CD and ITB are both chronic granulomatous diseases of the intestine that share similar clinical manifestations, endoscopic findings, CTE and histological features [1, 2]. In China, where tuberculosis is prevalent and the incidence of CD is increasing, it is important to accurately identify these two diseases in a timely and accurate manner. It is very important to make a correct differential diagnosis and take reasonable treatment measures in the early stage because of the distinct natural history and prognosis of these two disease. Thus, there is an urgent need to develop a better diagnostic tool to distinguish CD from ITB. In this research, we screened demographics, clinical manifestations, laboratory examination, endoscopic findings, CTE and pathological features in detail and searched for significant predictors for differentiating CD from ITB by univariate and multivariate logistic regression analyses. We found some clinical manifestations, laboratory examination, endoscopic findings, CTE and pathological features are helpful to distinguish CD from ITB. Among them, T-SPOT positive, cobblestone appearance, comb sign and granuloma are the most important features to distinguish CD from ITB. Cobblestone appearance and comb sign are the independent predictors of CD, while T-SPOT positive and granuloma are the independent predictors of ITB. On this basis, a nomogram prediction model for distinguishing CD from ITB was developed and assessment, which has high discrimination, calibration and clinical efficiency. It can be used as an accurate and convenient diagnostic tool to distinguish Crohn’s disease from intestinal tuberculosis, facilitating clinical decision-making.
The principal clinical manifestations of CD and ITB are abdominal pain, abdominal mass, and weight loss, accounting for more than 70%. However, the difference was not statistically significant, indicating that these main clinical manifestations lacked specificity in differential diagnosis. On univariate analysis of variables, diarrhea and perianal disease were more common in patients with CD, whereas ascites and pulmonary tuberculosis were suggestive of ITB, which is consistent with previous reports [1, 2, 17]. In addition, it is reported that the T-SPOT has superior sensitivity and specificity in the diagnosis of ITB compared with tuberculin skin test because it is not affected by previous Bacille de Calmette Guerin (BCG) vaccination and most nontuberculous mycobacteria infections [10, 18]. Our results showed that the sensitivity and specificity of T-SPOT are 92.2% and 93.0% in the differential diagnosis of CD and ITB, suggesting a valuable laboratory examination based on its high sensitivity and specificity, which is similar to previous reports .
Endoscopy plays an important role in the differential diagnosis of CD and ITB. The endoscopic features favoring CD included longitudinal ulcer, skip lesions and cobble stone appearance, whereas transverse (ring-shaped) ulcers favored the diagnosis of ITB in our study, as previously reported [2, 10, 20]. Our study also illustrated that intestinal ulcers are more common in the right colon in both Crohn’s disease and intestinal tuberculosis, especially in the ileocecum. One explanation is that the ileocecal region is commonly involved secondary to the high concentration of lymphoid aggregates in this area and prolonged contact between the bacilli and ileocecal mucosa . Involvement in the terminal ileum, transverse colon, descending colon, sigmoid colon and the rectum was significantly more frequent in patients with CD than ITB, which is also consistent with previous studies [12, 22].
Computed tomographic enterography (CTE) are the preferred imaging modalities for evaluating and differentiating between patients with CD and ITB . In our study, we found that asymmetrical bowel wall thickening, skip lesion, segmental small-bowel lesions, target sign, the comb sign, mesentery fibrofatty proliferation and homogeneous enhancement of lymph nodes were significantly more common in Crohn’s disease patients, which could provide reference for CD and ITB differentiation, as previously reported [3, 23]. Inflammatory stimuli leads to the increasing of small blood vessels around the lesion, which is shown as “comb sign” on CTE in patients with CD. Park et al. pointed out that a positive comb sign was the most suggestive finding of CD . Mao et al. found that segmental small bowel involvement and comb sign were independent predictors of CD. Combining CTE and colonoscopic findings increased the accuracy of diagnosing either CD or ITB . Visceral fat is a component of mesenteric fat and mesenteric fatty proliferation is one of the hallmarks of CD, being recognized as early as 1932 by Burril B. Crohn . Fat hypertrophy, fat wrapping, and creeping fat have been associated with active CD . We also found that mesentery fibrofatty proliferation favored the diagnosis of CD in our study.
Pathological features are the key to distinguishing ITB from CD. The “gold standard” typical granulomas with caseous necrosis for the diagnosis of intestinal tuberculosis limits its value in clinical application due to the low positive rate. In our study, caseous necrosis was found in only 7.0% of ITB patients, which was lower than the reported positive rate about 11.1% . Our data presented in this study demonstrated that compared with CD patients, granuloma was more common in ITB patients, and the difference was statistically significant, which could provide a reference for the differential diagnosis. Yu et al. also found that granuloma was more common in intestinal tuberculosis than in CD. Furthermore, night sweats, longitudinal ulcers and granulomas were the most important features to differentiate Crohn’s disease from intestinal tuberculosis on further multivariable logistic regression analysis . Tubercular granulomas are usually large (> 200 μm), dense, confluent, located in submucosa, however granulomas in CD are usually small (microgranuloma), discrete, sparse and can be situated either in the mucosa or in the submucosa [1, 13, 17]. Surgical specimens may reveal the presence of fissure-like ulcers, which are more common in CD and may extend till serosa, whereas they are rare in ITB, and if present they usually do not extend beyond the submucosa. In our study, fissure-like ulcers were found in a small proportion of CD patients about 3.3%, which are the limitation in clinical practice.
Although many valuable parameters of clinical manifestations, endoscopic findings, CTE and histological features discussed above help to differentiate between CD and ITB, the value of using a single parameter in distinguishing these two disease is very limited in clinical practice due to low sensitivity or specificity. Thus, establishing diagnostic model with multiple selected valuable parameters may be expected to obtain a diagnostic method with high accuracy and clinical maneuverability. Lee et al. established a diagnostic model base on eight colonoscopy parameters and showed a positive predictive value for CD as 94.4%, a positive predictive value for ITB as 88.9%, and accuracy 95.5% . Li et al. established a model based on clinical features and an endoscopic model in their study. Both these models had moderate sensitivity and specificity of approximately 80% in differentiating CD and ITB . In another study from China, two models were established based on clinical and CTE features, which showed diagnostic accuracy of 91.0% and 95.7%, respectively . All above models were established based on single or limited examination tools with quite different sensitivity, specificity and accuracy and no validation was performed.
In a study from India, Makharia et al. established a diagnostic model including blood in the stool, weight loss, histological focally enhanced colitis, and involvement of the sigmoid colon recruited parameters from clinical manifestation, endoscopic findings, and pathologic features with 83% sensitivity, 79.2% specificity, and 81.1% diagnostic accuracy . Yu et al. recruited independent predictors for diagnosis of CD and ITB included night sweats, longitudinal ulcers, and granulomas as variables for the predictive model, which had good diagnostic accuracy with an AUC of 0.86 . Jung et al. formulated a predictive model including age, sex, ring-shaped ulcers, suspicion of radiological pulmonary tuberculosis, longitudinal ulcers, diarrhea, and sigmoid colon involvement in a Korean population and showed a better performance, with a sensitivity of 95.9%, a specificity of 94.9%, and the AUC of 0.979 . The above models reported the diagnostic yield of differential diagnosis with big variations. Besides, most of these models included only one or some examination tools and have not evaluated all the features, in particular lack of radiological features. Especially, the formulae used in these prediction models are complicated and difficult to be applied in clinical practice.
Recently, He et al. establish two models based on 7 differential variables: age, transverse ulcer, rectum involvement, skipped involvement of the small bowel, target sign, comb sign, and interferongamma release assays (for model 1) or purified protein derivative (for model 2), respectively . Accordingly, two nomograms of the above two models were developed for clinical practical use respectively and the nomogram 1 with 92.4% specificity, 95.8% sensitivity, 94.7% accuracy for diagnosing CD, and the nomogram 2 with 90.9% specificity, 82.5% sensitivity, 82.1% accuracy for diagnosing CD, which can be conveniently used to identify some difficult CD or ITB cases, allowing for decision-making in a clinical setting. However, these models do not include pathological features, which has an important role in differentiating CD from ITB, and recruit feature (age) which may not be applicable to other populations. Furthermore, they do not evaluate the calibration and clinical usefulness of the models.
The present study comprehensively screened variables with statistical differences from demographics, clinical manifestations, laboratory examination, endoscopic findings, CTE and pathological features. Multivariate logistic regression analysis showed that T-SPOT positive, cobblestone appearance, comb sign and granuloma were significant predictors in differentiating CD and ITB. Base on the above multivariate analysis, a nomogram prediction model to distinguish CD from ITB was successfully established with the higher sensitivity, specificity, accuracy of 94.4%, 93.0%, 94.0%, respectively. The high C-index as well as the area under the curve showed that this prediction model can be widely and accurately used to distinguish CD from ITB. And the nomogram showed good internal calibration between the actual observation and the prediction in the derivation cohort. In addition, the DCA and clinical impact curves were employed to determine a clinical decision point that the patients could obtain the highest net benefit. Taken together, the present nomogram can be used as an accurate and convenient diagnostic tool to distinguish CD from ITB, facilitating clinical decision-making.
Our study has several strengths. First, this present study developed a diagnostic nomogram prediction model to differentiate between CD and ITB, with intuitive, easy-touse characteristics. Second, screening variables are more comprehensive, especially the inclusion of radiological and pathological features in our study. Third, this study employed for the first time the DCA and clinical impact curve to evaluate the clinical efficiency of diagnostic nomogram. However, there are some shortcomings in our study that this is a single-center study with a limited number of patients and we validated our model in the same data set due to the limited number of patients, so further studies with a larger sample size from multiple centers are needed to validate this predictive model. Besides, this model may not be applicable in other countries and regions, further research conducted among other populations is warranted to provide more evidence.
In conclusion, clinical manifestations, laboratory examination, endoscopic findings, CTE features and histological results are helpful to distinguish CD from ITB. T-SPOT positive, cobblestone appearance, comb sign and granuloma are the most important features to distinguish CD from ITB. On this basis, a nomogram prediction model for distinguishing CD from ITB was developed and assessed, which has high discrimination, calibration and clinical efficiency. It can be used as an accurate and convenient diagnostic tool to distinguish Crohn’s disease from intestinal tuberculosis, facilitating clinical decision-making.