Use of surrogate markers of inflammation and Rome criteria to distinguish organic from nonorganic intestinal disease

GASTROENTEROLOGY 2002;123:450 – 460

Use of Surrogate Markers of Inflammation and Rome Criteria to Distinguish Organic From Nonorganic Intestinal Disease JEREMY A. TIBBLE,* GUDMUNDUR SIGTHORSSON,* RUSSELL FOSTER,* IAN FORGACS,‡ and INGVAR BJARNASON* *Department of Medicine, Guy’s, Kings, St. Thomas’ Medical School; and ‡Department of Gastroenterology, Kings College Hospital, London, England

Background & Aims: Differentiating symptoms of irritable bowel syndrome (IBS) from those of organic intestinal disease is a familiar problem for physicians. The aim of this study was to assess the sensitivity, specificity, and odds ratios (ORs) of fecal calprotectin, small intestinal permeability, Rome I criteria, and laboratory markers of inflammation (erythrocyte sedimentation rate [ESR], C-reactive protein [CRP], blood count) in distinguishing organic from nonorganic intestinal disease. Methods: A total of 602 new referrals to a gastroenterology clinic who had symptoms suggestive of IBS or organic intestinal disease were studied for these parameters. All patients underwent invasive imaging (barium/ endoscopic examination) and other investigations as appropriate, with physicians blinded to the results of fecal calprotectin and intestinal permeability. Results: A total of 263 patients were diagnosed with organic disease and 339 with IBS. At 10 mg/L, the sensitivity and specificity of calprotectin for organic disease were 89% and 79%, respectively, and that of intestinal permeability for small intestinal disease were 63% and 87%, respectively. Sensitivity of positive Rome criteria for IBS was 85% with a specificity of 71%. An abnormal calprotectin test had an OR for disease of 27.8 (95% confidence interval [CI], 17.6 – 43.7; P < 0.0001) compared with ORs of 4.2 (95% CI, 2.9 – 6.1; P < 0.0001) and 3.2 (95% CI, 2.2– 4.6; P < 0.0001) for elevated CRP and ESR values. An abnormal permeability test gave an OR of 8.9 (95% CI, 5.8 –14.0; P < 0.0001) for small intestinal disease. The OR for IBS with positive Rome criteria was 13.3 (95% CI, 8.9 –20.0). Conclusions: Fecal calprotectin, intestinal permeability, and positive Rome I criteria provide a safe and noninvasive means of helping differentiate between patients with organic and nonorganic intestinal disease.

altered bowel habit, whereas other clinical features such as a predominance of diarrhea and rectal bleeding will increase the likelihood of inflammatory disease. Because the clinical differentiation remains problematic, many patients in the IBS category are investigated extensively with invasive radiographic and endoscopic imaging to make a diagnosis of exclusion. This has significant implications for health care costs as well as exposing patients to the inherent risks associated with such investigative procedures. The prevalence estimates for IBS indicate that 14%– 24% of women and 14%–19% of men are affected by this condition in Britain and the United States.1 These estimates have generally been based on patients fitting a number of symptom-based diagnostic criteria that have been developed to identify patients with IBS. The Manning2 and other criteria3– 6 have been widely used in clinical research; however, because there are no biological markers to define IBS, validation of such criteria has been difficult. There have been concerns regarding both the sensitivity and specificity of the criteria, which have ranged from 58% to 94% and 55% to 74%, respectively, in various studies,2,6,7 with their discriminate value possibly affected by sex.8 As a result, a consensus definition and criteria were developed, the Rome I criteria,9,10 for IBS and other functional gastrointestinal disorders. A number of investigators11–13 have recommended a straightforward approach to the evaluation and treatment of patients with IBS based on the use of the Rome criteria as a means of cost-effective management, avoiding the costly workup to sort through a confusing array of gastrointestinal symptoms. The stepwise assessment of pa-

astroenterologists are often faced with the diagnostic difficulty of differentiating patients with irritable bowel syndrome (IBS) from those with intestinal pathology, in particular inflammatory bowel disease (IBD). Many symptoms are common to both conditions, including abdominal pain, bloating, excessive flatus, and

Abbreviations used in this paper: CI, confidence interval; CRP, C-reactive protein; ELISA, enzyme-linked immunosorbent assay; ESR, erythrocyte sedimentation rate; IBS, irritable bowel syndrome; NPV, negative predictive value; OR, odds ratio; PPV, positive predictive value; ROC, receiver operator curve. © 2002 by the American Gastroenterological Association 0016-5085/02/$35.00 doi:10.1053/gast.2002.34755

G

August 2002

tients with suspected IBS based on positive Rome I criteria1,12 includes a full blood count, erythrocyte sedimentation rate (ESR)/C-reactive protein (CRP), serum chemistry, thyroid function tests, and stool examination for parasites and ova to exclude organic diseases; however, as shown by Tolliver et al.,14 these parameters have a disappointing diagnostic yield in patients meeting the Rome I criteria. Despite these recommendations, the Rome I criteria are rarely used formally in routine clinical practice, although they may often be used as entry criteria for research studies in an attempt to standardize characteristics of patient groups. The most striking difference between IBS and IBD is that the former is noninflammatory in nature. Therefore, one possibility is to measure surrogate markers of intestinal inflammation to differentiate between the two. Assessment of serologic markers (ESR, CRP) has in general been disappointing, probably because of their lack of sensitivity and specificity, because they are indirect measures of inflammation, and they can be affected by a number of nonintestinal diseases. The direct assay of feces for inflammatory markers has the potential to improve on the discriminant value of the serologic markers. Calprotectin, a calcium-binding protein found in neutrophilic granulocytes, monocytes, and macrophages,15 comprising up to 60% of the total cytosolic protein content of neutrophils,16 resists metabolic degradation16 and can be measured in feces. Its use has been extensively validated, showing consistent abnormalities in patients with IBD, colorectal carcinoma, and nonsteroidal enteropathy.17–23 Tests of intestinal permeability (differential urinary excretion of lactulose/L-rhamnose) have been shown to be sensitive for the detection of patients with inflammation of the small intestine, including that due to celiac disease, Crohn’s disease, and intestinal infections. The principle of the differential urinary excretion of 2 orally administered test probes is that premucosal and postmucosal determinants of their excretion are equal. Hence, the ratio of lactulose/L-rhamnose becomes a specific index of intestinal permeability and has been proposed as a diagnostic screen for small bowel disease.24 The aims of this study were to (1) determine if the use of fecal calprotectin and intestinal permeability are useful in differentiating between patients with organic and nonorganic disease and (2) compare the use of these 2 markers of intestinal inflammation with conventional laboratory parameters and the symptom-based Rome I criteria in differentiating between such patients.

ORGANIC VERSUS NONORGANIC INTESTINAL DISEASE

451

ment of a teaching hospital by general practitioners in South London. All patients had clinical symptoms suggestive of organic small or large intestinal disease or IBS that had not responded to therapy instituted by the primary care physician and were of sufficient severity for further consultation and investigation to exclude organic pathology. A total of 231 men and 371 women were recruited (median age, 40 years; range, 18 –90 years). Figure 1 shows the principal symptoms for which patients were referred according to their final diagnosis after all investigations were complete. A total of 275 patients referred for investigation of symptoms of esophageal reflux, symptoms clearly associated with gastroesophageal pathology, or functional or isolated dyspepsia (defined as a combination of symptoms related to food with localization in the upper abdomen, including epigastric pain or discomfort, bloating, belching, and nausea without symptoms of colicky abdominal pain or alteration in bowel habit at time of referral) were not included. Exclusion criteria were determined at first consultation, before investigation and classification by Rome criteria, and these patients were excluded from the study at this point. However, patients with dyspepsia in addition to intestinal symptoms were recruited into the study. Patients with a previously known diagnosis of IBD, colorectal carcinoma, and serious cardiopulmonary, hepatic, renal, neurologic, and psychiatric disease were also specifically excluded All patients underwent clinical evaluation by one of 4 gastroenterologists at Kings College Hospital, including history and full physical examination. All patients underwent an invasive diagnostic imaging procedure (barium enteroclysis/ enema and/or endoscopy/colonoscopy) of the gastrointestinal tract appropriate to their symptoms (see Results), with the decision on type of imaging procedure made with the investigating clinician blinded to results of laboratory parameters and Rome classification. Patients were specifically asked about those symptoms included in the Rome I criteria by questionnaire9,10 (Table 1) and were subsequently classified by a separate investigator as having positive or negative Rome I criteria. Rome I criteria were used because the study began before the introduction of Rome II definitions.25 All patients provided a single stool sample for fecal calprotectin assay, underwent a lactulose/L-rhamnose small intestinal permeability test, and provided blood for routine assay of full blood count, ESR, CRP, and thyroid and liver function tests; normal values were

Materials and Methods We prospectively studied 602 consecutive patients referred to and seen in a gastroenterology outpatient depart-

Figure 1. Principal symptoms for which patients were referred according to the final diagnostic group.

452

TIBBLE ET AL.

Table 1. The Rome I Diagnostic Criteria for IBS At least 3 months of continuous or recurrent symptoms of one of the following: Abdominal pain or discomfort 1. Relieved by defecation 2. Associated with a change in frequency of stool 3. Associated with a change in stool consistency Plus Two or more of the following for at least one quarter of occasions or days: 1. Altered stool frequency (⬎3 stools/day or ⬍3 stools/week) 2. Altered stool form (lumpy/hard/loose/watery) 3. Altered stool passage (strain/urgency/feeling of incomplete evacuation) 4. Passage of mucus 5. Bloating or feeling of abdominal distention

previously established by the relevant chemical pathology laboratory. Patients with diarrhea as a prominent symptom or a history of rectal bleeding underwent rigid sigmoidoscopy and rectal biopsy at first visit and provided 3 stool samples for microscopy, culture, and sensitivity. Additional investigation was left to the discretion of the investigating clinician, who was blind to the fecal calprotectin and intestinal permeability results. In view of the fact that use of nonsteroidal antiinflammatory drugs (NSAIDs) may affect intestinal permeability and levels of fecal calprotectin,23 use of this drug class in the preceding 3 months was determined. Fifty-seven patients had taken NSAIDs during this time period; 35 were taking a therapeutic dose on a regular basis and 22 had taken NSAIDs on an as-needed basis, none within 2 weeks of being seen as outpatients. The final diagnosis was based on normality of conventional investigation, including results of inflammatory parameters, imaging procedures, and the judgment of the investigating clinicians (who were blinded to formal Rome classification outcome as well as the calprotectin and permeability results). After patients had completed all diagnostic investigations and had a firm diagnosis, they were followed up for 3 months, at which time they were grouped as having organic or nonorganic intestinal disease for statistical analysis. The study was approved by the Kings Healthcare Local Research Ethics Committee, and all patients gave informed consent.

Lactulose/L-Rhamnose Small Intestinal Permeability Ratio Patients fasted overnight and ingested 100 mL of iso-osmolar test solution at 8 AM as previously described,26 followed by a further 2-hour fast. Complete 5-hour urine collections were made into a container containing 1 mL (10% wt/vol) mercurithiosalicylate (Merthiolate) as a preservative. Analyses of the sugars were performed by thin-layer chromatography and scanning densitometry,27 which is accurate (with recovery ⬎90%), sensitive (minimum level of detection ⬍1 mg/100 mL), and precise (coefficient of variation for the whole procedure between 2% and 8% without replication). The

GASTROENTEROLOGY Vol. 123, No. 2

normal differential urinary excretion of lactulose/L-rhamnose of ⬍0.05 (95% confidence interval [CI ], 0.012– 0.05) was established previously.26

Measurement of Fecal Calprotectin Patients provided a single stool sample for measurement of calprotectin that was submitted within 48 hours. Handling of stool and quantitation of calprotectin by enzymelinked immunosorbent assay (ELISA) was as previously described.18,23 In brief, 5-g aliquots were homogenized for 45 seconds at 20,000 rpm with an Ultra Turrax (Ika Werke, Germany) mechanical homogenizer with 10 mL Tris-buffered saline containing 10 mmol/L CaCl2 and 0.25 mmol/L thimerosal as an antimicrobial agent (pH 8.4). Supernatants were collected after centrifugation. In the ELISA, an immunoglobulin G fraction of rabbit anti-calprotectin15 was used for capture, and immunoaffinity purified alkaline phosphatase conjugated antibody was used for development.18 The ELISA has an intra-assay variation of 2% and interassay variation of ⬍15% with normal values ⬍10 mg/L (range, 0.5–10.9; n ⫽ 48).

Statistical Analysis Normal values for ESR, CRP, liver function tests, full blood count, fecal calprotectin, and lactulose/L-rhamnose permeability ratio were previously defined. Receiver operator curve (ROC) analysis28 was performed for each variable to determine sensitivity and specificity values for distinguishing organic and nonorganic disease, as well as calculating the positive predictive value (PPV) and negative predictive value (NPV) for each variable. For the purposes of assessing the value of the permeability ratio in detecting small bowel disease to which it is applicable, patients with large bowel organic disease were classified as nonorganic when calculating the above values for this measure. The Shapiro–Wilks W test for the normal distribution29 showed the results to be nonnormally distributed; therefore, median values and ranges were calculated for each continuous variable for the 2 disease group classifications. Statistical comparison was performed using the Mann–Whitney nonparametric test. The odds ratio (OR) of having organic disease with an abnormal variable was calculated for each parameter by logistic regression. Multiple logistic regression analysis using the Cox model30 was performed using all variables to determine those that best differentiated organic from nonorganic disease. Logistic regression analysis was used to determine the OR of having different subtypes of disease (small bowel, large bowel, or nonorganic disease) using a combination of fecal calprotectin, small intestinal permeability, and Rome criteria. Sensitivity, specificity, PPV, and NPV were calculated for each of the combinations of the above parameters in differentiating between subtypes of disease. Statistical significance was tested using ␹2.

August 2002

ORGANIC VERSUS NONORGANIC INTESTINAL DISEASE

Results Patient Diagnosis Each patient underwent one or more invasive diagnostic imaging procedures of the gastrointestinal tract as the gold standard, appropriate to their symptoms. In total, 372 patients had a full colonoscopy (including terminal ileal intubation), 5 had an endoscopic retrograde cholangiopancreatography, 15 had an enteroscopy, and 85 had an esophagogastroduodenoscopy, of whom 38 had both a colonoscopy and esophagogastroduodenoscopy. A further 65 patients had incomplete colonoscopy and underwent barium enema examination to visualize the remainder of the colon. Enteroclysis was performed if small intestinal disease was believed to be likely based on clinical symptoms and routine blood investigations. A barium enema was performed in 153 patients (over and above those performed following an incomplete colonoscopy), 150 patients had a barium enteroclysis, and 13 patients had both barium procedures. Of the 602 patients recruited, 263 patients were finally classified as having organic intestinal disease and 339 as nonorganic disease (Table 2). A total of 129 patients had small intestinal disease with and without

453

colonic disease, and 134 had colonic disease only. Patients with both small and large bowel Crohn’s disease were classified as having small bowel disease for the purposes of statistical analysis when assessing the value of the small intestinal permeability test. NSAID enteropathy was diagnosed in the presence of otherwise unexplained iron deficiency anemia in patients taking NSAIDs, improvement of diarrhea following discontinuation of NSAIDs, or following enteroscopy (also used for the diagnosis of alcoholic enteropathy) showing characteristic mucosal hemorrhagic lesions or discrete ulcers.31 Of the 35 patients taking NSAIDs regularly, 13 (37%) had an eventual diagnosis of NSAID enteropathy, whereas the remainder and those taking NSAIDs on an as-needed basis all had an eventual diagnosis of IBS. Table 2 shows the final diagnosis in the 602 patients, the site of disease, and fecal calprotectin/permeability levels in the different diagnostic groups. Figure 2 shows the fecal calprotectin levels in the different diagnostic groups. The fecal calprotectin levels in all organic diagnostic groups differed significantly (P ⫽ 0.001 to ⬍0.0001) from that in the nonorganic group except for the group with diabetic diarrhea (P ⫽ 0.06).

Table 2. Different Diagnostic Groups in the 602 Patients and Their Disease Classification for Statistical Analysis No. of patients Organic disease Small bowel Crohn’s disease Ileal Small and large bowel Celiac disease (Untreated) Infective diarrhea Giardia Campylobacter jejuni Small bowel enteropathy NSAID Alcoholic Radiation Diabetic diarrhea Colonic Colitis Crohn’s Ulcerative Microscopic Collagenous Diverticular disease Cancer Nonorganic disease IBS IBS ⫹ nonulcer dyspepsia IBS ⫹ other

73 11 12

Fecal calprotectin (mg/L; normal, ⬍10)a

Intestinal permeability ratio (normal, ⬍0.05)a

62 (1–128,000)

0.056 (0.01–0.54)

13 (1–141) 31 (2–145)

0.10 (0.02–0.49) 0.04 (0.01–0.14)

21 (2–195)

0.05 (0.02–0.22)

6 (1–68)

0.05 (0.01–0.11)

8 1 13 6 2 5

18 87 5 1 14 7

59 (5–5120)

0.03 (0.01–0.38)

21 (4–79) 47 (3–94) 4 (1–50)

0.03 (0.01–0.05) 0.04 (0.02–0.14) 0.03 (0.01–0.20)

275 38 26

NOTE. The IBS ⫹ other nonsymptomatic group included 5 patients with lactose intolerance who remained symptomatic on a lactose-free diet, 4 with angiodysplasia, 11 with (incidental) hemorrhoids, and 6 with melanosis coli. aValues expressed as median (range).

454

TIBBLE ET AL.

GASTROENTEROLOGY Vol. 123, No. 2

Figure 3. ROC for fecal calprotectin identifying patients with organic disease. At the cutoff value for normality (10 mg/L), fecal calprotectin has a sensitivity of 89% and specificity of 79% for organic disease. Figure 2. Dot plot of fecal calprotectin levels in the different diagnostic groups. The cutoff value for normality (10 mg/L) is shown by the horizontal line. The calprotectin levels in all organic diagnostic groups except those with diabetic diarrhea differed significantly (P ⫽ 0.001 to ⬍0.0001) from those in the IBS group.

Sensitivity and Specificity of Tests for Diagnostic Groups Table 3 shows the sensitivity and specificity of the individual different parameters in identifying patients with organic and nonorganic intestinal disease as determined by ROC analysis. ROCs for fecal calprotectin (Figure 3) and intestinal permeability (Figure 4) are shown. The maximal sum of sensitivity and specificity for fecal calprotectin is achieved at a value of 10 mg/L, the same as that previously defined as the cutoff value for normality. Fecal calprotectin had a significantly higher sensitivity (P ⬍ 0.0001) for organic disease than ESR and CRP. For intestinal permeability, the maximal sum of sensitivity and specificity for identifying patients with small bowel disease is achieved over the permeability ratio range of 0.04 (sensitivity, 71%; specificity, 79%) to 0.05 (sensitivity, 63%; specificity, 87%), with the latter our cutoff value for normality. ROC analysis for hemoglobin concentration, liver and thyroid function tests,

and patient age or sex did not show any significant discriminant value for these parameters. In view of the fact that patients with clinical “red flags” (anemia, weight loss, or rectal bleeding) are likely to need investigation irrespective of laboratory parameter results, we assessed the sensitivity and specificity of fecal calprotectin, ESR, and CRP in detecting disease in a subgroup of patients believed to be at low (Rome positive and red flag negative) or medium (Rome negative and red flag negative) risk of organic disease if a 3-tiered, clinically practical system of investigation was used (red flags and Rome criteria followed by laboratory inflammatory parameters). Of 281 low-risk patients, 40 (14%) were found to have organic disease (3 with diverticular disease, 3 with diabetic diarrhea, 5 with celiac disease, 16 with ileal Crohn’s disease, 7 with colonic Crohn’s disease, and 6 with microscopic/collagenous colitis). Fourteen of these would have been identified by performing an ESR and CRP (sensitivity, 35%; specificity, 73%), whereas 36 (sensitivity, 90%; specificity, 80%) were identified by the calprotectin assay, including all of those identified by ESR and CRP. Of 263 medium-risk patients, 95 were found to have nonorganic disease. The respective sensi-

Table 3. Sensitivity and Specificity of Different Parameters for Organic and Nonorganic Disease Parameter Fecal calprotectin ⬎10 mg/L For small and large intestinal disease For small intestinal disease only Positive Rome criteria for IBS Lactulose/L-rhamnose ratio ⬎0.05 for small intestinal disease CRP ⬎5.0 mg/L ESR ⬎10 mm/h

Sensitivity

Specificity

89 86 85

79 60 71

63 50 58

87 81 72

NOTE. Values for calprotectin, CRP, and ESR are for organic disease. The values for positive Rome criteria are for IBS and those for the lactulose/L-rhamnose ratio are for organic small intestinal disease.

Figure 4. ROC for intestinal permeability identifying patients with organic small intestinal disease. At the cutoff value for normality (0.05), intestinal permeability has a sensitivity of 63% and specificity of 87% for organic small intestinal disease.

August 2002

ORGANIC VERSUS NONORGANIC INTESTINAL DISEASE

tivity and specificity for detecting organic disease were 58% and 73% for ESR and CRP and 89% and 77% for calprotectin. PPV and NPV of Tests for Diagnostic Groups Table 4 shows the PPV and NPV for each of the individual laboratory parameters for organic disease; the PPV for the Rome I criteria are for IBS and those for intestinal permeability for organic small intestinal disease. Fecal calprotectin and the Rome I criteria seem to provide the best set of predictive values. Table 5 shows the median values and ranges for each of the laboratory parameters in the organic and nonorganic groups. Statistical significance of differences in individual parameters between organic and nonorganic groups is greatest for fecal calprotectin (P ⬍ 0.0001). ORs for Diagnostic Groups The ORs for having organic disease with an abnormal result are shown for each of the laboratory parameters in Table 4. The OR for having organic intestinal disease with an elevated fecal calprotectin level is 27.8 compared with an elevated CRP and ESR in which the OR for organic disease is 4.2 and 3.2, respectively. The OR for having IBS with positive Rome I criteria was 13.3 and for having organic small intestinal disease with an abnormal permeability ratio was 8.9. Combinations of Parameters in Distinguishing Between Diagnostic Groups Table 6 shows the sensitivity, specificity, PPV, NPV, and OR of combinations of different parameters in differentiating between diagnostic groups. When all parameters were considered together in a multiple logistic regression analysis to differentiate patients with organic/nonorganic intestinal disease, only fecal calprotectin (OR for organic disease, 21.3; 95% CI, 12.1–27.2; P ⬍ 0.0001), positive Rome I criteria (OR for IBS, 7.3; 95% CI, 4.2–12.8; P ⬍ 0.0001), and an

455

abnormal permeability ratio (OR for organic disease, 2.0; 95% CI, 1.0 – 4.0; P ⫽ 0.04) remained significant. Because an abnormal small intestinal permeability test should identify patients with organic small intestinal disease and an abnormal fecal calprotectin test should identify patients with organic disease within both the small and large intestine, the combination of these 2 parameters was analyzed to examine how well they identified the site of organic disease. Patients with an abnormal permeability ratio and elevated fecal calprotectin level compared with all other patients had an OR of organic small intestinal disease of 15.0 (95% CI, 8.9 –25.0; P ⬍ 0.0001). Forty-five of the 602 patients had an increased permeability test with a normal calprotectin level, of whom 10 had an organic diagnosis. Of these, 5 patients had untreated celiac disease, 1 patient had diabetic diarrhea with rapid small intestinal transit, 3 patients had quiescent terminal ileal Crohn’s disease determined from terminal ileal biopsy specimens, and 1 patient had proctitis. Patients with a normal intestinal permeability and elevated fecal calprotectin level had an OR for organic colonic intestinal disease of 13.3 (95% CI, 8.3–21.3; P ⬍ 0.0001), and patients with both a normal intestinal permeability ratio and fecal calprotectin level had an OR for IBS of 25.0 (95% CI, 15.3– 40.6; P ⬍ 0.0001). If patients had positive Rome I criteria and a combination of both a normal permeability and fecal calprotectin level, the OR for IBS was 46.1 (95% CI, 20.0 – 106.4; P ⬍ 0.0001).

Discussion This study shows both fecal calprotectin level and the Rome I criteria to be significantly better screening discriminates of patients with organic or nonorganic intestinal disease (OR, 27.8 and 13.3) than some other commonly used laboratory parameters, such as CRP (OR, 4.2) and ESR (OR, 3.2). In addition, the differential urinary excretion of lactulose/L-rhamnose is a good predictor of small intestinal disease (OR, 8.9). We have also

Table 4. OR, PPV, and NPV for the Different Laboratory Parameters for Organic Intestinal Disease Parameter Fecal calprotectin ⬎10 mg/L CRP ⬎5.0 mg/L ESR ⬎10 mm/h Permeability ratio ⬎0.05b Positive Rome criteriac aP

OR (95% CI) (17.6–43.7)a

27.8 4.2 (2.9–6.1)a 3.2 (2.2–4.6)a 8.9 (5.8–14.0)a 13.3 (8.9–20.0)a

⬍ 0.0001.

bOR/predictive cOR/predictive

values of the permeability ratio for organic small intestinal disease. values of the Rome criteria for IBS.

PPV

NPV

0.76 0.67 0.62 0.56 0.86

0.89 0.68 0.69 0.89 0.69

456

TIBBLE ET AL.

GASTROENTEROLOGY Vol. 123, No. 2

Table 5. Variable Parameters in the IBS (Nonorganic) and Organic Disease Diagnostic Groups

Parameters

Calprotectina (mg/L)

Permeability ratioa

ESRa (mm/h)

CRPa (mg/L)

Hemoglobina

Agea (y)

IBS (nonorganic) (n ⫽ 339) Organic disease (n ⫽ 263)

4b (0.5–50) 50b (0.5–12,800)

0.028c (0.005–0.216) 0.056c (0.012–0.83)

7c (1–86) 14c (1–94)

3c (0.3–63) 5c (0.3–200)

13.3c (7.9–17.4) 12.7c (7.1–18.1)

38 (13–90) 42 (16–88)

Rome criteria (no. positive/ no. negative) 240/99 39/224

aValues expressed as median (range). Permeability values in the organic disease group are for patients with small intestinal disease only (n ⫽ 129) because the permeability ratio is only applicable to small bowel disease. Values differ significantly (bP ⬍ 0.0001, cP ⬍ 0.01; Mann–Whitney U test) between groups.

shown that both the site and likelihood of organic disease can be reasonably predicted using a combination of a small intestinal permeability test and fecal calprotectin as an initial screening test in patients with symptoms that are compatible with both IBS and organic bowel disease. Up to 70% of persons with IBS do not seek medical attention.32–34 Despite this, IBS accounts for up to 12% of primary care consultations, 29% of whom are referred, and 28% of referrals to specialist gastroenterology practice.35,36 A significant proportion of those who are referred to gastroenterologists may have already been reassured that their symptoms are compatible with those of IBS but, failing to respond to a variety of therapies, these patients are referred for further reassurance and/or investigation. The lack of a gold standard for the diagnosis of IBS makes this type of study difficult. We have tried to firmly establish the diagnosis by various means; all patients underwent some form of invasive examination of the gastrointestinal tract, although the choice of procedure was determined by clinical judgment rather than using a standardized protocol. The 3-month follow-up of patients following classification with organic or nonorganic disease before statistical analysis was used to allow time to reclassify patients depending on symptoms and further investigation. The Rome I criteria were devel-

oped to aid physicians in the positive identification of patients with IBS in an attempt to develop a costeffective management strategy. It was hoped that this would reduce the number of invasive investigations performed to make what has traditionally been termed a diagnosis of exclusion. We have shown in accordance with others,2,4,9,10 who found sensitivities of 64%–90% and specificities of 55%–99% in similar cohorts of patients, that symptom-based criteria may be used to identify patients with a significant likelihood of having IBS (OR, 13.3; sensitivity, 85%; specificity, 71%). Despite recommendations regarding the use of such criteria in establishing a positive diagnosis at the first clinical encounter,1,37 their use is not universal and has largely been confined to use as entry criteria into research studies of patients with IBS.38 – 42 The concern for gastroenterologists is that, with the diverse symptomatology associated with IBS, a number of patients with organic intestinal disease will be incorrectly diagnosed if substantial reliance is placed on clinical criteria alone and therefore feel compelled to use invasive diagnostic investigations to increase diagnostic confidence. This concern is supported in part by our study, in which almost one third of all patients who were shown to have organic intestinal disease actually fulfilled the Rome I criteria for the diagnosis of IBS. It is reasonable to argue that patients with clinical red flags warrant invasive investigation irrespec-

Table 6. Combinations of Parameters in Distinguishing Between Diagnostic Groups Calprotectin ⬎10 mg/L plus permeability ratio ⬎0.05 for identification of patients with small intestinal disease Calprotectin ⬎10 mg/L plus permeability ratio ⬍0.05 for identification of colonic disease Calprotectin ⬍10 mg/L plus permeability ratio ⬍0.05 for identification of patients with nonorganic disease Calprotectin ⬍10 mg/L plus permeability ratio ⬍0.05 plus positive Rome criteria for identification of patients with nonorganic disease aP

⬍ 0.0001.

Sensitivity (%)

Specificity (%)

PPV

NPV

OR (95% CI)

55

93

0.70

0.74

15.0 (8.9–25.0)a

72

80

0.56

0.89

13.3 (8.3–21.3)a

69

92

0.92

0.70

25 (15.3–40.6)a

50

98

0.97

0.61

46 (20.0–106.4)a

August 2002

tive of laboratory parameters and that a more clinically practicable approach would be to assess the use of such parameters in subgroups of patients (low/medium risk of organic disease) following a clinical assessment and categorization by Rome criteria. In our study population, a surprisingly small number had red flags; most of these patients had ulcerative colitis. This low number may be due to our definition of rectal bleeding; if patients described blood confined to the tissue paper or dripping into the pan and were found to have hemorrhoids at rigid sigmoidoscopy, they were classified as red flag negative. In addition, our local referral policy for primary care physicians is that patients with rectal bleeding are referred to our colorectal surgeons, possibly causing a bias in the number of patients we see with red flags. Furthermore, some might argue that the list of red flags could be widened to include features such as nocturnal symptoms, family history, and so on. We have shown in a subgroup analysis that the sensitivity and specificity of calprotectin is unaffected when used in a 3-tiered system of investigation of patients believed to be at low or medium risk of organic disease. If used in the low-risk group, it correctly identifies 90% of those with organic disease, detecting an additional 22 patients (55%) compared with ESR and CRP. If used in the medium-risk group, it correctly identifies 77% of those who have nonorganic disease with a sensitivity for organic disease of 90%. Therefore, in the medium-risk group, fecal calprotectin may help the clinician determine the symptomatic threshold at which patients warrant invasive investigation. Fecal calprotectin is a sensitive marker of acute intestinal inflammation due to a wide variety of causes, and levels have been shown to correlate with intestinal inflammatory activity as assessed by 4-day fecal 111Indiumlabeled leukocyte studies.23,43 It identifies patients with IBD (in whom levels may be used to predict impending relapse of disease),17,43,44 patients with NSAID enteropathy,23 and patients with colorectal carcinoma/adenoma,19,22,45 in all of which there is increased neutrophil recruitment to the inflamed or ulcerated intestinal mucosa. To date, no study has assessed the use of fecal calprotectin in differentiating patients with commonly encountered organic intestinal diseases from those with IBS. The potential strength of fecal calprotectin assessment is that it is a direct measure of mucosal inflammatory activity that may be detected at a level insufficient to cause an increase in ESR and CRP, and levels in the stool seem to be unaffected by a variety of nonintestinal conditions, which may result in an elevation of the systemic

ORGANIC VERSUS NONORGANIC INTESTINAL DISEASE

457

inflammatory markers.23 This is reflected in the greater sensitivity, specificity, PPV, and NPV of fecal calprotectin compared with those for ESR and CRP (Tables 3 and 4). Both of the latter, when used in isolation, give significant ORs for predicting organic disease (ESR, 3.2; CRP, 4.2) but are inferior compared with fecal calprotectin (OR, 27.8). When assessed by the multiple logistic regression model, both ESR and CRP provide no additional discriminatory value over and above calprotectin and Rome I criteria in distinguishing between patients with organic and nonorganic disease. Patients with diabetic diarrhea were the one organic disease group that was not identified by the fecal calprotectin test. The pathogenesis of diabetic diarrhea is incompletely understood but likely to involve alterations in intestinal transit times as an important cause of symptoms in a substantial number of patients. If so, the inflammatory component of the syndrome may be absent or of insufficient severity to affect fecal calprotectin levels, and a case could be made for including this group within the nonorganic disease group for the purposes of this study. The same scenario might apply to patients with postinfectious IBS, in which the time frame between the infective inflammatory episode and investigation for IBS may affect the result of the fecal calprotectin test, such that, as in all cases, the clinician’s judgment regarding the degree of invasive investigation remains paramount. Patients with diverticular disease were included in the organic disease group because 10%–25% of such patients have an inflammatory component to the disease, as evidenced by an acute inflammatory infiltrate seen in peridiverticular biopsy specimens. Because biopsy specimens are not taken as a matter of routine from these patients at colonoscopy, we were unable to discriminate between those with and without active inflammation. If the 14 patients with diverticular disease are included in the nonorganic disease group, the ORs for the detection of organic disease using fecal calprotectin do not change significantly (OR, 26.5; 95% CI, 16.5– 42.4). We also did not exclude patients taking NSAIDs from our study. NSAID enteropathy was diagnosed in 13 patients (40% of those taking regular NSAIDs), in keeping with prevalence rates from previous studies using a variety of diagnostic techniques.23,46,47 Intestinal permeability is abnormal in a wide variety of conditions affecting the small intestine.24 This may be a cause or consequence of inflammation.27,48 We have shown differential urinary excretion of lactulose/L-rhamnose to have a sensitivity for organic small intestinal disease of 63% with a specificity of 87%. The permeability test had a lower sensitivity for small intestinal

458

TIBBLE ET AL.

disease than the fecal calprotectin test (sensitivity of 86%). The use of the intestinal permeability test in conjunction with a fecal calprotectin test can nevertheless be helpful in determining the site of intestinal inflammation. Patients in whom the fecal calprotectin level is elevated with an increase in intestinal permeability have an OR of small intestinal disease of 15.0 (Table 6). Combining the results of these diagnostic tests does not compromise the sensitivity or PPV (which is improved) in identifying patients with small intestinal disease and enhances the specificity when compared with the use of permeability in isolation. The significance of an elevated intestinal permeability with a normal calprotectin level is unclear. Although this could represent a level of intestinal barrier disruption of insufficient intensity to cause inflammation, it is more likely due to the fact that our cutoff limit of 0.05 is defined by the 95% CI of normal values and therefore 5% of apparently healthy individuals will be outside the normal range. Patients with an elevated fecal calprotectin level and normal intestinal permeability have an OR for colonic disease of 13.3, whereas normality of both tests gives an OR for nonorganic disease of 25. It should be emphasized that such combinations of test results are useful only in directing the clinician toward investigation of the small or large bowel and are not a diagnostic test. Finally, if this last combination of the permeability and calprotectin tests is combined with fulfillment of the Rome I criteria, the OR of IBS is 46, which is an exceptionally high likelihood associated with both a high specificity and PPV. Although this latter combination is associated with a sensitivity of 50%, it should be remembered that these are not diagnostic tests for IBS but rather that normality infers an absence of organic disease. By combining the results of 3 different tests (calprotectin, permeability, and Rome criteria) in a predictive model, a variance from the model of any of the 3 tests will have an adverse effect on the NPV. Hence, although this combination is only associated with an NPV of 61% (i.e., 49% of patients who do not fit the model will have organic disease), it is not being used to make a diagnosis of organic disease, which is better done using a calprotectin result alone. In addition, although predictive diagnostic models such as these are useful in helping clinicians decide how extensively to investigate patients, their ultimate decision will be influenced by clinical factors that may not be part of the Rome criteria, as may be the case for patients who fit into a high-risk group for colorectal cancer based on age or family history. The patients studied were those referred to a specialist gastroenterology unit, including those believed to have

GASTROENTEROLOGY Vol. 123, No. 2

refractory IBS, a proportion of whom will inevitably be found to have organic disease. Although this may have an impact on the results if applied to a wider population, it may be that patients with milder symptoms of IBS who are not referred may be less likely to have organic disease and even more likely to have a negative screening test, thereby improving the PPV and NPV of the Rome criteria and calprotectin/permeability tests. We do not propose the use of the fecal calprotectin test as an initial screen in a primary care setting until such a population has been studied, in which the different cutoff values would have to be validated, but our results support its use in the algorithm for the workup of suspected nonorganic disease13 following specialist referral. If the calprotectin method is considered a “screening” procedure for a similar group of patients as described in this study, the questions of practicality and cost arise. All the patients in this study provided a sample without major difficulty, and a recent audit of our services (in which patients referred as gastroenterology outpatients are sent instructions and materials to perform the test before consultation) showed a 96% compliance. The cost of the assay is more difficult to assess because local pay scales for laboratory technicians vary. However, the calprotectin assay is available as a standard ELISA (Calprotech Ltd., Oslo, Norway) that can be performed in any clinical diagnostics laboratory. The cost of the calprotectin assay ($500 for 40 patients) would seem to compare favorably with many other diagnostic inflammatory parameters used in the investigation of patients with gastrointestinal disease. In comparison with both endoscopy and contrast radiography, the use of the Rome criteria, intestinal permeability, and fecal calprotectin tests provides a noninvasive, safe, and effective means of screening patients for the presence of organic intestinal disease. Their combined use may help provide clinicians with objective evidence for the presence or absence of organic disease and aid in deciding which invasive investigations to request (small or large bowel studies) or possibly avoid in cases in which these parameters indicate that the diagnosis is IBS.

References 1. Drossman DA, Whitehead WE, Camilleri M. Irritable bowel syndrome: a technical review for practice guideline development. Gastroenterology 1997;112:2120 –2137. 2. Manning AP, Thompson WG, Heaton KW, Morris AF. Towards positive diagnosis of the irritable bowel. Br Med J 1978;6138: 653– 664. 3. Drossman DA, Sandler RS, McKee DC, Lovitz AJ. Bowel patterns among subjects not seeking health care. Use of a questionnaire to identify a population with bowel dysfunction. Gastroenterology 1982;83:529 –534.

August 2002

4. Kruis W, Thieme C, Weinzierl M, Schussler P, Holl J, Paulus W. A diagnostic score for the irritable bowel syndrome. Its value in the exclusion of organic disease. Gastroenterology 1984;87:1–7. 5. Whitehead WE, Bosmajian L, Zonderman AB, Costa PT Jr, Schuster MM. Symptoms of psychologic distress associated with irritable bowel syndrome. Comparison of community and medical clinic samples. Gastroenterology 1988;95:709 –714. 6. Talley NJ, Phillips SF, Melton J, Wiltgen C, Zinsmeister AR. A patient questionnaire to identify bowel disease. Ann Intern Med 1989;111:671– 674. 7. Talley NJ, Phillips SF, Melton J, Mulvihill C, Wiltgen C, Zinsmeister AR. Diagnostic value of the Manning criteria in irritable bowel syndrome. Gut 1990;31:77– 81. 8. Smith RC, Greenbaum DS, Vancouver JB, Henry RC, Reinhart MA, Greenbaum RB, Dean HA, Mayle JE. Gender differences in Manning criteria in the irritable bowel syndrome. Gastroenterology 1991;100:591–595. 9. Thompson WG, Drossman DA, Richter JE, Talley NJ, Corazziari E, Whitehead WE. Functional bowel disorders and functional abdominal pain. In: Drossman DA, ed. Functional gastrointestinal disorders: diagnosis, pathophysiology and treatment. Boston: Little, Brown, 1994:115–173 10. Appendix B. In: Drossman DA, ed. Functional gastrointestinal disorders: diagnosis, pathophysiology and treatment. Boston: Little, Brown, 1994:339 –345 11. Harris MS. Irritable bowel syndrome. A cost-effective approach for primary care physicians. Postgrad Med 1997;101:215–216, 219 –220, 223–226. 12. Camilleri M, Prather CM. The irritable bowel syndrome: mechanisms and a practical approach to management. Ann Intern Med 1992;116:1001–1008. 13. Drossman DA, Thompson WG. The irritable bowel syndrome: review and a graduated multicomponent treatment approach. Ann Intern Med 1992;116:1009 –1016. 14. Tolliver BA, Herrera JL, DiPalma JA. Evaluation of patients who meet clinical criteria for irritable bowel syndrome. Am J Gastroenterol 1994;89:176 –178. 15. Dale I, Brandtzaeg P, Fagerhol MK, Scott H. Distribution of a new myelomonocytic antigen (L1) in human peripheral blood leukocytes. Immunofluorescence and immunoperoxidase staining features in comparison with lysozyme and lactoferrin. Am J Clin Pathol 1985;84:24 –34. 16. Fagerhol MK, Andersson KB, Naess-Andresen CF, Brandtzaeg P, Dale I. Calprotectin (the L1 leukocyte protein). In: Smith VL, Dedman JR, eds. Stimulus response coupling: the role of intracellular calcium-binding proteins. Boca Raton, FL: CRC, 1990: 187–210. 17. Roseth AG, Aadland E, Jahnsen J, Raknerud N. Assessment of disease activity in ulcerative colitis by faecal calprotectin, a novel granulocyte marker protein. Digestion 1997;58:176 –180. 18. Roseth AG, Fagerhol MK, Aadland E, Schjonsby H. Assessment of the neutrophil dominating protein calprotectin in feces. A methodologic study. Scand J Gastroenterol 1992;27:793–798. 19. Kristinsson J, Roseth A, Fagerhol MK, Aadland E, Schjonsby H, Bormer OP, Raknerud N, Nygaard K. Fecal calprotectin concentration in patients with colorectal carcinoma. Dis Colon Rectum 1998;41:316 –321. 20. Meling TR, Aabakken L, Roseth A, Osnes M. Faecal calprotectin shedding after short-term treatment with non-steroidal anti-inflammatory drugs. Scand J Gastroenterol 1996;31:339 –344. 21. Teahon K, Roseth A, Foster R, Bjarnason I. Faecal calprotectin: a simple sensitive quantitative measure of intestinal inflammation in man (abstr). Gastroenterology 1997;112:A1103. 22. Tibble J, Sigthorsson G, Foster R, Sherwood R, Fagerhol M, Bjarnason I. Faecal calprotectin and faecal occult blood tests in the diagnosis of colorectal carcinoma and adenoma. Gut 2001; 49:402– 408.

ORGANIC VERSUS NONORGANIC INTESTINAL DISEASE

459

23. Tibble JA, Sigthorsson G, Foster R, Scott D, Fagerhol MK, Roseth A, Bjarnason I. High prevalence of NSAID enteropathy as demonstrated by a simple faecal test. Gut 1999;45:362–366. 24. Bjarnason I, MacPherson A, Hollander D. Intestinal permeability: an overview. Gastroenterology 1995;108:1566 –1581. 25. Drossman DA. The functional gastrointestinal disorders and the Rome II process. Gut 1999;45(Suppl 2):1–5. 26. Sigthorsson G, Tibble J, Hayllar J, Menzies I, Macpherson A, Moots R, Scott D, Gumpel MJ, Bjarnason I. Intestinal permeability and inflammation in patients on NSAID’s. Gut 1998;43:506 – 511. 27. Teahon K, Smethurst P, Levi AJ, Menzies IS, Bjarnason I. Intestinal permeability in patients with Crohn’s disease and their first degree relatives. Gut 1992;33:320 –323. 28. Henderson AR. Assessing test accuracy on its clinical consequence: a primer for receiver operating characteristics curve analysis. Ann Clin Biochem 1993;30:521–539. 29. Royston P. Estimating departure from normality. Stat Med 1991; 10:1283–1294. 30. Cox DR. Regression models and life tables. J Royal Stat Soc 1972;B34:187–202. 31. Morris AJ, Madhok R, Sturrock RD, Capell HA, MacKenzie JF. Enteroscopic diagnosis of small bowel ulceration in patients receiving non-steroidal anti-inflammatory drugs. Lancet 1991; 337:520. 32. Drossman DA, McKee DC, Sandler RS, Mitchell CM, Cramer EM, Lowman BC, Burger AL. Psychosocial factors in the irritable bowel syndrome. A multivariate study of patients and nonpatients with irritable bowel syndrome. Gastroenterology 1988;95:701–708. 33. Thompson WG, Heaton KW. Functional bowel disorders in apparently healthy people. Gastroenterology 1980;79:283–288. 34. Bommelaer G, Rouch M, Dapoigny M, Pais D, Loisy P, Gualino M, Tournut R. Epidemiology of intestinal functional disorders in an apparently healthy population. Gastroenterol Clin Biol 1986;10: 7–12. 35. Mitchell CM, Drossman DA. Survey of the AGA membership relating to patients with functional gastrointestinal disorders. Gastroenterology 1987;92:1282–1284. 36. Thompson WG, Heaton KW, Smyth GT, Smyth C. Irritable bowel syndrome in general practice: prevalence, characteristics and referral. Gut 2000;46:78 – 82. 37. Thompson WG. Irritable bowel syndrome: pathogenesis and management. Lancet 1993;341:1569 –1572. 38. Bennett EJ, Tennant CC, Piesse C, Badcock CA, Kellow JE. Level of chronic life stress predicts clinical outcome in irritable bowel syndrome. Gut 1998;43:256 –261. 39. Shaw M, Talley NJ, Adlis S, Beebe T, Tomshine P, Healey M. Development of a digestive health status instrument: tests of scaling assumptions, structure and reliability in a primary care population. Aliment Pharmacol Ther 1998;12:1067–1078. 40. Kay L, Jorgensen T, Lanng C. Irritable bowel syndrome: which definitions are consistent? J Intern Med 1998;244:489 – 494. 41. Bensoussan A, Talley NJ, Hing M, Menzies R, Guo A, Ngu M. Treatment of irritable bowel syndrome with Chinese herbal medicine: a randomized controlled trial. JAMA 1998;280:1585– 1589. 42. Patrick DL, Drossman DA, Frederick IO, DiCesare J, Puder KL. Quality of life in persons with irritable bowel syndrome: development and validation of a new measure. Dig Dis Sci 1998;43: 400 – 411. 43. Tibble JA, Teahon K, Roseth A, Sigthorsson G, Bridger S, Foster R, Sherwood R, Fagerhol M, Bjarnason I. A simple method for assessing intestinal inflammation in Crohn’s disease. Gut 2000; 47:506 –513. 44. Tibble JA, Sigthorsson G, Bridger S, Fagerhol M, Bjarnason I. Surrogate markers of intestinal inflammation are predictive of

460

45.

46.

47.

48.

TIBBLE ET AL.

relapse in patients with quiescent inflammatory bowel disease. Gastroenterology 2000;119:15–22. Gilbert JA, Ahlquist DA, Mahoney DW, Zinsmeister AR, Rubin J, Ellefson RD. Fecal marker variability in colorectal cancer: calprotectin versus hemoglobin. Scand J Gastroenterol 1996;31: 1001–1005. Allison MC, Howatson AG, Torrance CJ, Lee FD, Russell RI. Gastrointestinal damage associated with the use of nonsteroidal antiinflammatory drugs. N Engl J Med 1992;327:749 –754. Morris AJ, Wasson LA, MacKenzie JF. Small bowel enteroscopy in undiagnosed gastrointestinal blood loss. Gut 1992;33:887– 889. Somasundaram S, Hayllar H, Rafi S, Wrigglesworth JM, Macpherson AJ, Bjarnason I. The biochemical basis of non-steroidal antiinflammatory drug-induced damage to the gastrointestinal tract:

GASTROENTEROLOGY Vol. 123, No. 2

a review and a hypothesis. Scand J Gastroenterol 1995;30:289 – 299.

Received August 29, 2001. Accepted May 9, 2002. Address requests for reprints to: Jeremy A. Tibble, M.D., M.R.C.P., Digestive Diseases Centre, Level 9A, Royal Sussex County Hospital, Eastern Road, Brighton, East Sussex BN2 5BE, England. e-mail: [email protected]. Supported by NHS Executive South Thames Regional Office grant RDF 039 (to J.A.T.). The authors thank Prof. Magne Fagerhol (Department of Immunology, Ulleval University Hospital, Oslo, Norway) for supplying the anticalprotectin antibody for use in the enzyme-linked immunosorbent assay and for his technical support throughout the project.