The diagnostic value of the duodenal biopsy: A clinico-pathologic analysis of 28,000 patients

Results 28,210 patients with and 75,175 without duodenal biopsies were studied. Duodenal biopsy patients were younger (52 years versus 58 years, p < 0.001) and more likely to be female (OR 1.46; p < 0.0001); 87% of children and 27% of adults had duodenal biopsies. Suspicion of malabsorption or sprue, diarrhoea, aneamia, and weight loss were strong predictors of duodenal biopsy. More than 80% of patients had normal duodenum, except those biopsied for sprue, 64% of whom had normal findings. Marsh II–IIIc lesions were diagnosed in 23% of patients with suspected sprue, but in 3.0% of those with diarrhoea, weight loss, or aneamia, and in 1.5% of patients with dyspepsia or GERD. Conclusions A clinical suspicion of sprue produces the highest yield of histopathologic abnormalities; women are biopsied more often than men despite having less duodenal pathology. Keywords Celiac disease Duodenal biopsy Duodenum Gender bias Histopathology 1 Introduction Mucosal biopsies of the gastrointestinal tract are obtained during endoscopic procedures for a wide variety of reasons. Whereas some clinicians subscribe to the notion that an endoscopic procedure without biopsies, irrespective of the macroscopic findings, is an incomplete one [1,2] , others target their biopsies to address each patient's signs and symptoms. One of the most common indications for obtaining duodenal biopsies is believed to be the suspicion of celiac sprue as a cause of malabsorption, bloating, diarrhoea, and aneamia. Therefore, most studies designed to evaluate the scope of duodenal biopsies have focused on their sensitivity and specificity for the diagnosis of celiac disease [3–8] . However, the proportion of patients who have duodenal biopsies during esophagogastroduodenoscopy (EGD) remains unknown; furthermore, we are aware of only one study that has evaluated the diagnostic yield of such biopsies and the potential impact of the resulting histopathologic diagnoses on the patient's management [9] . The answers to these questions are unlike to be found in single-centre studies, in which results may be biased by local expertise, individual clinical interests, and financial considerations. Such studies are also likely to include limited numbers of subjects, unless they span retrospectively over long periods of time. When they do, changing practice guidelines, evolving management strategies, and operator and pathologist turn-over can cause remarkable shifts in the results during the course of the study. To evaluate the current status of the duodenal biopsy in private gastroenterology practice in the United States, we have carried out a nation-wide study of more than 100,000 patients who underwent EGDs over the course of a single 12-month period. The goals of this study were to determine what subsets of patients undergoing EGD in a private setting have a duodenal biopsy, and why. We also sought to assess the diagnostic yield of the biopsies with respect to the clinical indications and the endoscopic findings. 2 Materials and methods 2.1 Study setting This study was conducted at Caris Diagnostics, a specialised gastrointestinal laboratory receiving specimens from gastroenterologists operating in private outpatient endoscopy centres in 34 states, the District of Columbia, and Puerto Rico. Biopsies are interpreted by an experienced group of gastrointestinal pathologists who share a common approach to biopsy evaluation and have achieved a high level of diagnostic uniformity through a pre-determined approach to specimen handling, diagnostic criteria, and terminology. The study was approved by the Caris Diagnostics Institutional Review Board. Because all data in this study were collected entirely by reviewing existing de-identified records, no direct contact with either patients or providers was involved, and no individual information is revealed in any form, this study was considered exempt from the need for informed consent from participants. 2.2 Sources of data We analysed electronic data from the Caris Diagnostics database, which includes demographic and clinical information for each patient, a summary of the endoscopic findings or the entire endoscopic report, the site of origin of each specimen, and the histopathologic report for each biopsy. To identify the records for eligible duodenal biopsies we extracted data for all patients who had undergone an EGD with a biopsy within the 12-month period from April 2007 to March 2008, and whose biopsies were diagnosed at the Caris Irving, Texas, laboratory. We then created separate databases to include all patients who underwent duodenal biopsies. Information regarding the histopathologic features of the duodenal biopsies was obtained by analysing individual diagnostic lines using additional search terms and Boolean logic in Visual Basic for Applications (VBA). Information on the clinical indications that led to the performance of EGD in our patients was obtained from a specific field in the pathology requisition form, which in most patients is filled directly with data transferred from electronic report writers (ERW) such as ProVation ® , gMed ® , MD Reports ® , and Summit ® . Endoscopic impressions are derived from either ERWs or the full endoscopy report, which is available in pdf format in a portion of our patients. Further extraction of clinical indications, clinical features, and endoscopic findings was performed using simple queries in Microsoft Access and confirmed or expanded by manual review. 2.3 Diagnostic criteria In our group, duodenal biopsies are diagnosed using the criteria proposed by Marsh and later modified by Oberhuber et al. [10,11] . For the purpose of this analysis, the following diagnostic categories were considered: (1) normal duodenal mucosa; (2) duodenal intraepithelial lymphocytosis, or DIL (>30 intraepithelial lymphocytes per 100 enterocytes, equivalent to Marsh I or II lesions; CD3 or CD8 immunohistochemical stains are performed only to confirm equivocal cases); (3) variable villous atrophy (Marsh IIIa lesions); (4) villous atrophy consistent with gluten-sensitive enteropathy (Marsh IIIb or IIIc); (5) duodenitis (including peptic duodenopathy with gastric foveolar metaplasia, with and without active inflammation; patches of uninflamed heterotopic oxyntic mucosa are not included in this group) with and without superficial erosions [12] . Patients with diagnoses of infection (including giardiasis and cryptosporidiosis), tumours (duodenal adenomas, primary and metastatic carcinomas, and lymphomas), hamartomas and heterotopias, any lesions diagnosed in less than 1% of the biopsies, as well as those with equivocal diagnoses were excluded from the condition-specific analyses. 2.4 Statistical analysis Statistical calculations were performed using SigmaStat Version 3.5 (Systat Software, Inc., Point Richmond, CA). Distributions of categorical variables were compared by an uncorrected chi-square test. Medians, means and standard deviations were calculated for continuous variables (age in years), and comparisons between groups were made by the Student's t -test. When normality check failed, the Mann–Whitney Rank Sum Test for non-parametric data was used. Simple odd ratios were calculated using an online OR calculator. [13] 3 Results 3.1 Duodenal biopsies During the study period 103,385 unique patients, including 1099 children 17 years of age or younger, underwent an EGD with at least one biopsy specimen from esophagus, stomach, or duodenum. Duodenal biopsies were obtained from 28,210 of these patients, including 958 children (3.4%); the median age was 52 years, range 0–95; and 65.5% were female. Of the 75,175 patients (including 141 children, or 0.2%) who had biopsies from other organs of the upper gastrointestinal tract but did not have a duodenal biopsy, 32,656 were male and 42,519 female (56.6%); the median age was 58 (range 0–104). Thus, patients from whom a duodenal biopsy was obtained were younger than those who did not have one (52 years versus 58 years, p < 0.001) and more likely to be female (OR 1.46; 95% CI 1.42–1.50; p < 0.0001); 87.2% of children who underwent an EGD with biopsies had a duodenal biopsy, in contrast to 27.4% of the adults. The precise anatomic location from which duodenal samples were obtained was indicated in the endoscopic report in 66.6% of the patients ( Table 1 ). 3.2 Indications for EGD The conditions most commonly stated as indications for the EGD in patient with and without a duodenal biopsy are listed in Table 2 . A clinical suspicion or history of malabsorption or sprue was strongly related to a duodenal biopsy, as was a history of diarrhoea, aneamia, and weight loss. Although duodenal biopsies were obtained less commonly in patients whose stated indication for EGD was GERD, still a surprising 27.5% of these patients had a duodenal biopsy. 3.3 Endoscopic findings The most commonly reported endoscopic findings in the duodenum were: normal, scalloping, duodenitis, erosion or ulcer, and nodule or polyp. Specific information regarding these findings was provided in 4241 patients who had a duodenal biopsy (15.0%) and in 7314 (9.7%) of patients from whom a biopsy from the esophagus of the stomach, but not the duodenum, was obtained. The histopathologic diagnoses associated with each of those endoscopic descriptions are depicted in Table 3 . 3.4 Duodenal histopathologic findings A total of 26,846 (95.2%) patients had a histopathologic diagnosis that met the inclusion criteria detailed above. The remaining 1364 patients (4.8%) had infections, tumours (adenomas, lymphomas, neuroendocrine tumours, primary and metastatic carcinomas), gastric heterotopia, or multiple biopsies with discordant diagnoses ( e.g ., adenoma in one specimen and normal mucosa in another). These patients are not included in the analysis. Table 4 depicts the major histopathologic diagnoses for each major EGD indication. Irrespective of the EGD indication, 80–84% of all patients had normal duodenal morphology, with the exception of those biopsied for sprue, 64% of whom had a normal biopsy. Marsh II–IIIc lesions were diagnosed in 23.2% of patients in the sprue group, but only in 3.0–2.6% of those whose indication was diarrhoea, weight loss, or aneamia, and in 1.6% and 1.4%, respectively, in patients with dyspepsia and GERD. This latter group was more likely to have peptic-related duodenal conditions such as duodenitis (10.0%). There were significant differences in the histopathologic diagnoses in biopsies from men and women ( Table 5 ). Overall, women were more likely than men to have normal duodenal histology (82.3% versus 74.2%; OR 1.62; 95% CI 1.53–1.72; p < 0.0001). Women were half as likely as men to have duodenitis (6.5% versus 13.9%; OR 0.43; 95% CI 0.40–0.47). The prevalence of sprue (Marsh IIIb and IIIc) was similar in men and women (around 1%), except for patients worked up for diarrhoea, amongst which 2.3% of men, but only 1.3% of women had histologic sprue (OR 0.54; 95% CI 0.35–0.83; p < 0.01). Duodenal lymphocytosis was more common in women, particularly in those biopsied for aneamia (5.2% for women versus 1.9% for men; OR 2.82; 95% CI 1.94–4.11; p < 0.0001). 4 Discussion Data analysed for this study were derived from a pathology database, which by its very nature includes only patients who had at least one biopsy specimen. Therefore, we have no information about those patients who, during the same period, had EGDs without biopsies at the same centres. Nevertheless, existing data suggest that a reasonable estimate can be made. In a 2000 CORI study of 6788 EGDs performed in 37 private practices distributed throughout the country [14] , biopsies were obtained in 2552 EGDs (37.5%). This percentage is similar to the range of the responses obtained through an informal poll of several gastroenterologists who submit their biopsies to Caris Diagnostics, who estimated performing biopsies in 40–60% of all EGDs. If we use the low figure of 40%, the total number of patients who underwent endoscopies at these centres would be around 250,000, suggesting that duodenal biopsies are obtained from 10% to 12% of patients undergoing an EGD. Four out of five of these patients have a histologically normal duodenum. A high proportion of normal histopathologic results for certain types of specimens has been interpreted as evidence that biopsies from certain segments of the gastrointestinal tract are obtained indiscriminately and, therefore, the majority of them may be essentially useless. The fallacy of such reasoning is not difficult to expose; for example, 84% of patients who underwent a duodenal biopsy for the investigation of aneamia (in most cases iron-deficiency aneamia) had normal duodenal histology. This is a valuable and clinically relevant negative finding: celiac sprue is ruled out and other causes of aneamia are investigated. In fact, the overall 20% positive yield of duodenal biopsies compares favourably to that of the biopsy from the terminal ileum (4.6–18%) [15,16] . There were considerable gender differences both in the biopsy patterns and in the frequency of certain histopathologic diagnoses. More women than men had duodenal biopsies for all EGD indications, with one exception: the “incidental” duodenal biopsy from patients evaluated for GERD was almost 6 times more common in men (OR 5.87; 95% CI 5.55–6.21; p < 0.0001). The histopathologic results suggest that this tendency to take more duodenal biopsies in men with GERD may be reasonable, since twice as many men as women had duodenitis. Women whose primary stated reason for EGD was dyspepsia (which in our search included also the keywords “bloating,” “epigastric pain,” and “nausea”) were 1.6 times more likely than men to have a normal duodenal biopsy. This finding may related to the greater prevalence of functional dyspepsia reported in women [17] . Since women in this population had also a greater prevalence of normal gastric biopsies (data not shown), these findings may provide further support to the concept, not yet unequivocally confirmed, that there are significant gender-based differences in the prevalence and manifestations of functional dyspepsia [18] . A clinical suspicion, positive serology, or a past history of celiac sprue prompted the EGD in 688 females and 288 males and the prevalence of lesions classified as Marsh II or higher was similar in the two groups (around 23%); this similarity persisted when all EGD indication groups were amalgamated (2.3% for both females and males). Thus, this population did not show the usually reported 2:1 female:male ratio in the prevalence of celiac sprue [19] . The positive predictive values (PPV) for endoscopic descriptions were highest for duodenal nodules or polyps (51%), scalloping (38.2%), and duodenitis (18.1%). The endoscopic description of an ulcer corresponded to a histologic erosion or ulcer in 18.9% of the patients, but duodenitis was found histologically in an additional 40.3%, most likely because the surrounding mucosa rather that the ulcer itself is usually sampled; it should also be noted ( Table 3 ) that no duodenal biopsies were taken from more than 3000 patients in whom duodenitis was seen endoscopically and from more than 300 in whom an ulcer was seen. These data corroborate the perception, common amongst both gastroenterologists and pathologists, that there is often little correlation between endoscopic and histologic findings, particularly in inflammatory conditions. In an effort to improve the correlation, pathologists have traditionally exhorted clinicians to provide more detailed information about the appearance and the location of the lesions sampled in a biopsy. While endoscopic descriptions may still have some utility and magnification endoscopy may enhance their precision, we suspect that the almost universal availability of endoscopic images – in many settings only a few clicks away from the inquisitive pathologist – will soon relegate lengthy descriptive requisitions to obsolescence. Although our data were gathered from a wide geographic area and include patients from all regions of the United States, they cannot claim universal validity for the US population. First, 99.6% of the patients included in this study carried some type of health insurance (a private plan, Medicare with or without additional private coverage, and Medicaid); therefore, some segments of the population are underrepresented. Second, there may be differences in the approach to gastrointestinal biopsies between private practice and public and teaching hospitals, whose patients are not included in this study. In conclusion, it would appear that approximately 10% of patients who undergo an EGD in a private practice setting have at least a duodenal biopsy, and 20% of these have a histopathologic abnormality that may influence the patient's management, whether in the form of therapy or further testing. Conflict of interest statement Dr. Robert Genta is an employee of Caris Diagnostics, Irving, TX. No conflicts of interest relevant to this manuscript are declared. This manuscript was written entirely by the authors, with no external assistance. No grant support was received for this study. References [1] H.A. Carpenter N.J. Talley Gastroscopy is incomplete without biopsy: clinical relevance of distinguishing gastropathy from gastritis Gastroenterology 108 1995 917 924 [2] H.A. Carpenter N.J. Talley The importance of clinicopathological correlation in the diagnosis of inflammatory conditions of the colon: histological patterns with clinical implications Am J Gastroenterol 95 2000 878 896 [3] W. Dickey Endoscopic markers for celiac disease Nat Clin Pract Gastroenterol Hepatol 3 2006 546 551 [4] W. Dickey Endoscopy, serology and histology in the diagnosis of coeliac disease Dig Liver Dis 34 2002 172 174 [5] N. Savas S. Akbulut U. Saritas Correlation of clinical and histopathological with endoscopic findings of celiac disease in the Turkish population Dig Dis Sci 52 2007 1299 1303 [6] J.J. Tischendorf K. Wopp K.L. Streetz The value of duodenal biopsy within routine upper endoscopy: a prospective study in 1000 patients Z Gastroenterol 46 2008 771 775 [7] A. Ravelli S. Bolognini M. Gambarotti Variability of histologic lesions in relation to biopsy site in gluten-sensitive enteropathy Am J Gastroenterol 100 2005 177 185 [8] V. Villanacci The problem of biopsies in the diagnosis of celiac disease Gastrointest Endosc 69 2009 983 984 [9] T.G. Morales P.E. Jaffe M.B. Fennerty Yield of routine endoscopy beyond the duodenal bulb J Clin Gastroenterol 24 1997 147 149 [10] G. Oberhuber G. Granditsch H. Vogelsang The histopathology of coeliac disease: time for a standardized report scheme for pathologists Eur J Gastroenterol Hepatol 11 1999 1185 1194 [11] D.A. Antonioli Celiac disease: a progress report Mod Pathol 16 2003 342 346 [12] M.E. Robert Inflammatory disorders of the small intestine R.D. Odze J.R. Goldblum Surgical pathology of the GI tract, liver, biliary tract, and pancreas 2nd ed. 2009 Saunders Elsevier Philadelphia 321 354 [13] Lowry R. VassarStats for a 2x2 Contingency Table (Rates, Risk Ratio, Odds, Odds Ratio, Log Odds;Phi Coefficient of Association; Chi-Square Test of Association; Fisher Exact Probability Test). 9-15-2009. [14] T. McCashland R. Brand E. Lyden G.P. de The time and financial impact of training fellows in endoscopy. CORI Research Project. Clinical outcomes research initiative Am J Gastroenterol 95 2000 3129 3132 [15] K.K. Yoong T. Heymann It is not worthwhile to perform ileoscopy on all patients Surg Endosc 20 2006 809 811 [16] J.B. McHugh H.D. Appelman B.J. McKenna The diagnostic value of endoscopic terminal ileum biopsies Am J Gastroenterol 102 2007 1084 1089 [17] S.K. Ahlawat M.T. Cuddihy G.R. Locke III Gender-related differences in dyspepsia: a qualitative systematic review Gend Med 3 2006 31 42 [18] S.N. Flier S. Rose Is functional dyspepsia of particular concern in women? A review of gender differences in epidemiology, pathophysiologic mechanisms, clinical presentation, and management Am J Gastroenterol 101 Suppl. 2006 S644 S653 [19] M.T. Bardella C. Fredella V. Saladino Gluten intolerance: gender- and age-related differences in symptoms Scand J Gastroenterol 40 2005 15 19