Systemic consequences of coeliac disease in children with type 1 diabetes – is it worth following a gluten-free diet?

Introduction

Type 1 diabetes mellitus (T1D) is one of the most frequent chronic disorders in children and adolescents caused by auto- immune destruction of pancreatic beta cells leading to profound insulin deficiency requiring lifelong insulin treatment. This process is triggered by the exposure of generally predisposed persons to environmental factors [1]. T1D may coexist with other immune-mediated chronic disorders like coeliac disease (CD) which shares a common genetic background in HLA DQ2-DQ8 molecules [2]. The main abnormality in CD is malabsorption of macro- and micronutrients, vitamins and minerals, which may lead the impairment of weight gain and linear growth in younger children. Older children with CD may also have intestinal symptoms like abdominal pain, chronic anaemia, diarrhoea or constipation and vomiting. However, CD is a systemic disorder which is manifested by many extraintestinal symptoms (fatigue, iron deficiency anaemia, short stature, delayed puberty, osteoporosis, dermatitis herpetiformis) or the disease may be initially asymptomatic. The only treatment available for CD is a lifelong strict gluten-free diet (GFD) which should be introduced after unequivocal diagnosis of CD [3]. According to recent European (ESPGHAN) recommendations, in children with normal total serum IgA and IgA-antibodies against transglutaminase 2 (TGA-IgA) exceeding ≥ 10 times the upper limit of normal (ULN) after confirmation of positive endomysial antibodies (EmA IgA) in a second blood sample while on a diet containing gluten diagnosis of CD is justified. In case of positive TGA-IgA < 10 times ULN a small bowel biopsy with at least 4 samples from the distal duodenum and at least 1 from the bulb should be assessed to confirm the diagnosis of CD by demonstrating subtotal villus atrophy according to the Marsh classification [4]. CD can also be diagnosed without duodenal biopsies in asymptomatic children, using the same criteria as in patients with symptoms. ESPGHAN did not address diagnosic criteria of CD in diabetic children. According to ISPAD guidelines in the symptomatic children with T1D a biopsy-sparing approach may be considered in individual cases but after consultation with a paediatric gastroenterologist, the child and family. In asymptomatic children with positive serology the decision to perform duodenal biopsies for confirmation of gastrointestinal pathology and when to do that should be discussed with parents and the child [5]. In diabetic children transient TGA positivity with spontaneous normalization was reported thus follow-up (3–6 months) before the decision to perform the biopsy is warranted [6].

The frequency of coeliac disease in patients with T1D is 1.9% to 9.8% [7, 8]. In contrast to CD-only children, CD in T1D children is often asymptomatic, which raises the question of the necessity of introducing additional dietary restrictions which can be difficult to maintain in people with many duties related to diabetes treatment [9]. There are also individual opinions to be found in the medical literature on not forcing GFD in asymptomatic non adherent CD children with type 1 diabetes taking into account not consistent data on the influence of GFD on different parameters in T1D patients [10]. Such patients could be monitored carefully for the appearance of symptoms or unexplained difficulties in achieving glycemic control in which case prompt action could be taken [9].

Furthermore, GFD is characterized by a high glycaemic index, saturated fats content but a limited amount of fibre and proteins, which is contrary to nutritional recommendations in T1D and may adversely affect glycaemic control by increasing glucose variability [11, 12]. Strictly controlled nutrition is essential in both T1D and CD, but instituting GFD especially in asymptomatic paediatric patients with T1D seems questionable to many of patients and their parents, because adding another limitation to a diabetic dietary regimen might increase the financial and emotional burden. On the other hand, a possible impairment of growth, weight gain and bone health should be borne in mind if GFD is not followed, especially by poorly controlled diabetic children with CD.

Quality of life

Quality of life (QoL) is a subjective evaluation of one’s well-being, including both positive and negative elements at a specific point in time. It includes not only personal health (physical, mental, and spiritual), but also relationships, education/work environment, social status, wealth, a sense of security, freedom, autonomy in decision-making, social-belonging and physical surrounding. It has been proven that T1D may negatively affect QoL among children, especially girls [13]. It might seem obvious that the introduction of GFD further deteriorates QoL of patients with double diagnosis and their families. Gluten-free products are more expensive and less available, while GFD requires the elimination of commonly used carbohydrate products (e.g. breads, cereals, pastas, groats) and gluten-contaminated foods (e.g. sauces, sweets, processed meat), which may negatively influence GFD adherence. Non-compliance in CD is quite common, especially in youths [14, 15] who want to have an active social life and object to many treatment recommendations. There are few studies assessing the impact of GFD on QoL in youths with T1D and CD which have reported discordant findings.

In a case control study Pharm-Short et al. reported similar generic and diabetes-specific QoL in young patients with T1D and CD in comparison to subjects with T1D only but GFD nonadherent youths had lower QoL scores and so did their parents. Caregivers of symptomatic CD/T1D youths announced problems with family relationships as well as emotional and social functioning. T1D and CD duration as well as diabetes management regimen did not influence the results [16]. Similar results were obtained by Franceschi et al. who compared 86 double-diagnosed youths with 167 subjects with T1D only. There were no differences in QoL between the two groups and between their parents. However, QoL in the non-adherent GFD patients and their parents was significantly lower compared to the T1D-only group in physical, emotional, school, and disease domains. The GFD-adherent subjects and their parents presented higher family QoL scores than children with T1D only. In T1D/CD subjects the increased glycaemic control metric – TAR (time above range) was negatively correlated with QoL scores in the friends’ domain [17]. Sud et al. proved that an additional diagnosis of CD had little impact on QoL in children with T1D, but their parents reported significantly lower QoL scores in the psychosocial and social functioning domain as well as in the diabetes treatment barriers domain compared to their children. In turn, parents of children with T1D only also presented lower QoL scores than their children (in emotional functioning, diabetes treatment barriers, and communication domains). Quality of life was not impaired by adherence to GFD [18].

Neuman et al. and Söderström et al. demonstrated comparable QoL between children with new onset diabetes and no coeliac disease, who were subjected to a strict GFD just after the diagnosis for 12 months, and children who remained on gluten-containing diet (GCD) [19, 20]. Moreover, a GFD maintained over the first year after type 1 diabetes diagnosis was associated with better glycated haemoglobin (HbA_1c), a prolonged partial remission period and slightly slower C-peptide decline.

Weiman et al. recruited children and adults with T1D and asymptomatic CD into a prospective randomized study and assigned them to a GFD group or a gluten-containing diet group. Applying GFD over 12 months did not negatively affect QoL [21]. Similar results were obtained by Al Hayek et al. [22]. A recent systematic review, based only on studies of moderate-high evidence quality level and reporting data on objectively assessed adherence to GFD, showed that QoL was not different between T1D/CD and T1D only youths, as well as between pre- and post-CD diagnosed youths subjected to GFD [23].

Glycaemic control

The results of studies dedicated to the impact of CD and GFD in diabetic children on glycaemic control (glycated haemo- globin, acute complications, glucose variability) are inconsistent. Some authors found no significant differences in the frequency of diabetic ketoacidosis (DKA) [24, 25] and the mean HbA_1c between paediatric T1D patients with and without CD [26–33], even after a follow-up of 2 [34] or 5 years [35]. Abid observed that a GFD in T1D/CD children resulted in a reduction of severe hypoglycaemic episodes [31].

The mean HbA_1c in Sun’s study was lower in the 36-month period prior to diagnosis of asymptomatic CD in diabetic children compared with patients without CD. However, after 12 months of GFD treatment HbA_1c increased in the case subjects and achieved a similar level to that of the control group [36]. Somewhat different observations were made by Amin et al. in a case-control study with a 4-year follow-up. The mean HbA_1c was lower in the diabetic subjects with newly diagnosed CD compared with the control group and improved further after starting GFD [37]. In a multi-centre study by Kaspers et al. the HbA_1c values were lower in the patients with double diagnosis in comparison to diabetics without CD through the entire observation period, and daily insulin requirement did not differ between the groups [38].

A meta-analysis of 6 studies revealed no significant effect on HbA_1c in children with T1D and asymptomatic CD who were on GFD [39]. Another systematic review, which included only studies of moderate-high evidence quality level and reported data on objectively assessed adherence to GFD, showed no significant differences in HbA_1c, number of episodes of hypoglycaemia and total daily insulin requirement between youths with T1D/CD on GFD and those with T1D only [23]. In The CD-DIET Study patients with diabetes and biopsy proven asymptomatic CD were randomised to the GFD group or the GCD group. No HbA_1c differences were seen between the groups, although greater postprandial glucose increases emerged with the GFD group. The authors concluded that clinical vigilance was warranted with initiation of GFD [40].

Glycaemic control depends on the adherence to GFD. In the study by Nagl et al. only 36% (n = 218) of children with double diagnosis observed for 3 years were Ab-negative and their HbA_1c was lower than in Ab-positive and T1D-only patients at the end of the observation. There were no significant differences in severe hypoglycaemia or hospitalization episodes in the Ab-negative group compared to the Ab-positive group [41]. Comparable results were reported by Mozzillo et al. 87,3% of the youths in this study were fully adherent to GFD, those who were not strictly maintaining GFD had lower time in range (TIR), higher TAR in CGM and higher glycaemia risk index (GRI) compared to the T1D-only subjects [42].

Growth and nutritional status

The impact of CD on height, weight and BMI has been investigated in numerous studies. Some researchers found no differences in height/height-SDS [24, 27, 28, 30, 31, 36, 43], weight/weight-SDS [30, 31, 36, 43] and BMI/BMI-SDS [24, 27–31, 36, 43] in the CD/T1D children compared with the T1D-only, but others presented contradictory observations.

Simmons et al. conducted several studies dedicated to the problem of growth gain in double-diagnosed children. The authors observed that children with T1D and serologic evidence (TGA+) of coeliac disease had lower weight-SDS and BMI-SDS than serologically negative (TGA−) patients [25, 34]. After 2 years, the TGA+ group continued to have lower weight-SDS and BMI-SDS [25]. The younger age at CD diagnosis was associated with lower height-SDS in both sexes and lower weight only in boys. A total of 608 paediatric T1D + CD patients from the multicentre DPV registry were studied longitudinally by Nagl et al. [41]. Differences between Ab-negative (n = 218) and Ab-positive (n = 158) patients 3 years after biopsy were assessed and compared with 26,833 T1D patients without CD. No differences were found in BMI-SDS, height-SDS and weight-SDS between the Ab-positive group and the Ab-negative group at the time of CD diagnosis, but both groups had significantly lower values of anthropometric data than the T1D-only children with no change after 3 years [41]. Fröhlich-Reiterer et al. observed significantly lower weight-SDS and height-SDS in patients with biopsy-proven CD versus patients with T1D-only. After 5 years, the Ab-positive patients had decreased weight-SDS and height-SDS compared to the Ab-negative group [35]. Reduction in weight-SDS, height-SDS and also BMI-SDS was reported in other studies [37, 38, 43, 44]. In some cases a significant catch-up growth occurred after 12–24 months of GFD maintenance [45–47]. Only in one study by Sanchez-Albisua et al. was an improvement in height-SDS and weight-SDS observed among T1D/CD children with a good GFD compliance. However, the study was limited due to a small sample size and the lack of a control group [26].

Lipid profile and vascular risk factors

Relatively few studies have examined the impact of a GFD on lipid profile. In a retrospective multi-centre study by Salardi et al., at diagnosis of CD the diabetic children presented higher LDL cholesterol level compared to the controls. After 3 months of GFD treatment the LDL cholesterol level failed to normalize and the mean value of total cholesterol significantly increased but the mean triglycerides concentration significantly decreased [32]. Nagl et al. observed reverse trends in the cholesterol values. At baseline, the levels of total cholesterol, LDL cholesterol and triglycerides did not differ between children with double diagnosis and T1D only. After a 3-year GFD treatment, children with a good compliance had significantly lower total cholesterol and LDL cholesterol in comparison to the patients without CD, and dyslipidaemia occurred less frequently [41].

It is well known that low HDL cholesterol is associated with increased vascular risk. Warncke et al. analysed data from 392 centres cooperating under the DPV initiative. The CD/T1D patients presented significantly lower HDL cholesterol levels than the T1D-only group, even after adjusting for sex, age, diabetes duration, HbA_1c, BMI-SDS and insulin dosage. In the third year of observation the difference did not change and the HDL cholesterol level remained significantly higher in the control group [33]. The researchers compared vascular risk in diabetic patients one year before and after the CD diagnosis. The median HDL cholesterol levels and HbA_1c increased significantly between the two periods of time, and so did BMI-SDS. No differences between systolic and diastolic blood pressure SDS were found between the two periods [33]. The same finding was reported by Jessup et al., who noticed that patients with T1D and CD had lower HDL cholesterol values than the controls, with no differences in total cholesterol and non-HDL cholesterol between the groups [47].

Studies conducted in Sweden analysed the risk of diabetic retinopathy (DRP) and chronic renal disease in patients with a diagnosis of T1D before the age of 30 years. The authors found that patients with a duration of concomitant CD for more than 10 years were at a higher risk for DRP, and that long-standing CD was associated with chronic renal disease [48, 49]. CD was also an independent risk factor for microvascular complications after adjustment for confounders in patients from the German-Austrian DPV Database [50].

Bone health

Reduced bone mineral density (BMD) was confirmed in patients with untreated CD [51, 52]. In two recent reports 7–16% of children presented a reduced BMD at diagnosis of CD [53, 54]. T1D has also a well-known effect on lowering BMD and on the reduction of estimated bone strength. The decreased bone strength could be partly explained by reduced trabecular BMD and cortical area and adverse microarchitecture [55, 56]. Coexistence of T1D and CD may further exacerbate bone abnormalities. In the cross-sectional study of Simons et al. 123 children with T1D who were positive for TG had significantly worse BMD L1–L4 z-score compared with 129 children with T1D who were negative for TG. Additionally, both higher HbA_1c and TG were significant and independent predictors of lower BMD in multivariate analyses. The authors suggest a synergistic effect of hyperglycaemia and coeliac autoimmunity on low BMD [57]. Similarly, Pham-Short et al. demonstrated that youths with both T1D and CD have lower bone mineral content (BMC) relative to lean tissue mass and lower BMC, indicating abnormal trabecular and cortical bone development despite similar bone and muscle size [58]. It should also be emphasised that significant improvement was achieved in diabetic youths with CD after institution of GFD. Artz et al. reported an increase of BMD-SDS in diabetic patients with CD and severe villous atrophy who maintained strict gluten restriction for at least 12 months [43]. Valerio et al. proved that the quality of bone as assessed by phalangeal ultrasound in patients with T1D and CD, who strictly adhered to GFD, was similar to that found in diabetic patients without CD. A higher prevalence of ostaeopenia was present in patients with both diseases who reported non-adherence to GFD [59]. It is important to diagnose CD in diabetic children early enough to start treatment with GFD at the time of peak bone mass development, otherwise bone mass accrual may be compromised

Conclusions

Unlike the CD-only patients, the majority of children with T1D and CD remain asymptomatic and only 10% of them develop gastrointestinal symptoms [50]. The most challenging issue for children with double diagnosis is the necessity of maintaining dietary restrictions. Non-adherence to GFD is common among young patients with T1D and CD, especially those without symptoms. Some doctors raise the question of legitimacy of introducing additional dietary limitations arguing that avoiding gluten can be expensive both financially and emotionally with no measurable health benefits. Numerous studies prove the opposite – the quality of life of paediatric diabetic patients is not impaired by adding further restrictions. There is evidence suggesting that GFD has a protective role in many aspects – glycaemic control, growth, lipid profile, vascular complications and bone health. GFD treatment neither increases HbA_1c nor influences daily insulin requirement in CD/T1D children, but may impact short-term glucose variability. Moreover, diabetes control may improve significantly in children with a good GFD compliance. Many reports indicate that untreated CD may adversely affect the lipid profile, thus further increase cardiovascular risk in diabetic children. A GFD adherence at least stabilizes lipids and may improve triglycerides and LDL cholesterol concentrations. Numerous studies proved that diabetic patients with silent CD have decreased weight-SDS, height-SDS and BMI-SDS. GFD may improve these parameters. Finally, asymptomatic CD in diabetic children, especially those with suboptimal diabetes control, adversely affects bone mass accrual at the time of peak bone mass development. That is why early introduction of GFD is crucial to bone health. Current international guidelines recommend routine screening of patients with T1D for CD and introducing a properly balanced GFD if CD has been confirmed. GFD should be implemented in all children diagnosed with CD, even if they are asymptomatic.