Quantitative histology-based classification system for assessment of the intestinal mucosal histological changes in patients with celiac disease

Prasenjit Das; Gaurav PS Gahlot; Alka Singh; Vandana Baloda; Ramakant Rawat; Anil K Verma; Gaurav Khanna; Maitrayee Roy; Archana George; Ashok Singh; Aasma Nalwa; Prashant Ramteke; Rajni Yadav; Vineet Ahuja; Vishnubhatla Sreenivas; Siddhartha Datta Gupta; Govind K Makharia

doi:10.5217/ir.2018.00167

Articles

Page Path: HOME > Intest Res > Volume 17(3); 2019 > Article

Original Article Inflammatory bowel diseases Quantitative histology-based classification system for assessment of the intestinal mucosal histological changes in patients with celiac disease: Prasenjit Das^1,, Gaurav PS Gahlot¹, Alka Singh², Vandana Baloda¹, Ramakant Rawat², Anil K Verma^2,§, Gaurav Khanna¹, Maitrayee Roy¹, Archana George¹, Ashok Singh¹, Aasma Nalwa¹, Prashant Ramteke¹, Rajni Yadav¹, Vineet Ahuja², Vishnubhatla Sreenivas³, Siddhartha Datta Gupta¹, Govind K Makharia^2,; Intestinal Research 2019;17(3):387-397.
DOI: https://doi.org/10.5217/ir.2018.00167
Published online: April 22, 2019

¹Department of Pathology, All India Institute of Medical Sciences, New Delhi, India

²Department of Gastroenterology and Human Nutrition, All India Institute of Medical Sciences, New Delhi, India

³Department of Biostatistics, All India Institute of Medical Sciences, New Delhi, India

Correspondence to Prasenjit Das, Department of Pathology, All India Institute of Medical Sciences, Ansari Nagar, New Delhi 110029, India. Tel: +91-11-26594979, Fax: +91-11-26588663, E-mail: prasenaiims@gmail.com

Co-Correspondence to Govind K Makharia, Department of Gastroenterology and Human Nutritions, All India Institute of Medical Sciences, Ansari Nagar, New Delhi 110029, India. Tel: +91-11-26564425, Fax: +91-11-26588663, E-mail: govindmakharia@gmail.com

§Current affiliation: Celiac Disease Research Laboratory, Department of Pediatrics, Università Politecnica delle Marche, Ancona, Italy.

• Received: December 18, 2018 • Revised: March 8, 2019 • Accepted: March 13, 2019

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

prev next

8,576 Views
183 Download
13 Web of Science
15 Crossref
15 Scopus

Full Article

Download PDF

Abstract
INTRODUCTION
METHODS
RESULTS
DISCUSSION
NOTES
Acknowledgments
REFERENCES

Abstract

Background/Aims
The existing histological classifications for the interpretation of small intestinal biopsies are based on qualitative parameters with high intraobserver and interobserver variations. We have developed and propose a quantitative histological classification system for the assessment of intestinal mucosal biopsies.
Methods
We performed a computer-assisted quantitative histological assessment of digital images of duodenal biopsies from 137 controls and 124 patients with celiac disease (CeD) (derivation cohort). From the receiver-operating curve analysis, followed by multivariate and logistic regression analyses, we identified parameters for differentiating control biopsies from those of the patients with CeD. We repeated the quantitative histological analysis in a validation cohort (105 controls and 120 patients with CeD). On the basis of the results, we propose a quantitative histological classification system. The new classification was compared with the existing histological classifications for interobserver and intraobserver agreements by a group of qualified pathologists.
Results
Among the histological parameters, intraepithelial lymphocyte count of ≥25/100 epithelial cells, adjusted villous height fold change of ≤0.7, and crypt depth-to-villous height ratio of ≥0.5 showed good discriminative power between the mucosal biopsies from the patients with CeD and those from the controls, with 90.3% sensitivity, 93.5% specificity, and 96.2% area under the curve. Among the existing histological classifications, our quantitative histological classification showed the highest intraobserver (69.7%–85.03%) and interobserver (24.6%–71.5%) agreements.
Conclusions
Quantitative assessment increases the reliability of the histological assessment of mucosal biopsies in patients with CeD. Such a classification system may be used for clinical trials in patients with CeD.
Keywords: Celiac disease; Intestine, small; Morphometry; Image analysis; Histology, quantitative

INTRODUCTION

The existing histological classification systems for assessment of duodenal mucosal changes in celiac disease (CeD) include original Marsh classification, modified Marsh classification [1,2], Corrazza and Villanacci classification (2005), and Ensari (2010) classification. Because of the qualitative nature of the parameters included in all the above classification system, there are low interobserver agreements amongst pathologists in most of the parameters except for the intraepithelial lymphocyte (IEL) count [3-6]. The lowest interobserver agreement has been observed in differentiating the modified Marsh grades 3a and 3b with these classification systems [5]. Apart from the visual microscopic assessment techniques, quantitative histological assessment (Q-histology) of intestinal mucosa has gained acceptance in the recent past [7-10]. Although the Q-histology was introduced in the 1970s, their use, however, remained limited to the research settings only [11]. Even with this limited use, the Q-histology assessment technique evolved over a period of time and became more straightforward from older methods of measuring the mucosal surface to volume ratio, mucosal index (mucosal height to length of muscularis mucosa) assessment [12] and computer card based drawing technique [8] to a more simplified recent protocol of using the villous height (Vh) to crypt depth (Cd) ratio [9]. However, despite its acceptance and simplification, Q-histology has yet not become the standard for clinical care. One of the reasons being a lack of a well defined Q-histological classification system based on quantifiable histology parameters [13].

The primary objective of this study was to develop an objective Q-histological classification system for assessing the mucosal changes in the duodenal biopsies from patients with CeD.

METHODS

1. Ethics Statement

All study procedures were approved by the Institutional Ethical Committee (IEC/NP-312/2012 & RP-07/2012, dated October 10, 2012) and performed in the study were by the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from the subjects included in this study.

2. Derivation Cohort

Duodenal biopsies from 137 controls including those with functional dyspepsia (n=50), subjects undergoing endoscopic retrograde cholangiopancreatography for common bile duct stone(s), as well as carriers of HBV infection were included. All of them underwent appropriate workup for the exclusion of any organic disease. All of them had negative IgA-anti-tissue transglutaminase antibody (anti-tTG Ab) titers and had normal endoscopic findings. Duodenal biopsies from 124 adult patients with CeD diagnosed as per the European Society of Pediatric Gastroenterology, Hepatology and Nutrition 2012 criteria, were also included [14]. During the execution of this work as anti-tTG Ab ELISA kits from different manufacturers were used; for uniformity, we derived anti-tTG antibody fold rise value, as the (tTG value in a patient ÷ normal cutoff value for that particular ELISA kit). During endoscopic examination, 4 to 6 mucosal biopsy fragments were obtained from the post-ampulla part of the duodenum.

3. Identifcation of Quantitative Histological Parameters for Diferentiation between Duodenal Biopsies from Controls and Patients with CeD

Digitized images of duodenal biopsies were analyzed by a pre-calibrated (in micrometer scale) computer assisted image analysis (CIA) technique based software supplied by the Media Cybernetics, version 6.1 (Rockville, MD, USA). The following parameters were measured in the oriented part of biopsy fragments: Vh, Cd, epithelial cell height (Eh) at the villous tip, villous area and number of IELs/100 epithelial cells (ECs) at the villous tip. The Q-histological assessment was performed on an average of 4 mucosal biopsy fragments from each subject, including at least 5 to 15 villi (depending on the biopsy orientation). CD3 staining of the IELs was done for uniformity by using the rabbit anti-human CD3 antibody (Spring Biosciences, Pleasanton, CA, US; 1:200) and was detected by a polymer-based secondary detection kit (DAKO, Carpinteria, CA, USA). The stained IELs were counted by using the manual tagging tool of the software and expressed as the number of IELs/100 ECs, after counting at least 2,000 ECs in the upper one-third of villi. The Vh and Cd were measured by using the distance measuring tool and the average Cd to Vh ratio was calculated. The histological parameters which can differentiate a biopsy from patients with CeD and controls were identified by receiver operating curve (ROC) analysis (Table 1) followed by multivariate and logistic regression analyses (Figs 1, 2). Principal investigators (P.D. and A.S.) performed the CIA-based analyses in the derivation cohort in a blinded manner.

4. Reprodicibility Assessment of the Quantitative Histological Parameters

We then recruited a new set of validation cohort including biopsies from 120 patients with CeD and 105 controls. The Q-histological parameters found relevant in the derivation cohort were assessed in the validation cohort (Table 2).

5. Formation of a New Quantitative Histological Classification

A new Q-histological classification system for assessment of duodenal biopsies was derived based on the parameters which were found significant in both derivation and validation cohorts in differentiating mucosal biopsies from patients with CeD and controls (Table 3).

6. Validation of the New Quantitative Histological Classification System

The diagnostic sensitivity and specificity of proposed Q-histology classification were examined both in the derivation and validation cohorts separately, with and without using the anti-tTG antibody titers.

Intraobserver and interobserver validation: 4 qualified histopathologists (G.P.G., M.R., G.K., and A.N.), who were not involved in the derivation process, participated in the validation process in a blinded manner. A set of randomly selected 50 digitized histological slides of duodenal biopsies, including both controls and patients with CeD were handed over for Q-histological reporting using the CIA-based software. The participants were allotted different time slots on a single desktop computer to exclude technical bias. Each observer performed quantitative analysis twice, keeping a gap of 2 weeks in-between. Between the 2 cycles, the chronology of the digitized slides was altered. They were also asked to perform light microscopic grading of biopsies separately using the modified Marsh Oberhuber classification, original Marsh classification, Corrazza and Villanacci classification, and Ensari classifications in both the occasions. The interobserver and intraobserver agreements among the observers were estimated and were compared with those with the existing light microscopic classifications.

7. Statistical Analysis

A Stata 11 software (StataCorp LP, College Station, TX, USA) was used. The mean±SD of all parameters was calculated separately. For identifying histological parameters with good discriminative power, 2-sample t-test with equal variances was used, followed by 2-sample Wilcoxon signed-rank sum (Mann-Whitney) test. Diagnostic sensitivity and specificities of each parameter were calculated by ROC analysis, followed by univariate and multivariate analyses. Logistic regression analysis was performed with the calculation of OR and 95% CIs for each parameter. The intraobserver and interobserver variations were analyzed using McNemar’s test and weighted kappa analyses. A P-value <0.05 was taken as statistically significant.

RESULTS

1. Characteristics of Subjects Included in the Derivation Cohort

In the derivation cohort, duodenal biopsies from 261 subjects including 137 adult controls (mean age 32±10.9 years, males 108) and 124 adult patients with CeD (mean age 26.5±13.5 years, males 93) were included. The anti-tTG antibody fold in the control group and patients with CeD were 0.4±0.6 and 8.5±8, respectively.

2. Quantifiable Histological Parameters Assessed in the Derivation Cohort

1) IEL Counting

The IEL count in duodenal biopsies from patients with CeD was higher than that in controls (45.9±19.9 vs. 13.4±8.1, P< 0.001). IEL count ≥25/100 ECs was found to optimally differentiate between biopsies from patients with CeD and control with 88% sensitivity and 87% specificity (AUC, 93.7%; 95% CI, 91%–96%) (Table 1, Fig. 1). When the cutoff of IEL increased to ≥30/100 ECs (as per the Marsh classification), the sensitivity reduced to 83%, and specificity increased to 94%. When the IEL cutoff was considered as ≥40/100 ECs, the sensitivity dropped to only 56%, but specificity increased to 99%. The odds of a suspected case being CeD was 49.3 (P<0.001) if the IEL count was ≥25 IEL/100 ECs (Table 1, Fig. 1).

2) Villous Height

Vh was significantly lower in patients with CeD, than in control biopsies (484±212 μm [170–1,520 μm] vs. 2,108±81 μm [366–3,880 μm], P<0.001). As determined by the ROC analysis, Vh cutoff <335 μm was found to have good discriminative power between the duodenal biopsies of patients with CeD and controls (sensitivity and specificity of 91% each [AUC 96.4%]) (Fig 1). The odds of a suspected case to have CeD was 97.1 (95% CI, 41.8–225.7; P<0.001), if Vh was <335 μm (Table 1).

3) Crypt Depth

There was no significant difference in the Cd in the mucosal biopsies obtained from controls and patients with CeD (255± 124 μm [range, 108–865 μm] vs. 238±86 μm [range, 75–490 μm], P=0.09) (Table 1).

4) Cd to Vh Ratio

The mean Cd to Vh ratio in controls was 0.56±0.45 (range, 0.25–5.08) in comparison to 1.45±1.3 (range, 0.32–12.1) in patients with CeD (P<0.001). Using the Cd:Vh ratio cutoff >0.5 (1:2), the biopsies from controls and patients were classified correctly with a sensitivity of 95.1% and specificity of 53.3%. A Cd:Vh ratio cutoff ≥0.66 (1:1.5) identified from the ROC analysis showed the highest sensitivity of 81% and specificity of 80% (AUC 88%) (Fig. 1). As the lowest estimated Cd:Vh in the control group of biopsies was 0.25, we then examined the discriminative power of Cd:Vh >0.25 (1:4); while the sensitivity reached to 100%, specificity became nil. Our next objective was to examine the discriminative power of Cd:Vh >0.33 (1:3), as was described in the Marsh classification. With this cutoff, biopsies from patients with CeD and controls were classified with 99% sensitivity, but specificity was only 11%. Hence, the highest discriminative sensitivity and specificity was noticed when the pathological cutoff of Cd:Vh considered ≥0.5 (1:2) (Fig. 1).

5) Surface Eh

Mucosal Eh was 37±6.4 μm (range, 18–58 μm) in biopsies from patients with CeD and 35±1.9 μm (range, 17–177 μm) in controls (P<0.001). On ROC analysis, Eh of ≥35.4 μm was found to have a low discriminative power in differentiating biopsies from controls and that of patients with CeD (sensitivity, 63%; specificity, 60%; AUC, 66.1% [95% CI, 59%–72%]; OR, 2.55) (Table 1).

6) Villous Area

Duodenal biopsies from patients with CeD were found to have short and stubby villi with a higher overall villous area (943.15±436.2 μm²), in comparison to the biopsies from controls (829.5±745.5 μm²) (P=0.007). Villous area ≥750 μm² was found to discriminate biopsies from patients with CeD and controls with a sensitivity and specificity of 66% each, AUC 65.6%, and odds of a suspected case having CeD was 3.94 (P<0.001) (Table 1, Fig. 1).

3. Assessment of Quantifiable Histological Parameters in a Validation Cohort

The CIA-based Q-histological analyses were repeated in a validation cohort including duodenal biopsies from 225 subjects including 105 controls (mean age, 33.8±9.6 years) and 120 patients with CeD (mean age, 28.5±12 years) to re-check the reproducibility of the histological parameters (Tables 1, 2). While, the pathological cutoffs for the Eh and villous area differed widely (cutoff of Eh and cutoff of villous area in derivation cohort were ≥35.4 μm and ≥750 μm²; cutoff of Eh and cutoff of villous area in validation cohort were ≥29 μm and <1,330 μm², respectively), rest of the histological parameters, as Vh, Cd, Cd:Vd correlated well between these cohorts (Tables 1, 2).

4. Univariate and Multivariate Analysis of the Q-Histological Parameters

Based on univariate analysis of the results of the derivation cohort parameters such as IEL count ≥25/100 ECs, Vh <335 μm, Cd:Vh ratio ≥0.66, villous area ≥750 μm² and mucosal Eh of ≥35.4 μm were found to be significant in the differentiation of duodenal mucosal biopsies from controls and patients with CeD with sensitivity and specificity of 62.9% and 90.4%, respectively (Table 1). Of these, only 3 parameters, namely Vh <335 μm, IEL count ≥25/100 ECs, and Cd:Vh ratio ≥0.66, were found statistically significant on multivariable analysis. Finally, these 3 histological parameters were chosen for inclusion in the Q-histology based classification.

5. Logistic Regression Analysis of the Q-Histological Parameters

Based on the coefficients of logistic regression analysis, a histological scoring system was formulated to check their utility for differentiationg duodenal biopsies from patients with CeD and controls. The scoring system was as follows:

( IEL count ≥25/100 ECs×3.9)+(Vh <335 μm×4.8)+(Cd:Vh ratio ≥0.66×2.9)

Total score varied from 0 to 11. This scoring system was then re-applied on the raw data of derivation cohort including duodenal biopsies from 261 subjects. A score of ≥6, was found to discriminate the duodenal biopsies from patients with CeD and controls with a sensitivity of 90.3% (83.7%–94.9%), specificity of 93.5% (87.7%–97.2%), positive predictive value (PPV) 93.3% (87.3%–97.1%), negative predictive value 90.6% (84.2%–95.1%) and AUC 96.2%. In controls, the score was <6 (P< 0.001). These 3 Q-histological parameters together showed an area under ROC of 97.4% for discriminating between duodenal mucosal biopsies from patients with CeD and controls.

6. Modification of Histological Parameters to Exclude Technical Confounding Factors

While measuring the length between 2 points with the CIA-based software takes into account pixel density per inch, the pixel resolution of the camera sensor or the computer screen used for image analyses can alter the outcome. Hence, we decided to use a single camera for digitization and one computer terminal for CIA-based analysis. However, as these factors can vary among different set-ups, we modified the absolute cutoff value for Vh to Vh fold change by dividing the mean Vh in duodenal biopsies from patients by the average Vh in control biopsies. The IEL counting was based on manual tagging tool of the software, and the Cd:Vh ratio required no further adjustment. Hence, finally, the following modified histological parameters were incorporated in the proposed Q-histological classification system: IEL count ≥25/100 ECs, adjusted Vh fold change ≤0.7, Cd:Vh ratio ≥0.5.

7. Proposition of a Q-Histological Classification System

Based on the above mentioned adjusted parameters, we propose a new Q-histological classification system for classifying the changes in duodenal biopsies. We classified the biopsies to be normal (type 0) if IELs count is <25/100 ECs. We classified abnormalities in the biopsies as follows: type 1 or early enteropathic changes: IEL count ≥25/100 ECs with normal Vh fold change >0.7 and Cd:Vh ratio <0.5; type 2 or intermediate enteropathic changes: IEL count ≥25/100 ECs along with abnormality of any one of the other 2 Q-histology parameters, such as the Vh fold change ≤0.7 ‘OR’ the Cd:Vh ratio ≥0.5; type 3 or advanced enteropathic changes: IEL count ≥25/100 ECs along with Vh fold change ≤0.7, as well as the Cd:Vh ratio ≥0.5 (Table 3, Fig. 2).

8. Validation of the Newly Proposed Q-histological Classification System

The newly proposed Q-histological classification was reapplied on the raw data files of the derivation and validation cohorts separately. Based on the defined criteria, of 323 duodenal biopsies in the derivation cohort, 142 (43.9%) were classified as Q-histology type 0 (including 120 controls and 22 from patients with CeD), 11 (3.4%) as Q-histology type 1 (including 7 controls and 4 from patients with CeD), 131 (40.5%) as Q-histology type 2 (including 10 controls and 121 from patients with CeD) and 39 (12.07%) as Q-histology type 3 (including 1 control and 38 from patients with CeD). The sensitivity and specificities of the Q-histology type 1, type 2 and type 3 changes were as follows: type 1: 15.4% and 94.5%, which when combined with the anti-tTG Ab data became 16.7% and 99.2%; type 2: 84.6% and 92.3%, which when combined with tTG fold rise value became 82.1% and 99.2%; type 3: 63.3% and 99.2%, which when combined with tTG value became 64.3% and 100%, respectively. When both the serum tTG titre was high (both anti-tTG fold rise value >1 and >10 times above normal were considered separately) and either type 1, type 2 or type 3 Q-histological changes, none of the biopsies from patients with CeD was wrongly classified as biopsies from controls.

Of duodenal biopsies from 225 patients in the validation cohort (105 controls and 120 CeD), 108 biopsies (47.7%) were classified as Q-histology type 0 (all were from controls), 6 (2.6%) as Q-histology type 1 (all 6 were biopsies from patients with CeD), 33 (14.6%) as Q-histology type 2 (all from patients with CeD), and 78 biopsies (34.9%) as Q-histology type 3 (all from patients with CeD). When both Q-histology and serum tTG titres (both anti-tTG fold rise values >1 and >10 were considered) were abnormal, none of the disease biopsies were classified as controls in the validation cohort. When both serum anti-tTG titre was normal and Q-histology was 0, none of the biopsies were classified as biopsies from CeD. In 3 biopsies from patients with CeD, despite being serum anti-tTG titre high, Q-histology type was 0. Similarly, in 18 control duodenal biopsies with normal serum anti-tTG titer Q-histology analysis showed type 1 changes. Overall sensitivities and specificities of Q-histology type 1, type 2 and type 3 changes for diagnosing CeD were 66.7% and 100%; 91.7% and 100%; 96.3%, and 100%, respectively. Adding the tTG fold rise values did not alter these sensitivities and specificities further. Overall PPV of Q-histology type 2 changes were 92.6% and 100% in the derivation, and in validation cohorts respectively. All these findings show that Q-histological type 2 and 3 changes have good sensitivities, specificities as well as PPV in identifying abnormal intestinal biopsies, which become further specific for CeD when correlated with serum tTG values.

9. Reproducibility Assessment of the Q-Histology Classification System

The proposed Q-histology classification system showed the highest intraobserver agreement (69.7%–85.0%) and interobserver agreement (24.6%–71.5%) in comparison to other existing histological classification systems (Table 4).

DISCUSSION

After analyzing duodenal biopsies from 486 subjects including both controls and patients with CeD, we found that IEL count ≥25/100 ECs, Vh fold change ≤0.7, Cd:Vh ratio ≥0.5 are most reliable Q-histological parameters which can differentiate duodenal biopsies from patients with CeD and controls. The sensitivity, specificity, and PPV of the Q-histological classification system and better interobserver and intraobserver agreements suggest that Q-histology based classification is a better method for the assessment and documentation of duodenal mucosal biopsies [15-18]. We typed the duodenal mucosal histological changes based on the presence or absence of Q-histological parameters, as early enteropathy changes (type 1), intermediate enteropathy changes (type 2) and advanced enteropathy changes (type 3). This typing method of duodenal histological changes was however not based on the severity of Q-histological parameters, as that does not have any impact on patient management and the changes may not follow a steady pattern in reality. In this Q-histology classification system, the type 1 change only corresponds to increased IELs at villous tips; while in type 2 and type 3 changes, there will be either change in Vh or Vh:Cd or both. This way of graded typing of the histological changes will increase the likelihood of identifying the diseases biopsies more confidently, as we have seen in this study for diagnosing CeD [19-25]. Though, we performed CD3 immunostaining for highlighting the IELs (Fig 2B); in routine, it is not essential [25]. An experienced pathologist can identify the IELs and nuclei of ECs differently and can count easily. The other included parameters as Vh and Cd:Vh can either be measured by eye-piece reticules or with CIA-based software, more easily than previously described quantitation methods, as villous tortuosity and villous area [26,27], and spatiotemporal mucosal changes by in-vivo capsule endoscopy [3,8,9,28].

The Q-histology may prove to be better than the existing qualitative classification systems because it includes quantifiable histological parameters, hence can generate reliable and comparable data among research groups from different geographical regions and its ability to provide uniformity in the interpretation. As a long time is required for visual appreciation of histological improvement on the qualitative scale, another potential application of the Q-histology based classification system is likely to be assessment of even minor changes in the histology that occur over a shorter period of time especially during clinical trial of newer drugs such as as latiglutenase, zonulin inhibitors, therapeutic vaccines, etc [29]. Applicability of the modified Marsh classification is not optimum for interpreting follow-up biopsies from patients with CeD on GFD, because visible changes in Vh or Cd:Vh are slow to take place and minor changes are difficult to determine visually. Q-histology in such scenario can accurately measure any improvement or deterioration of Vh or Cd:Vh, which can be compared with baseline biopsy data. The outcome of the Q-histology assessment would depend on the exclusion of the possible confounding factors, such as the technical expertise of operators, adequacy of sampling, biopsy orientation, pixel densities of the digital camera sensor and the computer screen [29-33]. Also, the Q-histological assessment may be more labor intensive than the visual light microscopic analysis. While the light microscopic evaluation with use of eye-piece reticule takes around 5 minutes, the CIA-based assessment may take approximately 20 minutes for analyzing a digitized slide. We took care of the possible confounding factors and tried to justify why despite being more labor intensive Q-histology can be valuable.

While there is a merit of this study, there exits a few limitations. Recruitment of entirely healthy subjects for the derivation of normative histological characteristics was not ethically justifiable, and hence we recruited subjects with functional dyspepsia, HBV carriers and patients undergoing endoscopic retrograde cholangiopancreatography as controls.

In conclusion, this study demonstrates that the Q-histological assessment of duodenal biopsy can bring uniformity and reliability in the histological evaluation of mucosal alterations in patients with CeD. The Q-histology classification system has a potential for its use in clinical practice and in assessing histological response over a short period as required in many ongoing clinical trials.

NOTES

FINANCIAL SUPPORT

The work was supported by the Intramural Research Grant, All India Institute of Medical Sciences, New Delhi, India (grant number: A130).

CONFLICT OF INTEREST

No potential conflict of interest relevant to this article was reported.

AUTHOR CONTRIBUTION

Das P conceived idea, received the grant, executed the study & is the guarantor of the article. Makharia GK managed all patients and controls, conceived the idea, and performed the study. Gupta SD designed and executed the study. Ahuja V supervised the study. Gahlot GP, Singh A, Yadav R, Baloda V, Rawat R, Verma A recruited the patients, collected data and work execution. Nalwa A, Roy M, George A, Ashok S, Ramteke P, Khanna G participated as an observer. Sreenivas V statistical analyses. Approval of final manuscript: all authors.

Acknowledgments

We acknowledge the contribution of staffs of Histopathology Research Laboratory and Celiac Disease Clinic.

Fig. 1.

(A) Showing receiver operating characteristic (ROC) considering pathological cutoff of the IEL count ≥25/100 ECs for differentiating duodenal biopsies from patients with celiac disease and controls, with area under the ROC (AUC) of 93.7% (95% CI, 91.1%–96.1%). (B) ROC considering pathological cutoff of Vh <335 µm, with AUC 96.4% (95% CI, 94.3%–98.5%). (C) ROC considering the pathological cutoff for Cd:Vh ratio ≥0.66, with AUC 88.2% (95% CI, 84.1%–92.3%). (D) ROC considering villous area cutoff at ≥750 µm² , with AUC 65.6% (95% CI, 58.8%–72.4%). IEL, intraepithelial lymphocytes; ECs, epithelial cells; Vh, villous height; Cd, crypt depth.

Fig. 2.

(A) Photomicrograph shows a villous tip with increased intraepithelial lymphocytes (IELs; arrows) (H&E, ×200). (B) The method of manual tagging of anti-CD3 positive IELs by using the computer assisted image analysis-based software mentioned (immunostaining, ×100). (C) The method of measuring the mucosal epithelial heights at the villous tips (H&E, ×400). (D) Method of measuring the Cd:Vh and the villous areas (H&E, ×10). (E) A duodenal biopsy with increased IELs and Cd:Vh >0.5, hence, categorized as type 2 changes (H&E, ×100). (F) IEL count ≥25/100 ECs, Vh fold change ≤0.7 and Cd:Vh ratio ≥0.5, hence, categorized as type 3 changes (H&E, ×100). The black lines in the D-F are representing the imaginary shoulder of these biopsy fragments. Cd, crypt depth; Vh, villous height; ECs, epithelial cells.

Table 1.

Results of Quantitative Histological Assessment Performed with All Histological Parameters in the Derivation Cohort for Differentiation between Duodenal Biopsies from Controls and Patients with CeD

Histological parameter	Patients with CeD (n=124)	Control (n=137)	P-value	AUROC (%)	OR	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
Epithelial cell height (µm)	37 ± 6.4	35 ± 1.9		66.1 (59-72)	1.00	63.0 (54.0–71.4)	60.0 (51.2–68.3)	59.7 (50.9–68.1)	63.3 (54.3–71.6)
≥ 35.4	62.9	40.0	< 0.001		2.5 (1.5–4.2)
< 35.4	37.0	60.0	< 0.001
Villous area (µm²)	943.1 ± 436.2	829.5 ± 745.5		65.6 (58-72)	1.00	66.1 (57.1–74.4)	66.9 (58.3–74.7)	64.6 (55.6–72.8)	68.4 (59.8–76.2)
≥ 750	66.1	33.1	0.007		3.9 (2.3–6.6)
< 750	33.8	66.9	< 0.001
IEL count	45.9 ± 19.9	13.4 ± 8.1		93.7 (91-96)	1.00	88.1 (82.6–92.4)	87.0 (80.2–92.1)	90.1 (84.7–94.0)	84.5 (77.5–90)
≥ 25/100 ECs	88.1	13.0	< 0.001		49.3 (25.3–96.1)
< 25/100 ECs	11.8	86.9	< 0.001
Vh (µm)	484 ± 212	2,108 ± 81		96.4 (94-98)	1.00	91.1 (84.7–95.5)	90.4 (84.2–94.8)	89.7 (83–94.4)	91.8 (85.8–95.8)
< 335	91.1	9.5	< 0.001		97.1 (41.8–225.7)
≥ 335	8.9	90.4	< 0.001
Cd (µm)	238 ± 86	255 ± 124		49.6 (42-56)	1.00	51.6 (42.5–60.7)	51.5 (42.8–60.1)	49.2 (40.4–58.1)	53.8 (44.9–62.6)
≥ 215	51.6	48.3	0.090		1.13 (0.6–1.8)
< 215	48.3	51.4	0.610
Cd:Vh ratio	1.45 ± 1.3	0.56 ± 0.4		88.0 (84-92)	1.00	81.5 (73.5–87.9)	80.0 (72.3–86.4)	78.9 (70.8–85.6)	82.4 (74.8–88.5)
≥ 0.66 (1:1.5)	81.4	20.0	< 0.001		17.5 (9.4–32.6)
< 0.66	18.5	80.0	< 0.001

Values are presented as mean±SD or percent. Parentheses show 95% CI.

CeD, celiac disease; AUROC, area under the receiver operating characteristic; PPV, positive predictive value; NPV, negative predictive value; IEL, intraepithelial lymphocytes; ECs, epithelial cells; Cd, crypt depth; Vh, villous height.

Table 2.

Results of Quantitative Histological Assessment Performed with All Histological Parameters in the Validation Cohort for Differentiation between Duodenal Biopsies from Controls and Patients with CeD

Histological parameter	Patients with CeD (n=120)	Control (n=105)	P-value	AUROC (%)	OR	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
Epithelial cell height ≥ 29 μm	34.7 ± 4.6	24.6 ± 2.9	< 0.001	95.7 (93.0-98.0)	45.9 (21.2–100.2)	87.6 (80.4–92.9)	86.7 (78.6–92.5)	88.3 (81.2–93.5)	85.8 (77.7–91.9)
Villous area < 1,330 μm²	1,198.7 ± 267.9	1,395.9 ± 293.7	< 0.300	46.0 (38.0-54.0)	1.1 (0.6–1.87)	51.2 (42–60.4)	51.4 (41.5–61.3)	54.9 (45.2–64.2)	47.8 (38.3–57.4)
IEL count ≥ 25/100 ECs	36.1 ± 20.2	7.9 ± 3.5	< 0.001	99.9 (99.0-100.0)	51.2 (24.8–99.1)	97.5 (92.9–99.5)	100.0 (96.5–100.0)	100.0 (96.9–100.0)	97.2 (92.1–99.4)
Vh < 365 μm	379.1 ± 116.7	508.3 ± 130.9	< 0.001	86.2 (41.8-225.7)	39.4 (18.4–84.1)	86.8 (79.4–92.2)	85.7 (77.5–91.8)	87.5 (80.2–92.8)	84.9 (76.6–91.1)
Cd:Vh ratio ≥ 0.66 (1:1.5)	0.96 ± 0.7	0.34 ± 0.3	< 0.001	98.7 (97.0-100.0)	1,492.0 (285.0–7,539.0)	96.7 (91.8–99.1)	98.1 (93.2–99.8)	98.3 (94.1–99.8)	96.2 (90.6–99.0)

Values are presented as mean±SD. Parentheses show 95% CI.

Table 3.

The Proposed Quantitative Histological Classification for Assessment of Duodenal Mucosal Biopsies

Class	Computer image analysis based criteria
Type 0	IEL count < 25/100 ECs
Type 1	IEL count ≥ 25/100 ECs+Vh fold change > 0.7+Cd:Vh < 0.5
Type 2	IEL count ≥ 25/100 ECs+Vh fold change ≤ 0.7 'OR' Cd:Vh ratio ≥ 0.5
Type 3	IEL count ≥ 25/100 ECs+Vh fold change ≤ 0.7+Cd:Vh ratio ≥ 0.5

IEL, intraepithelial lymphocytes; EC, epithelial cells; Cd, crypt depth; Vh, villous height.

Notes: (1) Villous height fold change: calculated by dividing the villous height in a case ÷ mean villous height in the control population. (2) Type 1 Q histology refers to–early enteropathy changes; type 2–intermediate enteropathy changes; type 3–advanced enteropathy changes. (3) A duodenal biopsy showing IEL count <25/100 ECs irrespective of normal Vh fold change and Cd to Vh ratio or any abnormality of the latter 2 histological parameters should be assigned Q-histology type 0.

Table 4.

Intraobserver and Interobserver Agreements Using Histological Classification Systems

	Intraobserver agreements (based on weighted κ-value)	Interobserver agreements (based on weighted κ-value)
Histological classification systems
Marsh classification [2]	39.7 to 60.9	12.9 to 48.2
Modified Marsh [1]	46.6 to 63.7	28.3 to 48.7
Corazza et al. [5]	54.6 to 84.6	28.04 to 42.9
Ensari [6]	54.5 to 64.5	28.04 to 42.9
Proposed quantitative histological classification	69.7 to 85.0	24.6 to 71.5

All unit of data is percentage.

REFERENCES

1. Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac disease: time for a standardized report scheme for pathologists. Eur J Gastroenterol Hepatol 1999;11:1185–1194.Article PubMed
2. Marsh MN. Gluten, major histocompatibility complex, and the small intestine. A molecular and immunobiologic approach to the spectrum of gluten sensitivity (‘celiac sprue’). Gastroenterology 1992;102:330–354.Article PubMed
3. Revised criteria for diagnosis of coeliac disease. Report of Working Group of European Society of Paediatric Gastroenterology and Nutrition. Arch Dis Child 1990;65:909–911.Article PubMed PMC
4. Madan M, Piplani S, Sharma M, Bhasin TS, Manjari M, Kaur H. Celiac disease: an assessment of subjective variation and diagnostic reproducibility of the various classification systems. Glob J Med Res 2015;15(No 1-C): 882.
5. Corazza GR, Villanacci V, Zambelli C, et al. Comparison of the interobserver reproducibility with different histologic criteria used in celiac disease. Clin Gastroenterol Hepatol 2007;5:838–843.Article PubMed
6. Ensari A. Gluten-sensitive enteropathy (celiac disease): controversies in diagnosis and classification. Arch Pathol Lab Med 2010;134:826–836.Article PubMed PDF
7. Cummins AG, Alexander BG, Chung A, et al. Morphometric evaluation of duodenal biopsies in celiac disease. Am J Gastroenterol 2011;106:145–150.Article PubMed PDF
8. Meinhard EA, Wadbrook DG, Risdon RA. Computer card morphometry of jejunal biopsies in childhood coeliac disease. J Clin Pathol 1975;28:85–93.Article PubMed PMC
9. Taavela J, Koskinen O, Huhtala H, et al. Validation of morphometric analyses of small-intestinal biopsy readouts in celiac disease. PLoS One 2013;8:e76163.Article PubMed PMC
10. Kuitunen P, Kosnai I, Savilahti E. Morphometric study of the jejunal mucosa in various childhood enteropathies with special reference to intraepithelial lymphocytes. J Pediatr Gastroenterol Nutr 1982;1:525–531.Article PubMed
11. Ciclitira PJ, Evans DJ, Fagg NL, Lennox ES, Dowling RH. Clinical testing of gliadin fractions in coeliac patients. Clin Sci (Lond) 1984;66:357–364.Article PubMed
12. Tulloh EA, Baylis JM, Challacombe DN. Automated analysis of morphological change in the duodenal mucosa of children with coeliac disease. Arch Dis Child 1981;56:860–863.Article PubMed PMC
13. Ghoshal UC, Gwee KA. Post-infectious IBS, tropical sprue and small intestinal bacterial overgrowth: the missing link. Nat Rev Gastroenterol Hepatol 2017;14:435–441.Article PubMed PDF
14. Husby S, Koletzko S, Korponay-Szabó IR, et al. European Society for Pediatric Gastroenterology, Hepatology, and Nutrition guidelines for the diagnosis of coeliac disease. J Pediatr Gastroenterol Nutr 2012;54:136–160.Article PubMed
15. Adelman DC, Murray J, Wu TT, Mäki M, Green PH, Kelly CP. Measuring change in small intestinal histology in patients with celiac disease. Am J Gastroenterol 2018;113:339–347.Article PubMed
16. Elli L, Branchi F, Sidhu R, et al. Small bowel villous atrophy: celiac disease and beyond. Expert Rev Gastroenterol Hepatol 2017;11:125–138.Article PubMed
17. Singh P, Kurray L, Agnihotri A, et al. Titers of anti-tissue transglutaminase antibody correlate well with severity of villous abnormalities in celiac disease. J Clin Gastroenterol 2015;49:212–217.Article PubMed
18. Rahmati A, Shakeri R, Sohrabi M, et al. Correlation of tissue transglutaminase antibody with duodenal histologic marsh grading. Middle East J Dig Dis 2014;6:131–136.PubMed PMC
19. Hammer ST, Greenson JK. The clinical significance of duodenal lymphocytosis with normal villus architecture. Arch Pathol Lab Med 2013;137:1216–1219.Article PubMed
20. Trejdosiewicz LK. What is the role of human intestinal intraepithelial lymphocytes? Clin Exp Immunol 1993;94:395–397.Article PubMed PMC PDF
21. Veress B, Franzén L, Bodin L, Borch K. Duodenal intraepithelial lymphocyte-count revisited. Scand J Gastroenterol 2004;39:138–144.Article PubMed
22. Järvinen TT, Collin P, Rasmussen M, et al. Villous tip intraepithelial lymphocytes as markers of early-stage coeliac disease. Scand J Gastroenterol 2004;39:428–433.Article PubMed
23. Shalimar DM, Das P, Sreenivas V, Gupta SD, Panda SK, Makharia GK. Mechanism of villous atrophy in celiac disease: role of apoptosis and epithelial regeneration. Arch Pathol Lab Med 2013;137:1262–1269.Article PubMed
24. Ghosal UC, Das P. Diagnosis of celiac disease. In: Makharia GK. eds. Handbook of celiac disease. 1st ed. New Delhi: Kontentworx Publications, 2015:63–85.
25. Švajdler M, Daum O, Rychlý B. Diagnosing celiac disease: role of the pathologists. Int J Celiac Dis 2014;2:70–75.Article
26. Rosekrans PC, Meijer CJ, Polanco I, Mearin ML, van der Wal AM, Lindeman J. Long-term morphological and immunohistochemical observations on biopsy specimens of small intestine from children with gluten-sensitive enteropathy. J Clin Pathol 1981;34:138–144.Article PubMed PMC
27. Boruah D, Bhatia JK, Kamal KD, Malik A. Morphometric changes in jejunal mucosa in HIV positive patients presenting with enteropathy: an Indian study. Ann Pathol Lab Med 2017;4:A379–A387.Article
28. Hegenbart S, Uhl A, Vécsei A. Survey on computer aided decision support for diagnosis of celiac disease. Comput Biol Med 2015;65:348–358.Article PubMed PMC
29. Hagendorn E, Whitney-Miller C, Huber A, Potts SJ. Development of a tissue image analysis algorithm for celiac drug development. In: Potts SJ, Eberhard DA, Wharton KA. eds. Methods in pharmacology and toxicology. New York: Springer, 2015:141–152.
30. Vécsei A, Amann G, Hegenbart S, Liedlgruber M, Uhl A. Automated Marsh-like classification of celiac disease in children using local texture operators. Comput Biol Med 2011;41:313–325.Article PubMed
31. Ciaccio EJ, Tennyson CA, Lewis SK, Krishnareddy S, Bhagat G, Green PH. Distinguishing patients with celiac disease by quantitative analysis of videocapsule endoscopy images. Comput Methods Programs Biomed 2010;100:39–48.Article PubMed
32. Gottlieb K, Dawson J, Hussain F, Murray JA. Development of drugs for celiac disease: review of endpoints for Phase 2 and 3 trials. Gastroenterol Rep (Oxf) 2015;3:91–102.Article PubMed PMC PDF
33. Murray JA, Kelly CP, Green PH, et al. No difference between latiglutenase and placebo in reducing villous atrophy or improving symptoms in patients with symptomatic celiac disease. Gastroenterology 2017;152:787–798.Article PubMed

Figure & Data

REFERENCES

Citations

Citations to this article as recorded by

Follow-up of Celiac Disease After Diagnosis
Luca Elli, Govind K. Makharia, Daniel A. Leffler, Lucia Scaramella, Georgia Malamut
Gastrointestinal Endoscopy Clinics of North America.2025;[Epub] CrossRef
Celiac Disease Deep Learning Image Classification Using Convolutional Neural Networks
Joaquim Carreras
Journal of Imaging.2024; 10(8): 200. CrossRef
CD, or not CD, that is the question: a digital interobserver agreement study in coeliac disease
James Denholm, Benjamin A Schreiber, Florian Jaeckle, Mike N Wicks, Emyr W Benbow, Tim S Bracey, James Y H Chan, Lorant Farkas, Eve Fryer, Kishore Gopalakrishnan, Caroline A Hughes, Kathryn J Kirkwood, Gerald Langman, Betania Mahler-Araujo, Raymond F T Mc
BMJ Open Gastroenterology.2024; 11(1): e001252. CrossRef
Role of Serology, Dietary Assessment, and Fecal Gluten Immunogenic Peptides for Predicting Histologic Recovery in Children with Celiac Disease
Keerthivasan Seetharaman, Sadhna Bhasin Lal, Kaushal Kishor Prasad, Yashwant Kumar, Alka Bhatia, Sunita Malhotra
Digestive Diseases and Sciences.2023; 68(2): 529. CrossRef
Usefulness of a double immunofluorescence technique for detection of intestinal tTG-IgA deposits in diabetic and non-diabetic children with celiac disease
Raghav Lal, Ranjeet Bhardwaj, Ranjana Walker Minz, Kaushal Kishore Prasad, Sadhna Lal, Devi Dayal, Yashwant Kumar
Pediatrics & Neonatology.2023; 64(4): 388. CrossRef
The global burden of coeliac disease: opportunities and challenges
Govind K. Makharia, Prashant Singh, Carlo Catassi, David S. Sanders, Daniel Leffler, Raja Affendi Raja Ali, Julio C. Bai
Nature Reviews Gastroenterology & Hepatology.2022; 19(5): 313. CrossRef
Gluten Induces Subtle Histological Changes in Duodenal Mucosa of Patients with Non-Coeliac Gluten Sensitivity: A Multicentre Study
Kamran Rostami, Arzu Ensari, Michael N. Marsh, Amitabh Srivastava, Vincenzo Villanacci, Antonio Carroccio, Hamid Asadzadeh Aghdaei, Julio C. Bai, Gabrio Bassotti, Gabriel Becheanu, Phoenix Bell, Camillo Di Bella, Anna Maria Bozzola, Moris Cadei, Giovanni
Nutrients.2022; 14(12): 2487. CrossRef
The Role of Nerve Fibers in the Tumor Immune Microenvironment of Solid Tumors
Sharia Hernandez, Alejandra G. Serrano, Luisa M. Solis Soto
Advanced Biology.2022;[Epub] CrossRef
Case Report: Morphologic and Functional Characteristics of Intestinal Mucosa in a Child With Short Bowel Syndrome After Treatment With Teduglutide: Evidence in Favor of GLP-2 Analog Safety
Enrico Costantino Falco, Antonella Lezo, Pierluigi Calvo, Caterina Rigazio, Anna Opramolla, Ludovica Verdun, Giovanna Cenacchi, Marianna Pellegrini, Marco Spada, Gabriella Canavese
Frontiers in Nutrition.2022;[Epub] CrossRef
Quantitative histology as a diagnostic tool for celiac disease in children and adolescents
Mateus M. Vargas, Ricardo Artigiani Neto, Vera L. Sdepanian
Annals of Diagnostic Pathology.2022; 61: 152031. CrossRef
Best practices of handling, processing, and interpretation of small intestinal biopsies for the diagnosis and management of celiac disease: A joint consensus of Indian association of pathologists and microbiologists and Indian society of gastroenterology
Prasenjit Das, Kim Vaiphei, AnjaliD Amarapurkar, Puja Sakhuja, Ritambhra Nada, RoopaRachel Paulose, Rachana Chaturvedi, Anuradha Sekaran, Usha Kini, Archana Rastogi, Niraj Kumari, Anna Pulimood, Mala Banerjee, Prateek Kinra, Lavleen Singh, AmarenderSingh
Indian Journal of Pathology and Microbiology.2021; 64(5): 8. CrossRef
Artificial intelligence in small intestinal diseases: Application and prospects
Yu Yang, Yu-Xuan Li, Ren-Qi Yao, Xiao-Hui Du, Chao Ren
World Journal of Gastroenterology.2021; 27(25): 3734. CrossRef
Opportunities and challenges in the management of celiac disease in Asia
Ashish Agarwal, Ashish Chauhan, Vineet Ahuja, Govind K Makharia
JGH Open.2020; 4(5): 795. CrossRef
Digital histology in celiac disease: A practice changer
Daniel Vasile Balaban, Mariana Jinga
Artificial Intelligence in Gastroenterology.2020; 1(1): 1. CrossRef
Digital histology in celiac disease: A practice changer
Daniel Vasile Balaban, Mariana Jinga
Artificial Intelligence in Gastroenterology.2020; 1(1): 1. CrossRef

PubReader
ePub Link

Cite

CITE: export Copy Download Format; Close

Download Citation

Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.

Format:

RIS — For EndNote, ProCite, RefWorks, and most other reference management software
BibTeX — For JabRef, BibDesk, and other BibTeX-specific software

Include:

Citation for the content below
Citation and abstract for the content below

Quantitative histology-based classification system for assessment of the intestinal mucosal histological changes in patients with celiac disease

Intest Res. 2019;17(3):387-397. Published online April 22, 2019

DOI: https://doi.org/10.5217/ir.2018.00167

XML Download

Figure

Quantitative histology-based classification system for assessment of the intestinal mucosal histological changes in patients with celiac disease

Fig. 1. (A) Showing receiver operating characteristic (ROC) considering pathological cutoff of the IEL count ≥25/100 ECs for differentiating duodenal biopsies from patients with celiac disease and controls, with area under the ROC (AUC) of 93.7% (95% CI, 91.1%–96.1%). (B) ROC considering pathological cutoff of Vh <335 µm, with AUC 96.4% (95% CI, 94.3%–98.5%). (C) ROC considering the pathological cutoff for Cd:Vh ratio ≥0.66, with AUC 88.2% (95% CI, 84.1%–92.3%). (D) ROC considering villous area cutoff at ≥750 µm2 , with AUC 65.6% (95% CI, 58.8%–72.4%). IEL, intraepithelial lymphocytes; ECs, epithelial cells; Vh, villous height; Cd, crypt depth.

Fig. 2. (A) Photomicrograph shows a villous tip with increased intraepithelial lymphocytes (IELs; arrows) (H&E, ×200). (B) The method of manual tagging of anti-CD3 positive IELs by using the computer assisted image analysis-based software mentioned (immunostaining, ×100). (C) The method of measuring the mucosal epithelial heights at the villous tips (H&E, ×400). (D) Method of measuring the Cd:Vh and the villous areas (H&E, ×10). (E) A duodenal biopsy with increased IELs and Cd:Vh >0.5, hence, categorized as type 2 changes (H&E, ×100). (F) IEL count ≥25/100 ECs, Vh fold change ≤0.7 and Cd:Vh ratio ≥0.5, hence, categorized as type 3 changes (H&E, ×100). The black lines in the D-F are representing the imaginary shoulder of these biopsy fragments. Cd, crypt depth; Vh, villous height; ECs, epithelial cells.

Fig. 1.

Fig. 2.

Quantitative histology-based classification system for assessment of the intestinal mucosal histological changes in patients with celiac disease

Histological parameter	Patients with CeD (n=124)	Control (n=137)	P-value	AUROC (%)	OR	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
Epithelial cell height (µm)	37 ± 6.4	35 ± 1.9		66.1 (59-72)	1.00	63.0 (54.0–71.4)	60.0 (51.2–68.3)	59.7 (50.9–68.1)	63.3 (54.3–71.6)
≥ 35.4	62.9	40.0	< 0.001		2.5 (1.5–4.2)
< 35.4	37.0	60.0	< 0.001
Villous area (µm²)	943.1 ± 436.2	829.5 ± 745.5		65.6 (58-72)	1.00	66.1 (57.1–74.4)	66.9 (58.3–74.7)	64.6 (55.6–72.8)	68.4 (59.8–76.2)
≥ 750	66.1	33.1	0.007		3.9 (2.3–6.6)
< 750	33.8	66.9	< 0.001
IEL count	45.9 ± 19.9	13.4 ± 8.1		93.7 (91-96)	1.00	88.1 (82.6–92.4)	87.0 (80.2–92.1)	90.1 (84.7–94.0)	84.5 (77.5–90)
≥ 25/100 ECs	88.1	13.0	< 0.001		49.3 (25.3–96.1)
< 25/100 ECs	11.8	86.9	< 0.001
Vh (µm)	484 ± 212	2,108 ± 81		96.4 (94-98)	1.00	91.1 (84.7–95.5)	90.4 (84.2–94.8)	89.7 (83–94.4)	91.8 (85.8–95.8)
< 335	91.1	9.5	< 0.001		97.1 (41.8–225.7)
≥ 335	8.9	90.4	< 0.001
Cd (µm)	238 ± 86	255 ± 124		49.6 (42-56)	1.00	51.6 (42.5–60.7)	51.5 (42.8–60.1)	49.2 (40.4–58.1)	53.8 (44.9–62.6)
≥ 215	51.6	48.3	0.090		1.13 (0.6–1.8)
< 215	48.3	51.4	0.610
Cd:Vh ratio	1.45 ± 1.3	0.56 ± 0.4		88.0 (84-92)	1.00	81.5 (73.5–87.9)	80.0 (72.3–86.4)	78.9 (70.8–85.6)	82.4 (74.8–88.5)
≥ 0.66 (1:1.5)	81.4	20.0	< 0.001		17.5 (9.4–32.6)
< 0.66	18.5	80.0	< 0.001

Histological parameter	Patients with CeD (n=120)	Control (n=105)	P-value	AUROC (%)	OR	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
Epithelial cell height ≥ 29 μm	34.7 ± 4.6	24.6 ± 2.9	< 0.001	95.7 (93.0-98.0)	45.9 (21.2–100.2)	87.6 (80.4–92.9)	86.7 (78.6–92.5)	88.3 (81.2–93.5)	85.8 (77.7–91.9)
Villous area < 1,330 μm²	1,198.7 ± 267.9	1,395.9 ± 293.7	< 0.300	46.0 (38.0-54.0)	1.1 (0.6–1.87)	51.2 (42–60.4)	51.4 (41.5–61.3)	54.9 (45.2–64.2)	47.8 (38.3–57.4)
IEL count ≥ 25/100 ECs	36.1 ± 20.2	7.9 ± 3.5	< 0.001	99.9 (99.0-100.0)	51.2 (24.8–99.1)	97.5 (92.9–99.5)	100.0 (96.5–100.0)	100.0 (96.9–100.0)	97.2 (92.1–99.4)
Vh < 365 μm	379.1 ± 116.7	508.3 ± 130.9	< 0.001	86.2 (41.8-225.7)	39.4 (18.4–84.1)	86.8 (79.4–92.2)	85.7 (77.5–91.8)	87.5 (80.2–92.8)	84.9 (76.6–91.1)
Cd:Vh ratio ≥ 0.66 (1:1.5)	0.96 ± 0.7	0.34 ± 0.3	< 0.001	98.7 (97.0-100.0)	1,492.0 (285.0–7,539.0)	96.7 (91.8–99.1)	98.1 (93.2–99.8)	98.3 (94.1–99.8)	96.2 (90.6–99.0)

Class	Computer image analysis based criteria
Type 0	IEL count < 25/100 ECs
Type 1	IEL count ≥ 25/100 ECs+Vh fold change > 0.7+Cd:Vh < 0.5
Type 2	IEL count ≥ 25/100 ECs+Vh fold change ≤ 0.7 'OR' Cd:Vh ratio ≥ 0.5
Type 3	IEL count ≥ 25/100 ECs+Vh fold change ≤ 0.7+Cd:Vh ratio ≥ 0.5

	Intraobserver agreements (based on weighted κ-value)	Interobserver agreements (based on weighted κ-value)
Histological classification systems
Marsh classification [2]	39.7 to 60.9	12.9 to 48.2
Modified Marsh [1]	46.6 to 63.7	28.3 to 48.7
Corazza et al. [5]	54.6 to 84.6	28.04 to 42.9
Ensari [6]	54.5 to 64.5	28.04 to 42.9
Proposed quantitative histological classification	69.7 to 85.0	24.6 to 71.5

Table 1. Results of Quantitative Histological Assessment Performed with All Histological Parameters in the Derivation Cohort for Differentiation between Duodenal Biopsies from Controls and Patients with CeD

Values are presented as mean±SD or percent. Parentheses show 95% CI.

Table 2. Results of Quantitative Histological Assessment Performed with All Histological Parameters in the Validation Cohort for Differentiation between Duodenal Biopsies from Controls and Patients with CeD

Values are presented as mean±SD. Parentheses show 95% CI.

Table 3. The Proposed Quantitative Histological Classification for Assessment of Duodenal Mucosal Biopsies

IEL, intraepithelial lymphocytes; EC, epithelial cells; Cd, crypt depth; Vh, villous height.

Table 4. Intraobserver and Interobserver Agreements Using Histological Classification Systems