Prospective evaluation of MRI compared with CT for the etiology of abdominal pain in emergency department patients with concern for appendicitis

Rapid MRI Abdomen for Assessment of Clinically Suspected Acute Appendicitis in the General Adult Population: a Systematic Review

OBJECTIVES

To perform a systematic review on the use of magnetic resonance imaging (MRI) of the abdomen to evaluate clinically suspected appendicitis in the general adult population. We examined the diagnostic accuracy, the reported trends of MRI use, and the factors that affect the utility of MRI abdomen, including study duration and cost-benefits.

METHODS

We conducted a systematic literature search on PubMed, MEDLINE, Embase, Web of Science, and Cochrane Library databases. We enrolled primary studies investigating the use of MRI in diagnosing appendicitis in the general adult population, excluding studies that predominantly reported on populations not representative of typical adult appendicitis presentations, such as those focusing on paediatric or pregnant populations.

RESULTS

Twenty-seven eligible primary studies and 6 secondary studies were included, totaling 2,044 patients from eight countries. The sensitivity and specificity of MRI for diagnosing appendicitis were 96% (95% CI: 93-97%) and 93% (95% CI: 80-98%), respectively. MRI can identify complicated appendicitis and accurately propose alternative diagnoses. The duration of MRI protocols in each primary study ranged between 2.26 and 30 minutes, and only one study used intravenous contrast agents in addition to the non-contrast sequences. Decision analysis suggests significant benefits for replacing computed tomography (CT) with MRI and a potential for cost reduction. Reported trends in MRI usage showed minimal utilisation in diagnostic settings even when MRI was available.

CONCLUSIONS

MRI accurately diagnoses appendicitis in the general adult population and improves the identification of complicated appendicitis or alternative diagnoses compared to other modalities using a single, rapid investigation.

Magnetic resonance imaging (MRI) for diagnosis of acute appendicitis

BACKGROUND

Appendicitis remains a difficult disease to diagnose, and imaging adjuncts are commonly employed. Magnetic resonance imaging (MRI) is an imaging test that can be used to diagnose appendicitis. It is not commonly regarded as a first-line imaging test for appendicitis, but the reported diagnostic accuracy in some studies is equivalent to computed tomography (CT) scans. As it does not expose patients to radiation, it is an attractive imaging modality, particularly in women and children.

OBJECTIVES

The primary objective was to determine the diagnostic accuracy of MRI for detecting appendicitis in all patients. Secondary objectives: To investigate the accuracy of MRI in subgroups of pregnant women, children, and adults. To investigate the potential influence of MRI scanning variables such as sequences, slice thickness, or field of view.

SEARCH METHODS

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, and Embase until February 2021. We searched the references of included studies and other systematic reviews to identify further studies. We did not exclude studies that were unpublished, published in another language, or retrospective.

SELECTION CRITERIA

We included studies that compared the outcome of an MRI scan for suspected appendicitis with a reference standard of histology, intraoperative findings, or clinical follow-up. Three study team members independently filtered search results for eligible studies.

DATA COLLECTION AND ANALYSIS

We independently extracted study data and assessed study quality using the Quality Assessment of Studies of Diagnostic Accuracy - Revised (QUADAS-2) tool. We used the bivariate model to calculate pooled estimates of sensitivity and specificity.

MAIN RESULTS

We identified 58 studies with sufficient data for meta-analysis including a total of 7462 participants (1980 with and 5482 without acute appendicitis). Estimates of sensitivity ranged from 0.18 to 1.0; estimates of specificity ranged from 0.4 to 1.0. Summary sensitivity was 0.95 (95% confidence interval (CI) 0.94 to 0.97); summary specificity was 0.96 (95% CI 0.95 to 0.97). Sensitivity and specificity remained high on subgroup analysis for pregnant women (sensitivity 0.96 (95% CI 0.88 to 0.99); specificity 0.97 (95% CI 0.95 to 0.98); 21 studies, 2282 women); children (sensitivity 0.96 (95% CI 0.95 to 0.97); specificity 0.96 (95% CI 0.92 to 0.98); 17 studies, 2794 children); and adults (sensitivity 0.96 (95% CI 0.93 to 0.97); specificity 0.93 (95% CI 0.80 to 0.98); 9 studies, 1088 participants), as well as different scanning techniques. In a hypothetical cohort of 1000 patients, there would be 12 false-positive results and 30 false-negative results. Methodological quality of the included studies was poor, and the risk of bias was high or unclear in 53% to 83% of the QUADAS-2 domains.

AUTHORS' CONCLUSIONS

MRI appears to be highly accurate in confirming and excluding acute appendicitis in adults, children, and pregnant women regardless of protocol. The methodological quality of the included studies was generally low due to incomplete and low standards of follow-up, so summary estimates of sensitivity and specificity may be biased. We could not assess the impact and direction of potential bias given the very low number of high-quality studies. Studies comparing MRI protocols were few, and although we found no influence of MRI protocol variables on the summary estimates of accuracy, our results do not rule out that some MRI protocols are more accurate than others.

Abdominal fellowship-trained versus generalist radiologist accuracy when interpreting MR and CT for the diagnosis of appendicitis

OBJECTIVES

To compare the diagnostic accuracy of generalist radiologists working in a community setting against abdominal radiologists working in an academic setting for the interpretation of MR when diagnosing acute appendicitis among emergency department patients.

METHODS

This observational study examined MR image interpretation (non-contrast MR with diffusion-weighted imaging and intravenous contrast-enhanced MR) from a prospectively enrolled cohort at an academic hospital over 18 months. Eligible patients had an abdominopelvic CT ordered to evaluate for appendicitis and were > 11 years old. The reference standard was a combination of surgery and pathology results, phone follow-up, and chart review. Six radiologists blinded to clinical information, three each from community and academic practices, independently interpreted MR and CT images in random order. We calculated test characteristics for both individual and group (consensus) diagnostic accuracy then performed Chi-square tests to identify any differences between the subgroups.

RESULTS

Analysis included 198 patients (114 women) with a mean age of 31.6 years and an appendicitis prevalence of 32.3%. For generalist radiologists, the sensitivity and specificity (95% confidence interval) were 93.8% (84.6-98.0%) and 88.8% (82.2-93.2%) for MR and 96.9% (88.7-99.8%) and 91.8% (85.8-95.5%) for CT. For fellowship-trained radiologists, the sensitivity and specificity were 96.9% (88.2-99.5%) and 89.6% (82.8-94%) for MR and 98.4% (90.5-99.9%) and 93.3% (87.3-96.7%) for CT. No statistically significant differences were detected between radiologist groups (p = 1.0, p = 0.53, respectively) or when comparing MR to CT (p = 0.21, p = 0.17, respectively).

CONCLUSIONS

MR is a reliable, radiation-free imaging alternative to CT for the evaluation of appendicitis in community-based generalist radiology practices.

KEY POINTS

• There was no significant difference in MR image interpretation accuracy between generalist and abdominal fellowship-trained radiologists when evaluating sensitivity (p = 1.0) and specificity (p = 0.53). • There was no significant difference in accuracy comparing MR to CT imaging for diagnosing appendicitis for either sensitivity (p = 0.21) or specificity (p = 0.17). • With experience, generalist radiologists enhanced their MR interpretation accuracy as demonstrated by improved interpretation sensitivity (OR 2.89 CI 1.44-5.77, p = 0.003) and decreased mean interpretation time (5 to 3.89 min).