Our study describes the effects of a 3- 6
month diet of exclusively whole, unprocessed foods on clinical symptoms, mucosal pathology and celiac serology in patients with NRCD that were already properly following a strict GFD, as assessed by experienced dietitians with expertise in celiac disease. With the belief that the majority of such patients are continuing to react to miniscule amounts of gluten cross-contamination, the goal of this diet is to eliminate exposure to any possible source of gluten cross-contamination from packaged/processed foods, to include those labeled gluten-free. In this study, 82% of our study patients responded to the GCED, with resolution of symptoms and, when applicable, resolution of their prior celiac enteropathy. Most importantly, of the 6 study patients who met criteria for RCD prior to initiating the GCED, 5 (83%) responded to the diet; they did not, indeed, have RCD.
The GCED is not an attempt to enforce dietary adherence on a patient with detected persistent gluten ingestion, rather it modifies an already strict GFD to remove the possibility of trace gluten cross-contamination. Most CD patients can safely tolerate approximately 10
mg gluten cross-contamination daily, or 500 grams of food containing 20
ppm of gluten. However, there is a tremendous degree of variability within this population and some patients may have worsening histological changes with very low daily gluten exposure [22
]. The only grains permitted in the GCED are brown and white rice as even inherently gluten-free cereal grains have been found to be significantly cross-contaminated with gluten, presumably via comingling at harvest, transport and/or milling/processing. For example, in a recent study that tested 22 single-ingredient inherently gluten-free grains, seeds and flours, 32% of these products contained >20
ppm gluten; one product contained 2,925
ppm of gluten [25
]. Such a degree of cross-contamination illustrates how a significant amount of gluten may be ingested despite no apparent dietary indiscretions. Interestingly, the fact that the majority of the patients in our study were subsequently able to return to their previous strict GFD suggests that there is a degree of recovery that, once established, shifts these patients back to a more typical threshold of gluten reactivity.
There are a number of recent articles discussing the evaluation of NRCD, with a focus on the approach to finding other underlying etiologies, such as inadvertent gluten ingestion or microscopic colitis, that may explain persistent symptoms [6
]. It was not the intention of this study to duplicate this topic; rather we aimed to demonstrate that a number of GFD-adherent NRCD patients who are classified as RCD1 may not actually be refractory to dietary therapy. We propose a diagnostic algorithm (Figure
) that would avoid misclassification of non-refractory patients, prevent unwarranted immunomodulatory therapy and preclude further unnecessary and often costly testing and imaging modalities employed in the search for non-CD etiologies.
Figure 1 Proposed diagnostic algorithm for non-responsive celiac disease. 1. Dietary compliance should be assessed by an experienced dietitian to rule-out continued gluten exposure. 2. Consider colonoscopy with biopsies if symptoms clinically warrant. 3. To include (more ...)
Most of the patients referred to our celiac center for persistent symptoms, elevated serology, or enteropathy have unintentional dietary gluten secondary to a lack of proper dietary education. Therefore, our initial investigation begins with a comprehensive interview and evaluation by an experienced dietitian; only those patients without an apparent source of continued gluten exposure undergo further medical evaluation and consideration for the GCED.
After ensuring dietary compliance, the primary focus of an NRCD evaluation should be to rule-out the presence of RCD2, as this tends to follow an aggressive clinical course, with a high mortality rate primarily due to the development of intestinal lymphoma. Early recognition of RCD2 allows timely and more aggressive therapy, and encourages the use of diagnostic modalities such as video capsule endoscopy and CT enterography in order to exclude malignancy [15
]. We recommend repeat EGD with biopsies as the first investigation in GFD-adherent NRCD patients to differentiate those with celiac enteropathy, and therefore possibly RCD, from those without persistent mucosal disease. Patients with villous atrophy (i.e. Marsh 3 histology) should undergo immunophenotyping of the intraepithelial lymphocyte population by CD3/CD8 immunostaining or T-cell receptor clonal rearrangement analysis by PCR. Whereas a monoclonal population equates to RCD2, the presence of a polyclonal IEL population should not equate to RCD1 without first a trial of the GCED. Only those patients who do not
symptomatically and histologically respond to the diet should be diagnosed with RCD1 and treated appropriately [16
Persistent positive tTG IgA serology cannot be used to differentiate RCD from those without sustained mucosal damage, nor can it be used to differentiate RCD subtypes [16
]. Furthermore, studies have shown that normalization of tTG serology does not necessarily correlate with healed mucosa [29
]. These reasons highlight the importance of endoscopy in the evaluation of NRCD patients. Nonetheless, tTG is a useful marker for dietary gluten contamination [31
]. Our highest response rate was seen in that subset of patients with high celiac serology, with only 1 patient not responding to the GCED and, ultimately, being diagnosed with RCD1. The significance of a weak positive serology is less clear, and other studies of the NRCD population have also found little correlation between low titer positivity and underlying etiology [26
]. In our study, there was no difference between those patients who transitioned from high to weak positive serology and those patients who went from high to negative serology.
As Marsh 2 histology has, by definition, normal villi, we do not include this classification among those patients with potential RCD, despite the persistence of pathologic features. However, in those symptomatic CD patients with elevated serology, neither do we consider Marsh 2 lesions to be indicative of disease remission. 2 patients in this study had Marsh 2 histology prior to the GCED; both improved serologically and symptomatically and one underwent repeat biopsy with subsequent normal (Marsh 0) histology. For this reason we recommend that those patients with Marsh 2 histology undergo the GCED prior to any further work-up/evaluation. Our study also included two symptomatic patients with positive celiac serology yet Marsh 1 histology prior to initiation of the GCED; both of whom had subsequent resolution of symptoms and improved serology. Marsh 1 lesions would be consistent with disease remission, suggesting that these patients’ diarrheal symptoms were secondary to irritable bowel syndrome (IBS) or an underlying etiology other than gluten contamination. However, while we recognize that any diet modification carries with it some degree of placebo effect, the correlation of symptom improvement with a change in celiac serology is intriguing and the fact that both patients were subsequently able to return to their traditional GFD further argues for a non-placebo response. It should also be noted that one patient’s endoscopic report documented visualization of signs of villous atrophy. This raises the possibility of patchy villous recovery and subsequent missed villous lesions by biopsy [34
]. Given this response, we believe that a trial of the GCED should be considered in a symptomatic NRCD patient with positive celiac serology, despite apparently healthy villi.
Eighty-eight percent of our patients in this retrospective study were female. Although, as with other autoimmune diseases, there is a female predominance in CD, we were surprised that only 2 men were treated with the GCED since its development in 2005. As this study evaluated only those patients with NRCD who were treated with the GCED, we should not assume that this represents the entirety of our GFD-diet adherent NRCD population. However a previous study, after discounting those patients with inadvertent gluten ingestion, had an 81% female predominance in their NRCD population, supporting just such an assumption [27
]. Likewise, a study evaluating duodenal recovery in CD patients found that male gender was independently associated with achieving Marsh 0-1 histology, with females comprising 77% of the study population but 85% of their nonresponsive category [29
Owing to the retrospective study design, a major limitation of our study is the absence of repeat biopsies in some of our study patients, with 5 patients not undergoing endoscopy prior to the GCED and only 5 patients being biopsied after completing the GCED. Whereas the definition of RCD mandates persistent symptoms and, thus, an asymptomatic patient would no longer meet criteria for RCD, documentation of mucosal healing in those patients who responded clinically and serologically would strengthen the study results. Moreover, a disease-specific celiac symptom index would have aided in reliably assessing symptom resolution. Although compliance with the GCED was assessed by dietitian interview, we must consider the reality that strictly adhering to such a limited diet is difficult and likely more costly than a standard strict GFD given the focus on fresh, non-processed foods; these factors may affect adherence disclosure. Our study is also limited by lack of information on NRCD patients that were not started on the GCED, and we must recognize the possibility of a selection bias toward expected response to the GCED among our study subjects. Additionally, 8 patients who met NRCD criteria and were instructed on the GCED by our nutritionist never followed-up with our clinic and were therefore not included in our study. As it is our dietitian’s experience that GCED patients typically contact her periodically for clarification on diet, these patients likely did not start the diet. However, some of these patients may have complied with the diet and simply followed-up with their primary GI physician; as such, we may be missing useful data. Lastly, as it was our decision to include all NRCD patients who initiated the GCED, symptomatic patients with both negative serology and healed mucosa were not excluded; 75% of the noncompliant patients met this description, as well as one of the compliant patients. These patients essentially had little evidence to suggest that removal of trace gluten contamination would significantly affect their symptoms.