Recent statistics on the incidence of CRC report that 37,514 new cases of large bowel cancer were registered in the UK in 2006. Primary colonic tumours (any part of the colon other than the rectum) outnumber primary rectal tumours by approximately 2 : 1.⁴⁷

After the first diagnosis of a new CRC with colonoscopy and biopsy, and occasionally following an incidental finding on radiological imaging performed for another reason, more sophisticated medical imaging tests are required to accurately stage the tumour to determine the extent to which other tissues might be involved. The aim of radiological imaging is to obtain information regarding the primary tumour (T stage), local and distant nodal involvement (N stage) and distant metastatic disease (M stage).

The containment of the tumour to the mucosa or the muscularis propria is associated with improved survival.⁴⁸ However, because of delayed presentation before development of symptoms, it is not uncommon for CRCs to have either spread locally through the colonic wall, with involvement of local and distant nodal groups, or metastasised. Metastases tend to occur to the liver and lung.

For staging primary cancer of the colon, CT with spiral acquisition through the chest, abdomen and pelvis following intravenous contrast administration is the conventional imaging technique and is readily available and reproducible. MRI is not currently used for assessment of primary colonic tumours because of image degradation by peristalsis, and does not add any value to standard CT imaging in this area.

Magnetic resonance imaging is used as standard for staging of rectal cancers, with the benefits of imaging a part of the bowel that is not degraded by peristaltic motion resulting in more accurate assessment of the primary tumour and relationships to bowel wall. MRI is also capable of identifying involved local nodal groups, although, currently, standard practice is to rely on size criteria to suggest involvement. This results in a large inherent inaccuracy, as many studies have shown that size alone is a poor indicator of disease status within nodal groups. Spiral CT is also used to give information on local and distant nodal involvement and the presence or absence of metastatic disease, again usually to the liver and lungs.

FDG PET/CT is currently recommended only for the assessment of suspected recurrence of CRC and in pre-operative staging prior to metastectomy, and clinical opinion on the role of FDG PET/CT in the routine management of primary colon cancer varies. Some investigators suggest that in certain clinical circumstances it should be considered as part of the standard pre-operative assessment and acknowledge it may have an up-and-coming role in the initial staging of primary rectal cancer.^21,26–28 More recently, small studies have suggested that FDG PET/CT may offer a clinically useful addition for the routine staging of rectal cancers.²⁷ In one study, FDG PET/CT was able to identify involved nodes outwith the mesorectal fascia not seen with standard imaging, particularly with low rectal tumours in which iliac and inguinofemoral nodal involvement was a common finding.⁴⁹ This resulted in a change in planned management in 27% of patients, improving the accuracy of pre-treatment staging.

As 11% of the CRC population present with metastases at the time of first diagnosis,⁵⁰ the early identification of those with advanced disease might lead to improved survival. Replacing diagnostic CT with FDG PET/CT as the initial imaging investigation has considerable resource and cost implications; in a cost-effectiveness study in 2005, a single PET scan (using non-integrated equipment) was estimated to cost €1038 compared with €313 for a single CT scan.⁵¹ These cost considerations currently limit FDG PET/CT use to an add-on test in most centres where the technology is available.

The rationale for using FDG PET/CT (or PET/contrast-enhanced CT) as a replacement test, at the outset of the CRC diagnostic pathway, is the avoidance of unnecessary and expensive surgery in individuals who have advanced incurable disease at the time of the first diagnosis.⁵⁰

Replacing the diagnostic CT scan with FDG PET/CT implies that the CT component of both investigations should be of a similar standard. Modern FDG PET/CT systems tend to use low-dose spiral CT to localise a functional abnormality – this is because the patient has already undergone a high-quality, higher-dose spiral CT scan with intravenous contrast in almost all cases. Using FDG PET/CT as a replacement should involve performance of the CT component to the same standards as the diagnostic scan – this increases the complexity of the FDG PET/CT study but may have a role in the future.

It is anticipated that there will be a knock-on effect from the UK CRC screening programmes on NHS CRC radiology services, and this also deserves consideration.^50,52 Results from the second round of the UK CRC screening programme show that the majority of detected cancers are early tumours, classified as Dukes A or B. FDG PET/CT is more commonly used for the detection of recurrence and liver metastases, and the value of its use in the detection of small, contained primary CRCs remains under-researched. The limitation of the spatial resolution of modern scanners (5–8 mm) is a potential issue here and, although modern FDG PET/CT systems are capable of identifying primary CRCs, the standard roles for this imaging modality are concerned with disease spread, rather than assessment of the primary tumour.

Results

Our search did not identify any systematic reviews to evaluate the diagnostic test accuracy of integrated FDG PET/CT in the pre-operative staging of primary CRC.

We found no studies that met the stated objectives, but have presented all studies that included patients with primary CRC who received FDG PET/CT for pre-operative staging.

Number of included studies

The search identified two studies^53,54 that evaluated the diagnostic test accuracy of integrated FDG PET/CT for the detection of primary CRC.

Study characteristics and study designs

The study characteristics are shown in Table 1 and accuracy data are shown in Table 2.

TABLE 1

Study characteristics.

TABLE 2

Accuracy data.

One study⁵⁴ reported using a retrospective design. In the second study⁵³ the study design was unclear.

Study setting and country in which the research was conducted

Both studies were conducted in cancer centres in Japan.

Patient populations

The studies reported outcomes for a total of 141 patients, who were mostly men. The mean age of the patients in the two studies was 60⁵³ and 61⁵⁴ years, respectively, with a range of 23–89 years. Rectal cancer affected 104 patients and cancer of the colon affected 37.

Indication for FDG PET/CT

FDG PET/CT was undertaken in order to pre-operatively stage the primary CRC, and was reported to be specifically for the diagnosis of nodal disease in one study.⁵³

FDG PET/CT equipment and patient preparation

Full details of all the equipment used in the studies can be found in Table 1. FDG PET/CT equipment was manufactured by GE Healthcare (Fairfield, CT) and Siemens Medical Solutions (Surrey, UK). The fasting duration prior to the scan was at least 6 hours in both studies.

A range of injected FDG doses was reported with units ranging from 370 to 555 MBq. Where the information was reported, patients were scanned 60 minutes after the administration of the radioactive tracer.⁵³ Neither study used contrast-enhanced CT.

Image interpretation

The index and comparator tests, when one was performed,⁵⁴ were assessed by at least two individuals blind to clinical information. The qualitative and quantitative image interpretations were conducted in similar ways in each of the studies.

In one study,⁵³ lymph node metastases were qualitatively diagnosed by abnormal uptake regardless of node shape or size (visual diagnosis). A maximal nodal diameter (size diagnosis) cut-off value of 10 mm was used; maximum standardised uptake value (SUV_max) of lymph nodes with greater uptake than normal organs or surrounding tissue (SUV diagnosis): the optimal cutoff value was where accuracy was greatest; when more than one lymph node in the proximal or distant region was malignant, the one with the highest SUV was used in the analysis.

In the second study⁵⁴ the qualitative assessment judged lesions positive if there was an abnormal focal uptake at a location corresponding to lymph node chains on CT scans. Abnormal uptake was defined as a focal increased activity higher than that of the background soft tissue. The authors also reported using SUV_max in the quantitative diagnosis but did not report any other details.

Reference standard

Both studies reported histopathology from surgically resected specimens and lymph node resection as the reference standard in all patients.

Data synthesis – diagnostic performance

FDG PET/CT versus none

In proximal nodal staging based on patient-level analysis with an SUV threshold of 1.5, FDG PET/CT demonstrated a sensitivity of 51% (95% CI 36% to 66%) and a specificity of 85% (95% CI 72% to 92%). For distal nodal staging using an SUV threshold of 1.5, the sensitivity of FDG PET/CT was 62% (95% CI 30% to 86%) and specificity was 92% (95% CI 84% to 96%).

In analysis based on group-level data and compared with findings from surgical excisions and region lymph node dissection,FDG PET/CT was found to have a sensitivity of 28% (95% CI 18% to 42%) and specificity of 92% (95% CI 87% to 96%) when used for nodal staging of proximal and distal lymph nodes (n = 176).

In the detection of lymph nodes based on size (nodal maximum axial diameter) and a threshold of ≥ 10 mm, the authors found FDG PET/CT to have a sensitivity of 30% (95% CI 19% to 44%) and a specificity of 95% (95% CI 90% to 97%).

For SUV diagnosis using the optimal cut-off value of 1.5, the sensitivity of FDG PET/CT was 53% (95% CI 39% to 66%) and the specificity was 90% (95% CI 84% to 94%). At a threshold of 2.5, the sensitivity of FDG PET/CT was 38% (95% CI 26% to 53%) and the specificity was 94% (95% CI 89% to 97%), and, at a threshold of 3.5, the sensitivity was 24% (95% CI 15% to 38%) and the specificity was 100% (95% CI not calculable).

FDG PET/CT versus contrast-enhanced FDG PET/CT

Accuracy in diagnosing nodal staging appeared to be improved by adding contrast enhancement to the FDG PET/CT in 53 patients. FDG PET/CT had a sensitivity of 85% (95% CI 69% to 93%) and a specificity of 42% (95% CI 23% to 67%), and the respective accuracy of contrast-enhanced FDG PET/CT was 85% (95% CI 69% to 93%) and 68% (95% CI 46% to 84%).

FDG PET/CT produced poorer estimates of accuracy than contrast-enhanced FDG PET/CT in imaging all lymph nodes: a sensitivity of 75% (95% CI 57% to 87%) and a specificity of 52% (95% CI 32% to 71%) for FDG PET/CT compared with 90% sensitivity (95% CI 76% to 97%) and 76% specificity (95% CI 55% to 89%) for contrast-enhanced FDG PET/CT for pararectal nodes; a sensitivity of 31% (95% CI 14% to 55%) and specificity of 81% (95% CI 66% to 90%) for FDG PET/CT but 75% sensitivity (CI 95% 50% to 90%) and 86% specificity (95% CI 72% to 94%) for contrast-enhanced FDG PET/CT for internal iliac nodes; and a sensitivity of 67% (95% CI 39% to 86%) and a specificity of 61% (95% CI 46% to 74%) for FDG PET/CT compared with a sensitivity of 67% (95% CI 39% to 86%) and a specificity of 95% (95% CI 84% to 99%) for contrast-enhanced FDG PET/CT for obturator nodes.

Quality assessment

Fourteen items from the QUADAS checklist were used to assess the methodological quality of the results and the findings from this process are shown in Table 3.

TABLE 3

Quality assessment of studies evaluating FDG PET/CT in the pre-operative staging of primary CRC.

Only one study reported including a consecutive series of patients nor a random sample of adults undergoing pre-operative staging of primary CRC.⁵³ In both studies the assessors were blind to the clinical information and results of other studies. The reference standard test of histopathology was applied to all patients and is therefore likely to be 100% sensitive and specific, but neither study gave details of the execution of the reference standard. The reviewers considered 6 weeks to be the time limit after which disease progression might occur, and the time between the reference standard and the index test was reported to be ≤ 6 weeks in one study.⁵³ The withdrawals from the study were explained for both studies.

In addition to selection bias, the validity of the findings from these studies is also potentially compromised by review bias arising from the lack of information about the ‘blinding’ of individuals reviewing the scans.