CRD summary

This review concluded that magnetic resonance imaging (MRI) was accurate for assessing circumferential resection margin involvement and T-category staging in rectal cancer. This was a generally well-conducted review, and the conclusions reflect evidence presented, but some studies may have been missed and lack of study details precluded any assessment of the clinical diversity across studies and their generalisability.

Authors' objectives

To determine the accuracy of phased array magnetic resonance imaging (MRI) for T-category lymph node metastases and circumferential resection margin involvement in primary rectal cancer.

Searching

MEDLINE, EMBASE and Cochrane Central Register of Controlled Trials (CENTRAL) were searched from 2000 to March 2011 for studies in English; search strategy was reported. Bibliographies of retrieved articles were also searched.

Study selection

Studies that assessed the accuracy of MRI and specifically used phase-array coils compared to a reference standard of histopathology for the assessment of primary rectal cancer were eligible for inclusion. Studies had to report sufficient data to construct 2x2 tables of test performance. Studies of patients who received preoperative long-course radiotherapy or chemoradiotherapy alone were excluded. Most studies used one or two interpreters for imaging results. Where reported, definitions for a positive result for all three outcomes varied across studies. No further population or test characteristics were reported. At least two reviewers independently selected studies for the review.

Assessment of study quality

Study quality was assessed using a modified version of the QUADAS tool; the criteria relating to descriptions of the tests were altered to assess whether they were conducted according to accepted standards, and the reporting of selection criteria was omitted from the assessment. The authors did not state how many reviewers assessed study quality.

Data extraction

Two reviewers independently extracted data to construct 2x2 tables of test performance for each of three outcomes: T-category (T1/2 versus T3/4), lymph node involvement and circumferential resection margin involvement. Sensitivity and specificity were calculated. Study authors were contacted to obtain missing data.

Methods of synthesis

Pooled estimates of sensitivity and specificity with 95% confidence regions were calculated using a bivariate random-effects model; 95% confidence intervals (CI) were presented in the results table. The covariates used in the analyses were MRI protocol, Tesla strength, definitions of T3 tumours, lymph node metastatic and circumferential resection margin involvement, and the method of imaging assessment. The hierarchical summary receiver operating characteristic (HSROC) curves were used to estimate the diagnostic odds ratio. Sensitivity analyses were planned to investigate the impact of study quality.

Results of the review

Twenty one studies were included in the review; 11 were prospective, eight were retrospective, and the direction of data collection was unclear in 2 studies. All included studies used an acceptable reference standard, avoided partial and differential verification biases and incorporation bias, reported on uninterpretable results and explained withdrawals. Of the 21 studies, 11 recruited a representative patient spectrum, nine avoided progression bias, five reported blinding interpreters of the reference standard, 19 blinded interpreters of the index test, 16 avoided clinical review bias, 12 used an acceptable MRI protocol and seven used adequate histopathological assessment.

T-category assessment (19 studies; 33 data sets; 1,986 participants): The summary estimate of sensitivity was 87% (95% CI 81 to 92), specificity was 75% (95% CI 68 to 80), and diagnostic odds ratio was 20.4 (95% CI 11.1 to 37.3). Studies using two or more image assessors had higher sensitivity and diagnostic odds ratio than those where images were assessed by single assessors. MRI using higher Tesla strength magnets had higher specificity.

Lymph node assessment (12 studies; 23 data sets; 1,249 participants): The summary estimate of sensitivity was 77% (95% CI 69 to 84), specificity was 71% (95% CI 59 to 81) and diagnostic odds ratio was 8.3 (95% CI 4.6 to 14.7).

Circumferential resection margin involvement (10 studies; 13 data sets; 986 participants): The summary estimate of sensitivity was 77% (95% CI 57 to 90), specificity was 94% (95% CI 88 to 97) and diagnostic odds ratio was 56.1 (95% CI 15.3 to 205.8).

Authors' conclusions

MRI had good accuracy for both circumferential resection margin and T-category and should be considered for preoperative rectal cancer staging. In contrast, MRI was poor for lymph node assessment.

CRD commentary

The review addressed a clear question with reproducible inclusion criteria. Relevant sources were searched, but only studies published in English were included. In addition, diagnostic filters were used in the search strategy which meant studies could have been missed. Study selection and data extraction were conducted in duplicate; it was unclear whether similar methods to reduce error and bias were used during the assessment of study quality. Study quality was assessed using appropriate criteria, the results of which were reported in full. Robust methods of synthesis were used to derive pooled estimates of accuracy. It is worth noting that the higher overall accuracy for circumferential resection margin is due to higher specificity and for T-category a higher sensitivity compared to lymph node assessment.

This was a generally well-conducted review, and the conclusions reflect the evidence presented, but there was the possibility for missed studies and insufficient study details were provided to assess the clinical diversity across the studies and their generalisability to clinical practice.

Implications of the review for practice and research

Practice: The authors did not state implications for practice beyond those in the overall conclusion.

Research: The authors stated that research needed to be conducted to determine how accuracy for lymph node assessment can be improved. They also stated that definitions of essential staging elements and MRI protocols should be standardised.

Funding

Cancer Services Innovation Partnership (Canada).

Bibliographic details

Al-Sukhni E, Milot L, Fruitman M, Beyene J, Victor JC, Schmocker S, Brown G, McLeod R, Kennedy E. Diagnostic accuracy of MRI for assessment of T category, lymph node metastases, and circumferential resection margin involvement in patients with rectal cancer: a systematic review and meta-analysis. Annals of Surgical Oncology 2012; 19(7): 2212-2223. [PubMed: 22271205]

Original Paper URL

http://www.springerlink.com/content/d44533016413w476/

Indexing Status

Subject indexing assigned by NLM

MeSH

Humans; Lymphatic Metastasis; Magnetic Resonance Imaging; Prognosis; ROC Curve; Rectal Neoplasms /pathology /surgery; Review Literature as Topic

AccessionNumber

12012040616

Database entry date

22/11/2012

Record Status

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.