No Need for Routine Endoscopy in Children With Celiac Disease on a Gluten‐free Diet

See “Zero, One, or Two Endoscopies to Diagnose and Monitor Pediatric Celiac Disease? The Jury Is Still Out” by Leonard and Fasano on page 270. Recently, a press release with the head line “Twenty percent of children with celiac disease do not heal on a gluten-free diet” was published, based on a publication in this journal (1). This press release was based on a retrospective study on 103 patients with celiac disease (CD) from 2 children's hospitals in Boston, MA, and cites the first author, Dr. Maureen M. Leonard from the Mass General Hospital for Children: “While the long-term effects are not known, persistent enteropathy may predispose pediatric patients with celiac disease to future complications and suboptimal growth” and “These findings suggest the need not only for a baseline endoscopy to confirm the diagnosis of CD but also considerations of a repeat biopsy to evaluate for remission.” The authors call for further studies to confirm their findings, to better understand the response to a gluten-free diet (GFD), and to evaluate further treatment options. This press release was published on different Web sites including those of pediatric associations, and has raised considerable uncertainty among both pediatricians and patient organizations. In patients and their parents it has induced anxiety on further complications and prompted requests for additional endoscopic examinations. It raised questions about the potential need for other therapeutic interventions in addition to gluten exclusion, due to the incomplete mucosal recovery on GFD. The article, on which the press release was based, included chart reviews of 103 children with biopsy-proven CD who underwent a second upper endoscopy with duodenal biopsies at least 12 months after starting GFD. The endoscopies were not performed systematically. The reasons for repeating endoscopy were persistent or new symptoms in 72 children, and confirmation of mucosal healing only in 11. Results for autoantibodies against tissue transglutaminase (tTGA) were available in only 71 patients, with positive or borderline results in a third of them. Based on chart notes, adherence to GFD was considered to be excellent in 94 of the patients. The authors found villous atrophy (Marsh 3) in 20 patients. The presence of Marsh 3 lesions was neither related to tTGA positivity nor to symptoms reported in chart notes. We consider the study to have important limitations: 1. Selection bias: The children reported represent a highly selected group, with 70% undergoing repeat biopsy because of symptoms. Taavela et al (2) demonstrated a significant correlation of symptoms and the villous height/crypt ratio in 225 otherwise unselected celiac patients after following a GFD for at least 1 year (median 3 years; 11.8% with Marsh 3 lesions). The high prevalence of symptoms in the case series reported by Leonard et al suggests a selection bias toward patients with enteropathy, and selection bias coming from 2 quaternary centers is likely similar. 2. Assessment of adherence to GFD: The authors relied on physicians’ or dieticians’ subjective assessment and chart notes for assessing GFD adherence. The time intervals between these visit notes and the endoscopy are not reported and GFD adherence assessment was not based on a validated dietary questionnaire score as used in other follow-up studies (3). Of importance, for the evaluation of persistent histological alterations on GFD, children admitting nonadherence to GFD should have been excluded from the analysis. 3. Time interval between start of GFD and repeat endoscopy: Leonard et al re-endoscoped their patients after a median time of 29 months (IQR 16.8, 48.1) on GFD (1). This means that a quarter of the patients were rebiopsied between 12 and 17 months after starting the diet. Multiple serological follow-up studies in children with CD, however, indicated that 28% to 50% of patients were still tTGA positive 12 to 18 months after starting a GFD, 15% to 31% after 18 months and 22% after 24 months depending on baseline values and the ELISA test used (4–6). If monitoring is performed with a radiobinding assay, the time to seroconversion to a negative test result may take 5 years or longer in spite of strict GFD and disease remission (5). Therefore, a positive test within the first 18 to 24 months after starting a GFD may not indicate nonadherence to diet or persistent enteropathy, as long as tTGA-levels are constantly declining. On the contrary, negative tTGA results after an even more prolonged time on GFD make the presence of Marsh 3 lesions in childhood CD unlikely, as indicated by none of 22 seronegative children in the Austrian cohort (7) and none of 97 in an Australian study (8), both studies cited by the authors. 4. Lack of standardized tTGA testing: Serology had been performed in different laboratories using tests from different manufacturers. Results were considered for analysis if the test was performed within 4 months of repeat endoscopy. For a correlation between tTGA titers and histological findings the tTGA assessment should ideally be performed at the time of endoscopy and not >2 weeks apart. Incidental or voluntary ingestion of gluten during an otherwise compliant GFD leads to an increase in tTGA and mucosal lesions after challenge with even small amounts of gluten and occur within 2 to 4 weeks (4,9). Considering these major limitations, it does not surprise us that tTGA results in the present study did not correlate with histological findings. 5. Histological assessment: Biopsies were not read by an independent pathologist who was blinded to clinical information. No information was given on the location (bulb or descending duodenum), the quality of the biopsies or the orientation. It is well established that there is considerable interobserver variability in the interpretation of biopsies, particularly regarding biopsies with mild villous lesions (10–12). The villous atrophy prevalence of 19% reported by Leonard et al is considerably higher than the 5% and 9%, respectively, found in follow-up studies in the Australian and Austrian cohorts (7,8). The authors of these well performed studies concluded that complete mucosal recovery is highly likely in children reporting to be compliant with a GFD. They also conclude that negative serological tests identify mucosal recovery, which obviates the need for repeat endoscopic biopsies in such cases. All 3 studies included children who were on a GFD for at least 12 months. It appears possible that some children with complete flattening of the villi at the time of diagnosis (Marsh 3c) may have improved to Marsh 3a after introducing GFD. It is more likely that children with continuous enteropathy may not have been fully compliant with the GFD. Nonadherence was reported by Leonard et al in 8 of 103 children and may have been detected in further patients with more detailed and standardized dietary assessment. As already stated, these patients should have been excluded from the analysis. There is now good evidence also from unselected adult follow-up studies that an initially severe mucosal lesion takes several years to recover on a GFD and that persistent villous atrophy is mostly due to not adherence to a GFD (13,14) On the basis of the arguments given above we do not think that the data provided by Leonard et al (1) give sufficient support to the statement of the authors and the accompanying editorial (15) that routine repeat biopsies should be performed in all children at diagnosis and after the initiation of GFD to control for mucosal healing. Without good evidence subjecting children to routine endoscopy is not justified when enough data can be drawn from dietary history and serology. An approach with repeated biopsies in patients with CD had been recommended from 1979 until 1990 in the ESPGHAN CD guidelines, which proposed biopsies at 3 time points: at diagnosis before starting a GFD, on the GFD to prove mucosal healing, and after a gluten challenge to confirm the gluten-induced enteropathy (16). The development of EMA tests led to a change of the guidelines in 1990, which recommend repeated biopsies during GFD and after gluten challenge only in those children with negative celiac-specific antibodies, and in young children diagnosed below 2 years of age (17). The identification of tissue transglutaminase as the autoantigen in CD and the introduction of highly specific and sensitive tests, along with the well-documented correlation of tTGA titer levels with the severity of mucosal damage, led to the new ESPGHAN guidelines with the option to diagnose CD without biopsies, if certain conditions are fulfilled (18). One of these is the use of an adequate serological test, validated by EMA testing, a requirement that was not met in studies questioning the validity of the ESPGHAN guidelines (19). We conclude that the available information based on numerous studies and a long experience by a large group of pediatric gastroenterologists with thousands of celiac children indicates: 1. Most cases with persisting tTGA and/or EMA positivity after a GFD for at least 2 years with or without villous atrophy are related to incomplete compliance with the GFD. 2. Time to mucosal recovery after starting a GFD may take longer than 1 or even 2 years but has no real clinical consequences. The CD working group of ESPGHAN strongly advises against regular rebiopsy in children on a GFD, which involves unnecessary risks and a burden for patients and their families with a negative impact on quality of life and pointless high medical costs. Re-endoscopy should be reserved for symptomatic patients—in particular when seronegative—in which other conditions are suspected and for patients who need gluten challenge to establish the diagnosis.