Glucose Homeostasis and Genotype-Phenotype Interplay in Cystic Fibrosis Patients With CFTR Gene ΔF508 Mutation

We performed a retrospective study of all nonpediatric subjects (adolescents and adults 14–42 years old) with identified CFTR ΔF508 mutation who attended the Université Catholique de Louvain Cystic Fibrosis Unit for regular follow-up over the course of 2005. A total of 76 individuals were eligible for the study (age 25 ± 9 years, 47 and 53% male and female, respectively). The cohort included 33 patients who were heterozygous for the ΔF508 mutation while carrying various other corresponding alleles, collectively referred to as “other” (ΔF508/other) and 43 homozygous individuals (ΔF508/508). The reference population of nonmucoviscidosic subjects included consecutive diabetic outpatients with type 2 diabetes (n = 450) or chronic alcoholic pancreatitis (n = 35), followed in our center over the study period by the same investigators (M.P.H. and M.B.) (individual data not shown).

Nutritional status was evaluated by BMI (weight in kilograms divided by the square of height in meters). The presence of clinically significant overt pancreatic exocrine insufficiency was indirectly inferred from chronic requirements for oral pancreatine supplementation (on the basis of daily prescribed intakes of pancreatic enzyme capsules). Lung status was assessed by means of function tests, which included measurement of forced vital capacity (FVC) and forced expiratory volume over 1 s (FEV1).

The total group of heterozygous and homozygous individuals was subdivided according to glucose tolerance status. Glucose tolerance was estimated according to criteria for IFG, IGT, and diabetes based on the reports of the Expert Committee on the Diagnosis and Classification of Diabetes and of the Cystic Fibrosis Consensus Conference (7,8,17,18). As recommended by the Cystic Fibrosis Consensus Conference (8), diabetes was also substratified according to the presence or absence of fasting hyperglycemia. Due to the small number of patients with IFG and IGT (pre-diabetes) and the transient nature of these conditions toward full-grown diabetes those patients were combined with diabetic patients for modeling analysis and thereafter referred to as those with AGH.

β-Cell function and insulin sensitivity were assessed by homeostasis model assessment (HOMA) in patients with and without AGH at a time when patients were not on active steroid therapy (18,19). In normoglycemic subjects, HOMA was calculated from fasting glucose and insulin levels obtained at the last outpatient clinic before data collecting. HOMA testing in AGH subjects was performed an average of 3 years earlier than in NGT patients, since it usually corresponded with the fasting measurements that led to diabetes diagnosis, performed before implementation of lifelong glucose lowering therapy. HOMA of β-cell function (HOMA-B) values were plotted as a function of HOMA of insulin sensitivity (HOMA-S), defining a HOMA product (β-cell function[%] × insulin sensitivity [%] [B × S]; normal value 100%). Such a product corresponds with the true underlying β-cell function adjusted for individual insulin sensitivity and therefore underlies the need for successive therapies in any type of diabetes. Kahn et al. (20) previously reported that these two variables followed a hyperbolic relationship in individuals with NGT. Thus, if insulin secretory capacity decreases but HOMA-S also increases, according to this relationship, an NGT status should be maintained, albeit with a B × S function exhibiting different geometry. AGH develops when one variable or the other does not compensate, with the individual then departing from the normal hyperbolic relationship (21). In order to evaluate the magnitude of impairment of glucose homeostasis in cystic fibrosis patients, the cohort's results were plotted against a series of HOMA product means, calculated from two local reference populations with common types of diabetes (type 2 diabetes and diabetes secondary to chronic alcoholic pancreatitis).

BMI and lung function were assessed before initiation and after a mean of 4.5 years (n = 21) of insulin therapy. For genotype analysis, genomic DNA was isolated from whole-blood samples to investigate the presence of the CFTR ΔF508 mutation by PCR using allele-specific primers (INNOLIPA; Innogenetics, Ghent, Belgium). Insulin and/or C-peptide concentrations were measured with conventional radioimunoassay. A1C was measured by ion-exchange high-performance liquid chromatography. Islet cell cytoplasmic antibodies as well as GAD65 and antityrosine phosphatase (IA2) were determined by radioimmunoassay (CISBIO, Diegem, Belgium) in eight diabetic patients.

Results— are presented as means ± 1 SD, medians, or proportions. The significance of difference between groups was assessed by two-tailed, paired or unpaired Student's t test or Welch's test (22) for parametric and nonparametric data distributions. Differences between respective proportions were evaluated using χ² test. Since uncertainty in both HOMA-B and -S parameters could propagate error in their hyperbolic products, each measured value and its SE of estimate were used to obtain after error propagation, a z expression and its SE, that was used to compare groups. Differences in means or proportions were considered statistically significant at P values <0.05.

Main clinical data are described in Table 1. Age at diagnosis of cystic fibrosis was lower in homozygous than in heterozygous subjects (P = 0.001), while age, sex ratio, and BMI were comparable.

As shown in Fig. 1A, AGH was observed in 24% of ΔF508/other patients (n = 8) and in 40% of ΔF508/508 patients (n = 17) (P = 0.24). One heterozygous patient had IFG/IGT while seven (21%) had diabetes. Of homozygous subjects, 7% (n = 3) had IFG/IGT and 33% (n = 14) had diabetes (not significant). Five patients in the ΔF508/other group had fasting hyperglycemia when compared with six subjects in the ΔF508/508 cohort (not significant).

There was a trend for more impaired lung function, as reflected by absolute FEV1 (P = 0.084) and higher colonization rate with Pseudomonas aeruginosa, in ΔF508/508 subjects (P = 0.074). Pancreatic exocrine insufficiency was present in 52 and 100% of heterozygous and homozygous individuals, respectively (P = 0.001). Patients without oral pancreatine supplementation were older than those receiving daily substitution (29 ± 10 vs. 24 ± 8 years, P < 0.05). They were also older at the time of cystic fibrosis diagnosis than those receiving pancreatine (19 ± 12 vs. 3 ± 6 years, P < 0.05). BMI was 22 ± 4 and 21 ± 3 kg/m² in subjects without and with supplementation, respectively (not significant).

The clinical and biological characteristics of cystic fibrosis patients with NGT and AGH are described in Table 1. The prevalence of CFTR ΔF508/508 mutation was 51% in NGT and 68% in AGH subjects, respectively (P = 0.20; Fig. 1B). Patients with NGT were younger than those with AGH (23 ± 8 vs. 29 ± 10 years, P = 0.006). Age at cystic fibrosis diagnosis as well as sex ratio and BMI were comparable in subjects with NGT or AGH.

AGH was diagnosed at a mean age of 23 ± 8 years. The diagnosis of IFG/IGT was based on IFG in one subject and on an abnormal oral glucose tolerance test in three subjects. Diabetes was diagnosed as a result of fasting hyperglycemia (>126 mg/dl) in 5 patients, of a random glycemia >200 mg/dl with symptoms in 5 subjects, and of an abnormal oral glucose tolerance test in 10 subjects. One patient attended the Mucoviscidosis Unit after diagnosis of diabetes was made in another hospital.

We observed a lower HOMA-B in patients with AGH than NGT (93 ± 49 vs. 125 ± 51%, P = 0.011). In contrast, HOMA-S was comparable in subjects with NGT and AGH. Adjusting for potential confounders such as age, sex, and BMI did not affect HOMA values distribution between groups. As a result, the mean HOMA product was lower in hyperglycemic than in normoglycemic individuals (89 ± 50 vs. 116 ± 40%, P = 0.013 before adjustment for error propagation). Thus, AGH patients had a marked reduction in β-cell function, explaining the bulk of hyperbolic loss despite a small increase in insulin sensitivity (Fig. 2). The difference between HOMA products did not, however, reach statistical significance once propagation of error in product terms was taken into account. Comparison between HOMA products obtained in other common types of diabetes, on the other hand, showed significant differences in hyperbolic values between all cystic fibrosis groups and type 2 diabetic or chronic pancreatitis patients, with the two latter groups lying close to the 25% hyperbole, hence displaying a much more severe deficit in β-cell function adjusted for insulin sensitivity (Fig. 2).

A1C at diagnosis in AGH subjects was 7.3 ± 1.3% (5.8 ± 0.4 and 7.0 ± 1.2% in patients with IFG/IGT and diabetes, respectively; P < 0.01). Eighty-two percent of patients with AGH were treated with insulin (0.52 ± 0.38 units · kg⁻¹ · day⁻¹). Current A1C levels were 5.5 ± 0.5 and 6.8 ± 1.2% in patients with NGT and AGH, respectively (P < 0.0001). Pancreatic exocrine insufficiency was present in 73% of patients with NGT and 92% of those with abnormal glucose metabolism, a difference close to statistical significance (P = 0.098).

There was a higher rate of (past) resorting to systemic use of steroid therapy in patients with AGH than in NGT subjects (40 vs. 16%, P = 0.04). Liver cirrhosis was more frequent in patients with AGH (24 vs. 8%), but the difference did not reach the level of statistical significance (P = 0.11). There was a trend for lower FVC (P = 0.059) and FEV1 (P = 0.071) in patients with AGH than in those with NGT (Table 1). The higher prevalence of P. aeruginosa colonization in AGH patients was not significant.

After initiation of insulin therapy, there was a BMI increase of 1.8 ± 2.0% (P = 0.03). We observed a trend toward improvement in absolute FVC (0.3 ± 0.6 l, P = 0.06). A significant increment in predicted FVC was nevertheless observed (5.6 ± 9.1%, P = 0.046). As far as absolute and predicted FEV1s were concerned, the small increases assessed following insulin therapy (0.1 ± 0.6 and 1.5 ± 4.4%, respectively) did not reach the level of significance.

Our data show that CFTR ΔF508 homozygous subjects differ from heterozygous individuals particularly in terms of age at cystic fibrosis diagnosis and the presence of pancreatic exocrine deficiency, as also reported by Kerem et al. (5). As far as IFG, IGT, and diabetes are concerned, overall data from homozygous and heterozygous ΔF508 subjects show a prevalence of AGH close to figures reported in previous studies (1,17,23,24). In this regard, it is worth mentioning that Moran et al. (8) observed a 35% diabetes prevalence in a cohort of 105 individuals aged 20–29 years. As life expectancy is slowly increasing in cystic fibrosis subjects (as a result of more rational use of targeted antibiotics), it is not unexpected to witness an ever-increasing prevalence of diabetes—a rather late-occurring complication of cystic fibrosis (6,7,23,25). In our study, mean age at diabetes diagnosis was 23 years, in keeping with previous reports (23,25).

Concurrent genetic factors could contribute to the development of cystic fibrosis–related diabetes, as also indicated by Derbel et al. (26), who mentioned the role of single nucleotide polymorphism-19 in the calpain-10 gene. Additional factors, such as genetic modifiers, could eventually modulate phenotypic expression of the disease (27). In our patients with ΔF508/508 mutation, all of whom had marked exocrine pancreatic insufficiency, only a mere 40% developed AGH. This is in line with previous data showing that the CFTR ΔF508 genotype-phenotype association with diabetes was not very strong, although genotype is considered to be a potential predictor of pancreatic exocrine status (1,5,9–11,13–15). In our subgroup of heterozygous individuals, in whom one-half had pancreatic insufficiency, the prevalence of AGH was 24%—a figure somewhat lower when compared with homozygous subjects, although the difference did not reach significance. This could be due to the relatively small size of the groups. However, our data are in keeping with the report of Marshall et al. (24) and imply that, even in individuals who carry one single copy of ΔF508 and who could have milder cystic fibrosis—particularly pancreatic exocrine insufficiency—than homozygous patients, screening for abnormal glucose tolerance is mandatory, a point suggested in several other reports (7,23). It is likely, however, that increasing life expectancy will be associated with a higher prevalence of AGH in all subgroups of patients, as the phenotype of older cystic fibrosis patients is only vaguely known thus far.

The precise pathogenesis of IFG/IGT and diabetes in cystic fibrosis remains quite controversial. No difference in sex ratio, as reported by Mackie et al. (1) and Marshall et al. (24) but not by others (12), was observed between our cystic fibrosis patients with or without NGT. Moreover, in the present study, in scope with other data (25), autoantibodies against β-cells were absent when measured at the time of diabetes diagnosis. Cross-sectional studies using hyperinsulinemic-euglycemic glucose clamp have documented peripheral (28–33) or hepatic (34) insulin resistance, which could contribute to abnormal glucose tolerance in patients with cystic fibrosis. As recently reported, elevation of tumor necrosis factor-α and impaired translocation of GLUT-4 were proposed as contributive mechanism(s) (16). By contrast, several studies underscored the paramount importance of insulin deficiency (28,35–37). These observations are in keeping with morphological data published by Couce et al. (38).

In the present study, we observed in our patients with AGH a significant β-cell defect in the presence of rather normal insulin sensitivity indexes compared with a common type 2 diabetic cohort. The observed decrease in B × S product is not significant but is trending toward insufficient compensation for decreased insulin secretory capacity, as suggested by departure from the normal hyperbolic relationship illustrated in Fig. 2. A larger cohort size and analyses of patients’ parity in comparison groups on steroid therapy will be necessary to more accurately compare insulin secretory and sensitivity compensation effects among cystic fibrosis patients with AGH.

HOMA testing was no longer performed subsequent to formal diagnosis of diabetes on obvious medical and ethical grounds regarding frail patients. This could also have contributed to the lack of segregation between groups in terms of hyperbolic product, since HOMA-B was likely to keep on decreasing with time in patients with AGH. In contrast, and as predicted in NGT patients with cystic fibrosis, the B × S product was higher. It is, however, interesting to note that the B× S product in cystic fibrosis patients with AGH was not as altered as that observed in our type 2 diabetic cohort nor as that in diabetic subjects with chronic pancreatitis; yet, 80% of AGH patients had a formal diagnosis of diabetes. This difference in secretory function adjusted for individual insulin sensitivity could account for the relative mild clinical expression of AGH in cystic fibrosis (13). In keeping with this, only five subjects had diabetes-related clinical symptoms at diabetes diagnosis.

Our results are in scope with, and expand, previously published data trending toward a β-cell defect rather than classical insulin resistance in cystic fibrosis patients with IFG/IGT and diabetes (28,35–37). In this view, it appears rational, in terms of a therapeutic prospect, to consider that most cystic fibrosis patients with AGH will at some time be considered eligible for insulin supplementation, especially since their lean body mass was usually low, in accordance with cystic fibrosis phenotype. In agreement with this observation, most of our diabetic cystic fibrosis patients indeed received daily insulin injections and may potentially have benefited to a certain degree from the anabolic and anticatabolic effects of insulin on lean body mass.

In summary, impaired glucose metabolism, either IFG/IGT or diabetes, is a frequent comorbidity in young adults who carry the ΔF508 mutation in the homozygous or heterozygous state. The interplay between genetic status and combined endocrine-exocrine pancreatic deficiencies seems to play the major role in the development of glycemic abnormalities. Impairment in insulin secretion is an important determinant of this condition, and as HOMA suggests, considerably contributes to glycemic dysregulation when compared with the magnitude of compensation from insulin resistance as documented by HOMA-S. Given the high prevalence of abnormal glucose tolerance in cystic fibrosis ΔF508 patients, regular screening for (pre-)diabetes is mandatory. On a pathophysiological basis, insulin supplementation seems a rational therapy for consideration in cystic fibrosis ΔF508 patients with diabetes, although this does not preclude the possibility that therapy directed toward a component of insulin resistance could also interact.

The authors appreciated the help of G. Reychler for screening the patients for CFTR ΔF508 mutations and I. Istasse and R. Lecocq for excellent secretarial assistance.