol/l (range 12.6–25.8). A total of 67% (4/6) also had glycosuria after fasting. A total of 38% (5/13) of the nondiabetic mutation carriers had glycosuria at 120 min compared with 0% (0/23) of the family control subjects (P = 0.0034) (Fig. 2). The nondiabetic mutation carriers who had glycosuria had a higher peak glucose value during OGTT than mutation carriers without glycosuria (10.2 ± 1.9 vs. 7.4 ± 0.7 mmol/l, P = 0.002). Mutation carriers with glycosuria had peak plasma glucose values ranging between 8.1 and 11.8 mmol/l compared with 6.2 and 8.4 mmol/l in individuals without glycosuria.

To assess if the differences remained when the study was limited to mutation carriers with normoglycemia, we removed the two mutation carriers and the one family control with IGT and repeated the analysis. The only change in the results was that the difference in the 2-h glucose was no longer significant, but the AUC for glucose in the OGTT remained significantly higher in mutation carriers (P = 0.04). All other results remained significant, the insulinogenic index was reduced in mutation carriers (55.8 [32.5–95.7] vs. 229.3 [126.5–415.5], P < 0.001), insulin sensitivity (HOMA-S) was higher (148.2 [83.3–264.1] vs. 100 [66.4–150.7], P = 0.024), and glycosuria was present in mutation carriers (4/11 vs. 0/22, P = 0.008).

We have shown that many of the phenotypes seen in subjects with diabetes due to a mutation in the HNF-1α gene are also seen in young mutation carriers before hyperglycemia develops. The key phenotypes present before hyperglycemia were marked β-cell deficiency, increased insulin sensitivity, and a low renal threshold for glucose.

Recruitment was difficult; even though this study was limited to OGTT (most often done at home), eight families were reluctant to take part. The high penetrance of HNF-1α mutations means it is difficult to find nondiabetic mutation carriers, especially when an OGTT is performed. Our study of 13 nondiabetic mutation carriers is second only to the Scandinavian cohort described by Groop and colleagues (9,10) and is bigger than other studies (8,11). Our subjects are considerably younger than previously reported series, with a median age of 15 years compared with 29 years (9,10). The mean age of diagnosis of diabetic subjects from the Exeter Diabetes U.K. MODY collection was 20.4 years (4), which is similar to other countries (9–11). Therefore, our cohort is more likely to include subjects who will go on to develop diabetes rather than representing a group in whom either genetic or environmental factors have reduced the penetrance of the mutations.

Although there was no difference in FPG, mutation carriers had a higher 2-h plasma glucose and AUC glucose compared with family control subjects. It has previously been shown that subjects with HNF-1α mutations have normal FPG and are classified as having diabetes by a raised 2-h plasma glucose during OGTT (27). This can be explained as the insulin secretion rate in HNF-1α mutation carriers is only reduced when plasma glucose levels exceed 8 mmol/l (8). This emphasizes the need to carry out OGTT when screening for diabetes in these families.

Mutation carriers had β-cell dysfunction and reduced insulin secretion compared with family control subjects even when the glucose tolerance was normal. This is in keeping with previous studies of nondiabetic mutant carriers that have used the OGTT, the intravenous glucose tolerance test, and hyperglycemic clamps (9–11).

There was increased insulin sensitivity by HOMA analysis of fasting glucose and insulin. This suggests normoglycemia may be maintained by increased insulin sensitivity despite reduced β-cell function. This is in keeping with a previous study that has shown increased insulin sensitivity in nondiabetic HNF-1α mutation carriers compared with NGT subjects from families with type 2 diabetes using an OGTT-derived insulin sensitivity index (10). Insulin sensitivity was not increased in glucose clamp studies of three normoglycemic nondiabetic mutation carriers (11). HOMA has limitations, because it is a fasting assessment rather than a dynamic one and therefore predominantly reflects hepatic rather than muscle insulin resistance, although it has been validated against the intravenous glucose tolerance test and euglycemic clamps in normoglycemic subjects (23,28). We cannot exclude the possibility that these mutation carriers have not developed diabetes as a result of their increased insulin sensitivity. In this case, it would reflect a characteristic of nondiabetic mutation carriers rather than of subjects with HNF-1α mutations per se. The finding in young subjects (mean 15 years) ∼5 years below the mean age of diagnosis suggests insulin sensitivity is a characteristic of all mutation carriers.

We found no significant difference in fasting or 2-h total proinsulin and proinsulin-to-insulin ratio between mutation carriers and family control subjects. This contrasts previous studies that have shown a reduced fasting proinsulin (9) or an increased proinsulin-to-insulin ratio (6). The differences may reflect smaller numbers, or the difference in proinsulin-to-insulin ratio (6) may be characteristic of diabetic HNF-1α subjects. The previous study only classified diabetes on the basis of FPG, so some of the “nondiabetic” subjects may have been identified as having diabetes if an OGTT had been performed.

A total of 38% of subjects with mutations had glycosuria at 2 h post-OGTT despite 2-h plasma glucose values <11.1 mmol/l. This is explained by the low renal threshold (14,16,17), which in diabetic MODY subjects, has been shown to be due to reduced glucose reabsorption due to reduced expression of the low-affinity/high-capacity glucose cotransporter (SGLT2) that is under direct transcriptional control by HNF-1α (17). The peak values seen suggest glycosuria occurs if plasma glucose rises above ∼8.4 mmol/l during OGTT. The renal threshold for HNF-1α subjects with diabetes has formally been measured at a whole-blood value of 6.5 mmol/l (equivalent to plasma glucose of 7.4 mmol/l). The previously described nondiabetic mutation carrier had glycosuria and a corresponding random plasma glucose of 7.7 mmol/l (16). The presence of 2-h glycosuria in all subjects with diabetes and a peak value in an OGTT of >8.4 mmol/l suggests that this could be used as a noninvasive screening tool in young children. A positive urine test for glycosuria after a large oral glucose load would guide the need for a formal OGTT and genetic testing.

In conclusion, we have shown that β-cell deficiency, increased insulin sensitivity, and glycosuria are all found in HNF-1α mutation carriers before the development of hyperglycemia. This study emphasizes the marked insulin deficiency seen in young subjects even before hyperglycemia develops. The presence of glycosuria at 2 h after an oral glucose load is of practical importance and should help to guide the need for an OGTT or genetic screening in young family members.

This study and the U.K. MODY collection were funded by Diabetes U.K. A.T.H. is a Wellcome Trust Research Leave Fellow.

We thank all patients and their referring clinicians. We acknowledge the help of Sue Ayres, Harriet Castleden, and Gill Salt during the sampling for this study; Lisa Allen and Emma Edghill for their help with genetic testing; and Ewan Pearson and Rinki Singh for their help with the manuscript.