Portal de Periódicos da CAPES

Evaluation of the Diagnostic Accuracy of Plasma Markers for Early Diagnosis in Patients Suspected for Acute Appendicitis

2013; Wiley; Volume: 20; Issue: 7 Linguagem: Inglês

10.1111/acem.12160

1553-2712

Dirk H. S. M. Schellekens, Karel W. E. Hulsewé, B.A.C. van Acker, Annemarie A. van Bijnen, Tom M. H. de Jaegere, S H Sastrowijoto, Wim A. Buurman, Joep P. M. Derikx,

Abdominal Trauma and Injuries

The main objective of this study was to evaluate the diagnostic accuracy of two novel biomarkers, calprotectin (CP) and serum amyloid A (SAA), along with the more traditional inflammatory markers C-reactive protein (CRP) and white blood cell count (WBC), in patients suspected of having acute appendicitis (AA). The secondary objective was to compare diagnostic accuracy of these biomarkers with a clinical scoring system and radiologic imaging. A total of 233 patients with suspected AA, presenting to the emergency department (ED) between January 2010 and September 2010, and 52 healthy individuals serving as controls, were included in the study. Blood was drawn and CP and SAA−1 concentrations were measured using enzyme-linked immunosorbent assay (ELISA). CRP and WBC concentrations were routinely measured and retrospectively abstracted from the electronic health record, together with physical examination findings and radiologic reports. The Alvarado score was calculated as a clinical scoring system for AA. Final diagnosis of AA was based on histopathologic examination. The Mann-Whitney U-test was used for between-group comparisons. Receiver operating characteristic (ROC) curves were used to measure the diagnostic accuracy for the tests and to determine the best cutoff points. Seventy-seven of 233 patients (33%) had proven AA. Median plasma levels for CP and SAA−1 were significantly higher in patients with AA than in those with another final diagnosis (CP, 320.9 ng/mL vs. 212.9 ng/mL; SAA−1, 30 mg/mL vs. 0.6 mg/mL; p < 0.001). CRP and WBC were significantly higher in patients with AA as well. The Alvarado score was helpful at the extremes ( 7). Ultrasound (US) had a sensitivity of 84% and a specificity of 94%. Computed tomography (CT) had a sensitivity of 100% and a specificity of 91%. The area under the ROC (95% confidence interval [CI]) was 0.67 (95% CI = 0.60 to 0.74) for CP, 0.76 (95% CI = 0.70 to 0.82) for SAA, 0.71 (95% CI = 0.64 to 0.78) for CRP, and 0.79 (95% CI = 0.73 to 0.85) for WBC. No cutoff points had high enough sensitivity and specificity to accurately diagnose AA. However, a high sensitivity of 97% was shown at 7.5 × 109/L for WBC and 0.375 mg/mL for SAA. CP, SAA−1, CRP, and WBC were significantly elevated in patients with AA. None had cutoff points that could accurately discriminate between AA and other pathology in patients with suspected AA. A WBC < 7.5 × 109/L, with a low level of clinical suspicion for AA, can identify a subgroup of patients who may be sent home without further evaluation, but who should have available next-day follow-up. Evaluación de la Certeza Diagnóstica de Marcadores Plasmáticos para el Diagnóstico Precoz en los Pacientes con Sospecha de Apendicitis Aguda El principal objetivo de este estudio fue evaluar la certeza diagnóstica en los pacientes con sospecha de tener una apendicitis aguda de dos nuevos biomarcadores, la calprotectina (CP) y el amiloide A sérico (AAS), junto con los marcadores inflamatorios más tradicionales, como la proteína C reactiva (PCR) y el número de leucocitos. El objetivo secundario fue comparar la certeza diagnóstica de estos biomarcadores con un sistema de puntuación clínica y la imagen radiológica. Se incluyeron en este estudio 233 pacientes con sospecha de apendicitis que acudieron al servicio de urgencias (SU) entre enero de 2010 y septiembre de 2010, y 52 individuos sanos que sirvieron como controles. Se extrajo sangre y se midió la CP y el AAS mediante un ELISA. La PCR y el número de leucocitos se midieron de forma rutinaria y se recogieron de forma retrospectiva de las historias clínicas electrónicas, junto con los hallazgos de la exploración física y los informes radiológicos. La escala Alvarado (Alvarado score) se calculó como un sistema de puntuación clínica para la apendicitis aguda. El diagnóstico final de apendicitis aguda se basó en el examen histopatológico. Se utilizó el test de la U de Mann-Whitney para la comparación entre los grupos. Las curvas ROC se utilizaron para medir la certeza diagnóstica de las pruebas diagnósticas y para determinar los mejores puntos de corte. Setenta y siete de 233 pacientes (33%) tuvieron apendicitis aguda con certeza. Las medianas de los niveles plasmáticos de CP y AAS fueron significativamente más altas en los pacientes con apendicitis aguda que en aquellos con otro diagnóstico final (CP 321 ng/mL vs. 213 ng/mL; AAS 30 mg/mL vs. 0,6 mg/mL; p < 0,001). La PCR y los leucocitos estaban también significativamente más elevados en los pacientes con apendicitis aguda. La escala de Alvarado fue de ayuda en los extremos (< 3 o > 7). La ecografía tuvo una sensibilidad de un 84% y una especificidad de un 94%. La tomografía computarizada (TC) tuvo una sensibilidad de un 100% y una especificidad de un 91%. La curva ROC (intervalo de confianza 95%) fue de 0,67 (IC 95% = 0,60 a 0,74) para la CP, de 0,76 (IC 95% = 0,70 a 0,82) para la AAS, de 0,71 (IC 95%= 0,64 a 0,78) para la PCR y de 0,79 (IC 95%= 0,73 a 0,85) para los leucocitos. La CP, la AAS, la PCR y el número de leucocitos estaban significativamente elevados en los pacientes con apendicitis aguda. No hay un punto de corte que pueda discriminar de forma certera entre la apendicitis aguda y otra patología en los pacientes con sospecha de apendicitis aguda. Un número de leucocitos < 7,5 × 109/L, con un bajo nivel de sospecha clínica para la apendicitis aguda, puede identificar un subgrupo de pacientes que puede ser enviado a casa sin más evaluaciones, aunque estos pacientes deberían disponer de un seguimiento al día siguiente. Acute appendicitis (AA) is the most common abdominal emergency worldwide, with an overall lifetime risk of 8.6% in males and 6.7% in females.1 It remains a challenge to make the correct diagnosis, mainly because the clinical presentation may be nonspecific, and there is a lack of adequate specific biomarkers for AA.2 Frequently used laboratory tests for the diagnosis of AA are white blood cell count (WBC) and C-reactive protein (CRP). Leukocytosis appears in 70% to 90% of the patients with AA. However, the specificity is low, because other acute abdominal complaints are also frequently associated with leukocytosis.3, 4 Elevated CRP may be an indicator of appendiceal perforation or abscess formation in patients suspected for AA.5 The nonspecific presentation and limited diagnostic accuracy of laboratory testing often leads to an inconclusive diagnosis, necessitating imaging techniques of the abdomen. Ultrasound (US) and computed tomography (CT) scanning can improve diagnostic accuracy in patients with acute abdominal pain, as was recently reported in the OPTIMA trial.6, 7 US gives good results in the hands of an experienced operator.8 Expertise for US, on the other hand, is not readily available at all times in all emergency departments (EDs). CT scanning has a high accuracy rate for diagnosing AA. However, clinicians should be reluctant to use CT because of the radiation exposure to patients.9-11 Furthermore, the use of imaging techniques takes precious time in the ED and radiology department. In conclusion, there is a need for accurate plasma markers, which can streamline the diagnosis and reduce the requirement for abdominal imaging. Several recent studies have explored the role of plasma markers in improving the diagnosis of appendicitis in both children and adults.12-17 Two markers that have shown promising results in diagnosing AA are calprotectin (CP) and serum amyloid A (SAA) protein. The pathophysiology of AA is characterized by an increased permeability of the mucosal barrier, leading to extensive influx of activated neutrophils into the gut wall.18 Lactoferrin (LF) and CP are released upon neutrophil activation and can be detected in the systemic circulation. In a pilot study, the usefulness of both CP and LF for diagnosing AA was analyzed.15 CP, a cytosolic protein constituting 60% of soluble proteins in human neutrophil granulocytes,19 was most accurate in differentiating patients with AA from healthy individuals. Moreover, it gave additional information to the conventional markers WBC and CRP, making it an interesting new marker for the diagnosis of AA.15 SAA protein is a family of proteins synthesized in response to inflammation-related cytokines such as interleukin (IL)-1, IL-6, and tumor necrosis factor-α. There are several isoforms of SAA; we used the isoform SAA-1, which increases within 2 to 3 hours after activation of the immune system and returns to normal levels at 5 to 7 days.20 Circulating SAA levels were better able to discriminate between children with AA and those without than WBC and CRP.13 The aim of this study was to further evaluate the diagnostic accuracy of these markers (CP and SAA-1), as well as that of the classical inflammatory markers (CRP, WBC), in patients with suspected AA. We also studied cutoff points that potentially could be used to diagnose or exclude appendicitis. We further compared the diagnostic values of the biomarkers with the clinical findings and radiologic imaging. This was a phase 3 diagnostic study in patients with acute abdominal complaints, suspected for appendicitis. We obtained a signed informed consent from all patients, parents, or caretakers. The study was approved by the local ethics committee and carried out according to the revised declaration of Helsinki. The principles of good clinical practices were followed during this study. This study was conducted at a large teaching hospital in the Netherlands (Orbis Medical Center, Sittard-Geleen, the Netherlands). Between January 2010 and September 2010, consecutive patients presenting to the ED with acute abdominal pain of less than 48 hours duration who were clinically suspected of having AA as determined by their general practitioners (GPs) were enrolled in this study. In the Netherlands, patients routinely visit their GPs first in the case of most health problems.21 The GPs control access to specialized medical care and triage patients before they enter the ED. The GPs based their clinical suspicion of AA on several variables, including clinical status, disease history, and physical examination. Patients were excluded from the study if they were pregnant, had radiologic imaging of the abdomen prior to ED admission, or had a history of abdominal trauma within 7 days of presentation. Patients were evaluated following a standard diagnostic procedure of history taking, physical examination, and routine WBC and CRP laboratory tests. According to the OPTIMA study for acute abdominal complaints,7 a patient only underwent surgery after imaging of the abdomen was performed using US or CT. The choice of imaging technique was based on the attending surgeon's preference. The attending radiologist and surgeon interpreted all CT scans. All patients diagnosed with AA underwent appendectomies following their diagnostic work-ups. When a patient was considered at low risk for AA, imaging was deemed unnecessary and the patient was seen the next day at the outpatient clinic. Two members of the research team (DS, JD) collected all necessary clinical and radiologic information from the medical files of all participating patients. The records of every fifth enrolled patient were randomly checked by a third independent research-member (KH). The participating hospital has over 10 years of experience with an electronic health record system, in which all patient medical information is collected. There were no missing data. The clinical information was used to retrospectively calculate the Alvarado score.22 The Alvarado score is a 10-point clinical scoring system, based on history, physical examination, and two laboratory investigations; it is often used to help risk-stratify patients with clinical suspicion of appendicitis.23 The Alvarado score was not applied in the clinical management of these patients, but we chose to include it in order to compare a well-known and validated clinical scoring system to the performance of biomarkers. From each patient an extra 5 mL of blood was collected in an EDTA tube (BD Vacutainer, Becton Dickinson Diagnostics, Breda, the Netherlands) in the ED. These samples were immediately centrifuged after collection for 12 minutes at 2100 rpm and cooled down to 5°C. Laboratory technicians then directly removed the plasma portion from each EDTA tube and transferred it into multiple cryopreservation tubes and froze the samples at −20°C until analysis. Calprotectin and SAA-1 plasma concentrations were measured using commercially available enzyme-linked immunosorbent assay (ELISA) kits (Hbt, Uden, The Netherlands). The ELISA kits are based on the sandwich principle with a working time of 3.5 hours. The detection limits for CP and SAA-1 were 10 and 3.1 ng/mL, respectively. Samples were assayed according to the manufacturer's recommended procedures by trained laboratory technicians at the specialized laboratory facilities of the Department of General Surgery. The laboratory personnel were unaware of the final diagnosis and of the CRP and WBC values. Samples were run in duplicate, and a variability of 5% between sample duplicates was accepted. CRP and WBC concentrations were determined in standard fashion by the laboratory of clinical chemistry and hematology. The primary outcome was the diagnostic accuracy of CP and SAA−1 to identify patients with appendicitis. Secondary outcome measures were the diagnostic value of plasma CRP concentration, WBC, the Alvarado score, and radiologic imaging findings. In addition, we analyzed the receiver operating characteristic (ROC) curve of the laboratory tests to determine cutoff points that potentially could be used to diagnose or exclude appendicitis. The final diagnosis of AA was based on histopathologic examination of the appendix as the gold standard. AA was histologically demonstrated by infiltration of the mucosa of the appendix by neutrophilic granulocytes, with or without local peritonitis. Quantification of the neutrophilic infiltrate was not recorded because this was not the purpose of this study. Diagnosis of a perforated appendicitis was based on review of the attending surgeon's written postoperative report. All patients without initial diagnoses of AA were successfully followed up at 24 hours and then by telephone at 3 months. For this study, diagnoses other than AA were categorized as other acute abdominal pathology, nonacute abdominal complaints, negative appendix, and urologic and gynecologic pathology. Statistical analysis was performed using GraphPad Prism 5 for Windows (GraphPad Software Inc., San Diego, CA) and SPSS 17.0 for Windows (IBM SPSS, Armonk, NY). Based on the CP data of a previous study,15 a power analysis was performed. The number of patients needed for this study was calculated using the difference in CP levels between controls and patients with AA15 (α = 0.05, 1 − β = 0.90). This produced a minimum number of 220 patients suspected of having appendicitis. To obtain reference values for the plasma markers CP and SAA−1, 52 healthy individuals undergoing elective orthopedic operations were included. No formal sample size calculation was conducted for the number of healthy individuals. Normality was tested using Kolmogorov-Smirnov test. None of the laboratory tests were normally distributed. For comparison of CP, SAA−1, CRP, and WBC between the group with AA and the other diagnoses, a two-tailed Mann-Whitney U-test was used. For comparisons between AA and each of the separate diagnostic categories, a Kruskal-Wallis test with a Dunn's post hoc test was performed. All data are presented as medians with interquartile ranges (IQRs). With respect to the categorical data, Fisher's exact tests were performed, with a Bonferroni correction when comparing all other four categories separately with the AA group as reference. Receiver operating characteristic (ROC) curves were used to study diagnostic accuracy of the markers in patients with AA, represented by area under the curve (AUC), with an AUC of 0.5 indicating no discrimination ability and an AUC of 1.0 indicating maximal discrimination ability. The ideal cutoff points to discriminate patients with or without AA among patients suspected of AA were assessed as maximum sum of sensitivity and specificity. The combined diagnostic value of the variable markers was studied using a direct multivariable logistic regression model.24 The guidelines for accurate logistic regression modeling were followed in this study.25 The continuous variables were checked for the absence of influential multicollinearity, lack of strongly influential outlier values, and the assumption of linearity in the logit. The overall model performance was measured by the Hosmer-Lemeshow goodness-of-fit test. Different markers were entered as continuous variables and the presence of appendicitis as a categorical dependent variable (patients with AA were coded 1, and other diagnoses were coded with 0). Next, the predictive probabilities were calculated. The diagnostic accuracy of different multimarker combinations was determined using ROC curves (with the predictive probabilities) to calculate the AUCs. A p-value below 0.05 was considered statistically significant. The STARD statement for reporting studies of diagnostic accuracy was used in this study.26 A total of 233 consecutive patients seen in the ED with suspected appendicitis (101 males, 132 females; mean age = 30.3 years, range = 5 to 83 years) were enrolled. Eighty-six patients underwent appendectomies, 77 (33% of the total population) had histopathologically proven AA, of whom 16 had perforated appendicitis. The remaining 156 patients were divided into four diagnostic categories: other acute abdominal pathology (n = 15), nonacute abdominal pathology (n = 107), negative appendix (n = 9), and urologic and gynecologic pathology (n = 25). Patient and disease characteristics are shown in Table 1. Nonperforated appendicitis (n = 61) Perforated appendicitis (n = 16) Negative appendix (n = 9) • Irritable bowel syndrome (n = 2) Constipation (n = 1) • Gastroenteritis (n = 1) • Acute aspecific abdominal pain (n = 5) Other acute abdominal pathology (n = 15) • Diverticulitis (n = 12) • Small bowel ischemia (n = 1) • Diaphragmatic hernia (n = ) • Omental torsion (n = 1) Nonacute abdominal pathology (n = 107) • Gastroenteritis (n = 26) • Irritable bowel syndrome (n = 13) • Constipation (n = 21) • Mesenteric lymphadenitis (n = 8) • Acute aspecific abdominal pain (n = 31) • Pancreatitis (n = 1) • Crohn's disease (n = 2) • Diverticulosis (n = 1) • Pyrosis (n = 2) • Cholecystolithiasis (n = 1) • Pneumonia (n = 1) Urologic and gynecologic pathology (n = 25) • Adnexitis (n = 2) • Urolithiasis (n = 7) • Pelvic inflammatory disease (n = 2) • Ovarian tumor (n = 2) • Ovarian abscess (n = 1) • Adnexal torsion (n = 1) • Urinary tract infection (n = 4) • Prostatitis (n = 1) • Pylonephritis (n = 3) • Ovulation bleeding (n = 2) Of the 233 patients, 168 received imaging of the abdomen. In 123 patients, US was the imaging modality used, and 60 patients underwent CT scans. In 15 cases the US was inconclusive and was followed by CT. Prior to surgery, all patients underwent either US or CT scan. None of the patients who did not undergo surgery at their initial ED visit required surgery following their discharge over a 3 month follow up period. Median plasma concentrations for CP were significantly higher in the 77 patients with proven AA (320.9 ng/mL, IQR = 193.4 to 492.4 ng/mL) than in either the 156 patients with other diagnoses (212.9 ng/mL, IQR = 104.3 to 323.0 ng/mL) or the 52 healthy control patients (166.4 ng/mL, IQR = 91.5 to 299.2 ng/mL; p < 0.001). Between-group analyses showed significant differences between the patients with AA and those with nonacute abdominal pathology (p < 0.05; see Data Supplement S1, available as supporting information in the online version of this paper). Serum amyloid A−1 protein plasma concentrations were significantly higher in the 77 patients with AA (30.7 mg/mL, IQR = 2.9 to 129.6 mg/mL) than in either the 156 patients with other diagnoses (0.6 mg/mL, IQR = 0.3 to 17.3 mg/mL) or the 52 healthy control patients (0.3 mg/mL, IQR = 0.1 to 0.6 mg/mL; p < 0.001). Significantly elevated concentrations in the AA group were observed compared to the nonacute abdominal pathology group, the urologic and gynecologic pathology group, and the healthy control group (p < 0.05; see Data Supplement S1). The median plasma concentrations of CRP and WBC were also significantly higher in patients with AA than in patients with other diagnoses (CRP—24 mg/L, IQR = 6.5 to 65.5 mg/L vs. 4.0 mg/L, IQR = 1.0 to 15.5 mg/L, p < 0.001; WBC—14.4 × 109/L, IQR = 11.4 to 16.8 × 109/L vs. 8.9 × 109/L, IQR = 6.8 to 12.0 × 109/L, p < 0.001). Between-group comparisons showed significant differences in the plasma concentrations between patients with AA versus the urologic and gynecologic pathology group and AA versus the nonacute abdominal pathology group (p < 0.05; see Data Supplement S1). The Alvarado scores were significantly different between patients with AA (8 of 10, IQR = 5.0 to 9.0) and patients with other diagnoses (5 of 10, IQR = 3.0 to 6.0). Between-group comparisons showed a significant difference between AA versus urologic and gynecologic pathology and AA versus nonacute abdominal complaints (p < 0.05). In particular, the finding of rebound tenderness, which was present in 109 patients, had a 71% accuracy for the diagnosis of AA (77 of 109, see Data Supplement S1). Of 123 patients who underwent US, 58 were negative for appendicitis, 50 were positive, and 15 were inconclusive. Of the 58 patients with negative US, nine had AA on subsequent CT, which was performed because of high clinical suspicion. All 50 patients with US suggestive for appendicitis underwent surgery. Of these, 46 had appendicitis and four had normal appendices. All 15 patients with nondiagnostic US received additional CT scans. The sensitivity for US was 84% (95% confidence interval [CI] = 71% to 92%), and the specificity was 94% (95% CI = 82% to 98%; see Data Supplement S2, available as supporting information in the online version of this paper). Sixty patients underwent CT scans; 28 were interpreted as positive for appendicitis. All 28 patients underwent surgery. Of these, 25 had appendicitis and three had normal appendices. Thirty-two were interpreted as negative for appendicitis. None of these patients were diagnosed with AA over the 3-month follow-up period. The sensitivity for CT was 100% (95% CI = 86% to 100%), with a specificity of 91% (95% CI = 77% to 98%; see Data Supplement S2). The value of diagnostic tests for AA depends on more than significantly elevated plasma concentrations. The clinical usefulness depends largely on cutoff points that most accurately discriminate between patients with AA and those without. To find this cutoff point, ROC curves were calculated and the ideal cutoff points were assessed as the maximum sum of sensitivity and specificity of the marker. The overall accuracy of the markers was represented by the area under the ROC curve. The areas under the curve for CP, SAA-1, CRP, and WBC were 0.67 (95% CI = 0.60 to 0.74), 0.76 (95% CI = 0.70 to 0.82), 0.71 (95% CI = 0.64 to 0.78), and 0.79 (95% CI = 0.73 to 0.85), respectively. The optimum cutoff points, sensitivity, specificity, likelihood ratios (LRs), and AUC are shown in Table 2. The results show that the cutoff points for the ROC that yield the best combination of sensitivity and specificity lead to poor discrimination between patients with and without AA. Because AA is a diagnosis better not to be missed, a cutoff point with maximum sensitivity is considered most useful. When applied to the current data, the sensitivity of WBC (with a cutoff point of 7.5 × 109/L), CP (100 ng/mL), and SAA−1 (0.375 mg/mL) yield a sensitivity of 97% (95% CI = 89% to 100%). All cutoff points with the highest sensitivity are presented in Table 3. No single marker accurately differentiated between patients with AA and patients with other diagnoses. Therefore, we combined the various tests using a multivariable logistic regression model. Combining all laboratory variables revealed that WBC together with SAA-1 had the best overall accuracy, resulting in an AUC of 0.80 (95% CI = 0.74 to 0.86). Next, ideal cutoff points were assessed as maximum sum of sensitivity and specificity. This revealed both a sensitivity and a specificity of 75% for the combination of WBC and SAA-1. The combination of WBC and CRP led to an AUC of 0.80 (95% CI = 0.73 to 0.85) with a sensitivity of 82% and a specificity of 70%. No other potential combination resulted in a statistically different increased overall accuracy compared to a single test (see Data Supplement S3, available as supporting information in the online version of this paper). Correctly and expeditiously diagnosing AA is essential to minimize the number of patients with perforated appendicitis, as well as to prevent unnecessary surgery. To improve the diagnostic accuracy for AA, several investigators have studied the potential role of various plasma markers. We and others showed promising results for CP and SAA-1 for diagnosing AA in phase 1 and phase 2 diagnostic studies, comparing biomarker concentrations in patients with the disease to those of controls.13, 15 Our current study is a phase 3 study, aiming to evaluate if the concentration of specific biomarkers can distinguish between subjects with AA and those without, among a population of patients with suspected AA. A considerable number of studies have examined the value of WBC and CRP in improving the diagnostic accuracy for appendicitis. However, there is as yet no consensus whether WBC and/or CRP can support the clinical diagnosis of appendicitis.2, 5, 27 Therefore, imaging has evolved to play a pivotal role in the diagnosis of AA.6, 28 Our study reported a significant difference in plasma CP and SAA-1 levels between patients with AA and other abdominal complaints (and a healthy control group). Furthermore, CRP and WBC concentrations were also significantly increased in patients with AA. However, the determined cutoff points using ROC curves led to poor discrimination between patients with and without AA, which renders these cutoff points not useful clinically. This led us to a multimarker strategy to discriminate between patients with and those without AA. Unfortunately, AUC values for the combination of the markers were not significantly better than the highest AUC value of the single markers. Instead of using a cutoff point that has a maximal combination of sensitivity and specificity, we can use cutoff points with maximum sensitivity at the expense of specificity or vice versa. Because appendicitis is a disease better not to be missed during the initial diagnostic evaluation of a patient, we considered a cutoff point that represented maximum sensitivity as the best to differentiate between patients with and without appendicitis in a population of patients with acute abdominal complaints. When a test has a high sensitivity, a negative result effectively rules out the diagnosis. This is also known as the mnemonic "SnNout."29 When this principle was applied to the studied markers, the sensitivity of WBC (with a cutoff point of 7.5 × 109/L), CP (100 ng/mL), and SAA-1 (0.375 mg/mL) was 97% (95% CI = 89% to 100%). WBC and SAA-1 were the most accurate of all tested variables, with AUCs of 0.79 and 0.76, respectively. However, a combination of SAA-1 and WBC did not improve diagnostic accuracy over the single marker strategy. The WBC is a test available in every hospital and has long been integrated in the diagnostic work-up for patients with acute abdominal complaints. Because our data suggest that the WBC has a sensitivity for AA equivalent to or better than that of other biomarkers, we propose that WBC is the preferred biomarker for patients suspected of AA. Localization of the inflammatory process seems to be the key element in the diagnosis of patients suspected for AA, making abdominal imaging necessary. This is supported in recent studies, such as the OPTIMA trial, showing that the diagnostic accuracy in patients suspected of having AA increases when they undergo radiologic imaging.6 Our results further support the role of imaging in detecting AA.30 However, in the OPTIMA trial, the role of plasma markers was not taken into account. Using imaging for every patient suspected of having AA is time-consuming for both the emergency and the radiology departments. The use of imaging also has negative financial consequences, and imaging with ionizing radiation increases the risk of radiation-induced cancer.31, 32 Therefore, reducing the use of CT in patients with abdominal pain would be beneficial. The usefulness of the WBC for excluding AA has been supported by previous studies, showing that a cutoff point < 10 × 109/L of WBC is useful for ruling out appendicitis.12, 33 A recent phase 3 diagnostic study in children with appendicitis showed that plasma CP and leucine-rich alpha glycoprotein-1 (LRG) are elevated. However, similar to our study, the WBC (at a cutoff of 8.85 × 109/L) performed better than the newer biomarkers (CP and LRG) for the purpose of diagnosing AA.12 Using a WBC< 7.5 × 109/L to identify low-risk subjects in our study would have led to 26 fewer imaging procedures (a reduction of 15.5%). Furthermore, 54 of the 233 patients in our study (23%) could have been sent home without further evaluation. However, two of these patients (3.7%) were found to have AA at 24-hour follow-up. The diagnosis of these two patients was initially missed by their physicians because they had fewer signs and symptoms of AA compared to patients whose diagnoses were not missed. The following day, both patients presented for routine outpatient follow-up and had worsening of their abdominal complaints, which led to appendectomy showing mild inflammatory changes. The methods used for this study have several limitations that must be considered when evaluating the results. First, the sample size limits the possibility to perform subgroup analyses. Thus we were unable to analyze children or patients with perforated appendix to see if there were differences in these populations. Second, it should be noted that the test results for CP and SAA take 3.5 hours to complete due to the fact that the current assays are for laboratory research only. This currently limits the diagnostic use in the clinic. Third, it is worthwhile to mention the fact that the time between the start of the abdominal complaints and presentation at the ED was not recorded. This might be of importance since the diagnostic value of biomarkers is time-dependent. Moreover, data on the time between admission to the ED and the diagnosis in patients suspected for AA were not recorded. Also, our system relies on patients being referred to the ED by their GPs. Thus our population had already been screened by physicians who suspected AA, and this may alter the incidence of AA in our population. Finally, our patients are able to get proximate follow-up, allowing us to reevaluate discharged patients the following day. Diagnostic strategies may need to differ if this is not available. This phase 3 diagnostic study suggests that neither calprotectin, serum amyloid A-1, C-reactive protein, or white blood cell count have high enough sensitivity and specificity to be clinically useful in the evaluation of subjects with suspected acute appendicitis. A white blood cell count < 7.5 × 109/L appears to have high sensitivity, and in conjunction with a low level of clinical suspicion for acute appendicitis, can identify a subgroup of patients that may be sent home without further evaluation, but who should have available next day follow-up. Furthermore, if WBC is > 7.5×109/L, in patients with suspected AA, no discrimination with other acute abdominal pathology was possible since most of these diseases are also characterized by inflammation. Therefore WBC > 7.5×109/L selects patients in whom imaging should be considered to localize the inflammatory process. The authors thank all staff at the emergency department and laboratory of clinical chemistry and hematology of the Orbis Medical Centre for their contributions in sampling, data collection, and processing of the samples and taking care of the informed consent of the patients. The authors wish to thank HyCult biotechnology (Hbt, Uden, the Netherlands) for providing the CP and SAA ELISA kits. The authors also thank R. Visschers for his help during the statistical analyses of the data. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.