Use of flow cytometric analysis of IELs as a routine diagnostic procedure in CD was introduced by Camarero et al. in the year 2000 in a pediatric series of 54 CD patients (24). This study evidenced the permanent imbalance in the density ratio of γδ⁺ vs sCD3^- IEL subsets that was almost pathognomonic of celiac conditions with independence of the clinical stage. It was termed celiac lymphogram. The authors designed a mathematical model to calculate the probability of suffering from CD that yielded a sensitivity of 94.4%, a specificity of 94.9%, a positive likely ratio (LR+) of 18.5 and a negative LR of 0.06. More recently, the same group has validated the diagnostic value of the lymphogram in an extensive pediatric series comprising 602 active CD patients, 92 on a GFD, 45 potential CD and 470 non-CD controls (27). When using stringent cut-offs of γδ⁺ (>15%) and sCD3^- (<6%), 90% of the active CD patients fitted in a complete “celiac lymphogram” with a 100% probability of disease. The remaining 10% of the active CD group displayed a “partial lymphogram” with lower disease probability. In this scenario, an isolated decrease on sCD3^- IELs yielded higher specificity and sensitivity for disease detection, with a disease probability of 76% vs 21% when the partial lymphogram was at expenses of an isolated increase of γδ⁺. When patients were on a GFD, the majority showed partial lymphograms at expenses of an isolated increase of γδ⁺>15% accompanied by a rising tendency on the density of sCD3^- IEL subset >6%, which lead to conclude that an increase on γδ⁺ IELs is the unequivocal marker of CD enteropathy while the size of sCD3^- IEL subset represents a useful marker to monitor diet compliance and mucosal integrity. The same was true for potential CD patients, where the increase of γδ⁺ was a constant hallmark at any time of the natural progression of this condition. γδ⁺ IEL counts has also been pointed out as a risk factor of overt disease in the largest cohort of potential CD patients so far published (28). By contrast, a persistent increase of sCD3^- IEL subset seemed to protect against progression to villous atrophy, being a good biomarker in the clinical management of potential CD.

Nowadays, lymphogram has become a routine diagnostic tool in many hospitals (29–38), useful not only in pediatric CD diagnosis but also in the much more complex scenario of adults’ gastroenteropathology, as discussed in the next paragraph.

There are several pivotal studies in adults addressing the role of phenotyping of IELs by flow cytometry in the diagnosis of CD. Most studies considered together children and adults, but some ones yielded results only for adults. In 2017, Valle et al. described a series of 161 CD patients, including 66 adults. Specifically in adults, sensitivity was 89% and specificity 96%, although similar results were obtained in children and adults (93% sensitivity and 96% specificity considering the whole data) (38). In 2019, Nijeboer et al. evaluated 95 adult patients with a mean age of 53 years, they only considered the increase in γδ⁺ cells (cut-off value >14%), showing a sensitivity of 66% and a specificity of 97% (37). It is noteworthy that the CD patients with normal γδ⁺ IELs in that study were significantly older than those with abnormal values. The authors argued that advanced age may be a factor preventing the characteristic increase of γδ⁺ IELs in CD for unknown reasons, and thus might explain the lower sensitivity found in that study. In this sense, the effect of sex, age and the degree of histological damage was evaluated (32) in a study including 169 CD patients with mean age 18.8 ± 1.5 years (range 1–83 years). Statistically significant differences in the median percentage γδ⁺ cells between age groups, including children, were not detected. Differences for sex, histological damage (Marsh 1 vs. Marsh 3) or titers of IgA anti-transglutaminase antibodies were also not observed either. Results showed 82% sensitivity, 100% specificity and 100% positive predictive value (PPV). However, considering that the number of CD patients and controls older than 61 was very small, the authors stated that further studies on this population group will be awaited. Another recent study evaluated 107 adult CD patients with a mean age of 60 years (35). Applying the described ‘IEL lymphogram’ criteria suggested by Ruiz-Ramírez et al. (≥8.5% TCRγδ⁺ and ≤10% sCD3^- IELs), 64% sensitivity and 92.5% specificity was observed. Rather than using single linear cut-offs for sCD3^- and γδ⁺ IELs, the authors identified a discriminant function as % C D3⁺ I E L s+2x(%γδ⁺ I E L s)>100. This function differentiated CD from control biopsies in the hypothesis generating group. The results were replicated in a validation group and found to be independent of histology in patients on long-term GFD up to 12 years (combined sensitivity, 98.5%; specificity, 98%). A recent study was focused on elderly patients (31). Eighty-seven patients ≥50 years were included at baseline (45 aged 50–59 years, 23 aged 60–69 years and 19 aged ≥70 years). A sensitivity of 95%, 89% and 87% was observed for each age group, respectively, using a cut-off value of γδ⁺>10%; and a sensitivity of 84%, 83% and 53%, respectively, for a cut-off value >14% (p=0.02; 50–69 vs. ≥70 years), with statistically significant difference between applying a cut-off of 10% or 14% (p=0.008). The median γδ⁺ count in the ≥70 years group was lower than in the other groups (p=0.014). Thus, maintaining a high specificity, results in adult populations suggest that sensitivity of the celiac lymphogram might decrease with increasing age above 70 years old.

As has been explained, the flow cytometry phenotyping of IELs shows a high specificity for CD diagnosis in adult patients and can be of diagnostic help in cases where diagnosis is not straightforward. A multicenter study was aimed on the diagnosis of seronegative villous atrophy (SNVA) (30). Sixty-seven patients with SNVA were included, 37 with CD and 30 with non-celiac villous atrophy. The celiac lymphogram was associated with a sensitivity of 87% and a specificity of 97% for CD. Among patients with a pre-test CD probability of 30%, post-test probabilities were 92% and 5% for positive and negative celiac lymphogram. The authors concluded that the celiac lymphogram was associated with a high level of diagnostic evidence either against or in favor of CD in patients with SNVA. The celiac lymphogram has also been shown to be a useful tool to consider Marsh 1 lesions as CD (29, 31, 32, 39–41). Using dermatitis herpetiformis as a model disease in which there are gluten-related symptoms despite of a non-atrophic enteropathy, even with negative celiac serology in a high percentage of patients, Popp and Mäki argue about the existence of a ‘celiac trait’, consisting in a Marsh 1 lesion, positive celiac genetics and increase in γδ⁺ cells, that should be identified and treated since they present clinical and histological remission after a GFD (42). Recently, a ‘low-grade celiac score’, including data on serology, intraepithelial lymphogram, HLA and histology, was derived statistically to identify patients likely to respond to a GFD and be diagnosed with Marsh 1 celiac enteropathy with a sensitivity of 86% and a specificity of 85%, and an AUC value of 0.91 (41). This score uses the γδ⁺ count, being the parameter that scores higher in seronegative patients. However, the diagnostic accuracy of the score was only maintained if there was a concomitant decrease of sCD3^- IELs, i.e., when the celiac lymphogram was present (43). The ‘low-grade celiac score’ is thus a semiquantitative approach to characterize the ‘celiac trait’ described by Popp and Mäki, which may be useful in clinical practice.

Persistence of increased γδ⁺ cells after gluten withdrawal opens the possibility of diagnosing CD in patients who started on a GFD over several years and without any persistent histological changes on microscopic examination of duodenal biopsies, and without the need for undergoing a gluten challenge. Several studies focused on this aspect with promising results, but the studies were performed on small samples of patients, with the follow-time after a GFD not always described or only describing changes of mean values before vs. after the diet (24, 33, 35–37). Further studies on this item will be acknowledged.

IEL flow cytometry requires obtaining one single duodenal biopsy from the second portion of the duodenum, which is obtained in the same procedure as for histological analysis and commonly processed immediately, giving the results in the next four hours. However, the sample can be stored at 4 °C in complete medium 24 h before its processing and analysis (44).

This methodology is highly reproducible. In Spanish multicenter studies (30, 31) all participating centers used similar gating strategies to select both sCD3⁻ and TCRγδ⁺ cells, which were measured as CD45⁺CD3⁻CD103⁺ and CD45⁺CD103⁺TCRγδ⁺, respectively, over the total CD45⁺CD103⁺ cells. The authors performed comparative studies with samples in parallel between the hospitals and the concordance for both sCD3⁻ and TCRγδ⁺ cells was almost 100% in terms of absolute percentages [appendix A in (31)]. There were minor differences in the cut-offs for TCRγδ⁺ and sCD3⁻ cells used in the different participating centers, which imply slight variations in sensitivity and specificity between them, that is, the lower the cut-off, the higher the sensitivity and the lower the specificity, and vice versa. Some centers consider specificity and others sensitivity as priority. Thus, the quantitative individual values were interpreted in that studies with a single cut-off value that allowed for high specificity, thereby maintaining good sensitivity.

Flow cytometry of intestinal IELs is the recommended method to identify the “aberrant” sCD3^- intracytoplasmic CD3 positive (icCD3⁺) IELs that expand in refractory CD type II (RCDII). Specifically, flow cytometry can precisely distinguish and enumerate the aberrant sCD3^- icCD3⁺ IEL subset and also the different CD8^- intraepithelial innate lymphoid cells and γδ⁺ cells, the latter being expanded in CD and RCD type I. A cut-off of 20% aberrant IELs has been proposed for a presumptive diagnosis of RCDII (45, 46).

More recent works include new flow cytometry markers in the assessment of this pre-malignant sCD3^- icCD3⁺ IEL subset. García-Hoz et al. showed that the increased intracytoplasmic fluorescent intensity (icCD3⁺) on the aberrant IEL subset predicts the risk of RCDII progression to enteropathy-associated T cell lymphoma (EATL), in good correlation with the presence of clonal TCR rearrangements (47). The authors conclude that the use of flow cytometry represents a powerful tool in the diagnosis and follow up of RCD: the enumeration of the sCD3^- icCD3⁺ aberrant IELs provides the initial presumptive diagnosis of RCDII, and the quantification of the icCD3 fluorescence intensity is of a prognostic value in determining the risk of clonality and progression to overt lymphoma. On the same line, Cheminant M et al. proposed NKp46, a Natural Killer receptor, as a useful biomarker for diagnostic and therapeutic stratification of gastrointestinal T lymphoproliferative disease (48).

A precise flow cytometry profiling of aberrant RCD IELs should help not only in the diagnosis but also in the clinical stratification and manage of patients, predicting survival and recommending therapies in this serious condition (49).

Detection of gluten-reactive T cells in peripheral blood remained elusive until CD patients on a GFD were challenged for three days aimed to identify CD relevant immunodominant gliadin epitopes (50). Maximum IFN-γ responses were observed on PBMCs collected six days after starting the 3-day gluten challenge by ELISPOT (Enzyme-linked immunospot) assay with overnight culture using a pool of antigenic peptides. This IFN-γ production mainly required CD4⁺ T cells and HLA-DQ, suggesting it was the result of HLA-DQ restricted CD4⁺ T cells specifically responding to gliadin. Particularly, the HLA-DQ2.5 immunodominant epitope was identified in α-gliadin 57-73, QLQPFPQPELPYPQPQS (p57–73 QE65). Abolishment of the IFN-γ response after depletion of the β7 integrin but not of αE (CD103), also suggested that those cells expressed the α4β7 integrin, which is associated with homing to the intestinal lamina propria (51). It was presumed that memory CD4⁺ T cells were causing the response. These cells most likely persist due to the presence of traces in the so-called gluten-free food or to the occasional gluten exposure that commonly occur in CD patients. It was assumed that they react to gluten presented by antigen presenting cells in mesenteric lymph nodes or gut associated lymphoid tissue, activate, clonally expand and via the thoracic duct reach peripheral blood, where they circulate until they migrate to the small intestine, which constitutes the effector site.

After identifying the gluten T cell epitopes and the transitory presence of gluten-specific T cells in blood, HLA-DQ2 tetramers could be generated to enable researchers to obtain directly the CD4⁺ T cells of interest (52, 53). Tetramers allow the study of antigen-specific T cell responses. They consist of multimerization using streptavidin of four biotinylated monomers of an antigenic peptide bound to MHC II molecules, which is necessary to obtain stable binding between them and the TCR. Streptavidin is coupled to a fluorophore for analysis by flow cytometry. Initially, DQ2-αI or DQ2-αII epitopes of α-gliadin were used as antigens. T cells detected with both DQ2-tetramer constructs were detected in all or the majority of patients. The presence of the gut-homing integrin β7 was confirmed by flow cytometry, but additionally, they were characterized as expressing CD28 and CD95. The state of differentiation was not homogeneous, but the markers observed suggested an effector memory phenotype (CD45RA⁺CD45RO^-CCR7^-CD27^-) or a transitory state towards that phenotype (52).

In 2013, Han et al. demonstrated using time-of-flight mass cytometry (CyTOF) that CD8⁺ and γδ⁺ T cells could be also detected in blood after the 3-day gluten challenge in parallel to the described CD4⁺ T cell response (6). These cells expressed the integrin receptor CD103β7, which is indicative of migration to the gut intraepithelial compartment, and the activation marker CD38. Exhaustive analyses including a high number of markers suggested that these cells corresponded to effector memory T cells. Interestingly, it was observed that the detected cells resemble intestinal T lymphocytes present in mucosal biopsies of CD patients (5).

After the first works above described, IFN-γ ELISPOT and HLA-tetramers were proposed as possible tools for CD diagnosis (52, 54, 55). The last one was presented as faster and independent of the function of the cells, detecting naïve as well as memory T cells, and adding the possibility of a deeper characterization using flow cytometry. Interestingly, both approaches would enable diagnosis of individuals on a GFD. However, Brottveit et al. alerted in 2011 that tetramer test is quite laborious and the tetramer reagents have limited stability, making hardly probable the widespread use of the tetramer test (54). They reported that the use of tetramers was 100% specific and 85% sensitive for HLA-DQ2.5⁺ CD.

The findings of Han et al. (6) were used by our research group to try a new test to diagnose CD and thus we showed that the study of activated gut-homing CD8⁺ T cells by flow cytometry offered 95% specificity and 97% sensitivity for detecting seropositive CD (39). This test has a presumed easy implementation in clinical practice.

The three methodological approaches were based on a similar gluten challenge and the increase of IFN-γ, tetramers or activated gut-homing CD8⁺ T cells from day 0 to 6.

Although the three T cell subsets studied (gluten-specific CD4⁺, CD8⁺ and γδ⁺ T cells) show a similar kinetics after the 3-day gluten challenge, peaking at day 6-8 after the start of the challenge (6), the magnitude of the response observed was quite variable among patients. This has practical consequences when evaluating these cell responses by flow cytometry with diagnostic purposes. The number of γδ⁺ T cells in blood is low, making difficult to visualize the subgroup of γδ⁺ T cells mobilized after gluten challenge in subjects with a low response. Gluten-specific CD4⁺ T cells can be also hard to detect. CD4⁺ T cells recognize gluten peptides when presented by APC carrying the CD associated HLA-DQ molecules. The degree of immunogenicity depends on the specific gluten derived peptide/s and the individual HLA genetics. Some gluten epitopes elicit strong responses in most patients, which are called immunodominant epitopes. These are used to identify the CD4⁺ T cells of interest directly using HLA tetramers or indirectly by IFN-γ ELISPOT. However, the term gluten involves a heterogeneous group of proteins, a broad range of epitopes have been discovered and it is very likely that some epitopes are not represented in the antigen preparation used for research. The most extensively used peptide p57–73 QE65, is generally the immunodominant epitope in CD present in wheat, but this is not so in all patients. Accordingly, in the published works there are several cases of patients who respond to deamidated gliadin but they do not to the specific peptide used. The use of restricted sequences (antigen stimulation) may reduce the response to be observed although a cocktail of immunodominant peptides is used (56). Moreover, the use of antigenic peptides also needs to determine the optimal antigen concentration to obtain maximum responses.

The adequacy of flow cytometry to identify gluten-induced T cells was also evidenced by the work of Christophersen et al., who showed that gluten reactive CD4⁺ T cells could be identified from biopsy based solely on phenotypic markers (56).

There are some a priori limitations common to the three suggested approaches. First, all require at least a 4-week GFD period prior to the 3-day gluten challenge. It has not been determined the requirement of a very strict GFD, but nowadays, GFD can be easily monitored by determining gluten immunogenic peptides (GIPs) in stool or urine. GIPs result of the hydrolysis of gluten and have been shown to be sensitive and specific markers to detect ongoing gluten intake (57). Second, there is a high variability in the response among individuals; however, this is not a problem if extensive studies are performed in order to know the range of response and proper cut-offs for diagnosis are established.