CT Colonography

The evidence base comparing CT colonography and conventional colonoscopy consists of three systematic reviews (Chaparro et al, 2009, Mulhall et al, 2005 and Halligan et al, 2005). There was a high degree of overlap between the three meta-anlayses in relation to the studies included in each with the more up to date review (Chaparro e al, 2009) including the majority of studies which had been used in the two previous reviews along with a number of studies published since. In total, 85% of studies included in Mulhall, 2005 and 67% of studies included in Halligan, 2005 were included by Chaparro, 2009. Reasons for discrepancies in the included studies might be due to slight differences in research questions in each review or more up to date versions of studies used in the later review, though it is not clear that this is the case.

The evidence base is of generally high quality based on assessment using the QUADAS checklist to evaluate the 47 studies assessed as part of a systematic review (Chaparro et al, 2009) . The two remaining systematic reviews did not provide a detailed report of the quality assessment for the included studies though both studies did assess quality using standardised methods. Due to the fact that there was a high degree of overlap in the reviews, the quality assessment provided by Chaparro (2009) was considered an adequate reflection of the quality of evidence from all 3 reviews.

Figure 2.1

Summary of the quality of included studies comparing CT Colonography and Conventional Colonscopy.

The only area for which the quality of the individual studies was low related to the inadequacy of reporting of uninterpretable results in the majority of studies.

The area of largest uncertainty was whether or not readers of CT colonography and colonoscopy results had access to all relevant clinical information necessary to accurately interpret images as this is not generally reported in studies.

Diagnostic studies are susceptible to a particular bias known as spectrum bias which describes the effect a change in patient mix may have on the performance of a given test. In the case of the studies used to inform this topic, the majority of the study populations consist of a representative spectrum of the patients that are likely to be referred for diagnostic interventions which would suggest that spectrum bias is not a particular concern for this topic. This does not present the whole picture however as the spectrum of patients referred for a particular intervention may be impacted by local practices and therefore a representative patient population in the UK may not be the same as one in the US and so this should be considered when examining the evidence; for example a patient may be referred for a particular intervention based on the severity of their symptoms. This topic also related to the effectiveness of interventions in symptomatic patients and the studies included in each systematic review included a variety of patients including both symptomatic and asymptomatic patients.

Two retrospective case series’ examined the potential adverse effects of CT colonography (Burling et al, 2006 and Sosna et al, 2006).

A small number of individual studies examining the effectiveness of CT colonography which were not included in any of the systematic reviews were identified (Hoppe et al, 2004; Laghi et al, 2002; Pescatore et al, 2000; Reuterskiold et al, 2006). No reason for why these studies were excluded from the most recent systematic review could be found and so an evidence table for each study has been produced and included, though the results reported are not discussed in this summary.

Flexible Sigmoidoscopy plus Air Contrast Barium Enema

Two randomised trials compared flexible sigmoidoscopy plus barium enema with conventional colonoscopy (Rex et al, 1995 and Rex et al, 1990). From figure 2.2 it can be seen that the quality of the two randomised trials is very low with a high risk of bias in both studies.

Figure 2.2

Summary of quality of studies comparing flexible sigmoidoscopy plus air contrast barium enema with conventional colonoscopy.

Per Polyp Sensitivity

From two systematic reviews and meta-analysis (Chaparro et al, 2009 and Halligan et al, 2005), per polyp sensitivity of CT colonography was similar and both reviews reported higher sensitivities for larger polyps.

Chaparro (2009) reported sensitivities which ranged from 28–100% for all polyps >6mm with an overall pooled sensitivity was 66% (95% CI 64–68%). Halligan et al (2005) did not report an overall pooled sensitivity.

Sensitivity was reported to increase with polyp size with a pooled sensitivity of 59% (95% CI 56%–61%, range 16%–90%) for polyps 6–9mm and a pooled sensitivity of 76% (95% CI 73–79%, range 50–100%) for polyps >9mm.

There was significant heterogeneity between studies in all three comparison groups with the I² value >50% for all three groups (Chaparro et al, 2009).

Halligan et al (2005) similarly reported the the performance of CT colonography was affected by polyp size; average sensitivity for large polyps was 77% (95% CI, 70%–83%) and 70% (95% CI 63%–76%) for medium polyps. Due to heterogeneity the data for small polyps were not pooled in this study.

Different thresholds for polyp size were used in both systematic reviews maybe which impact on the outcomes and so ought to be considered when interpreting the results.

Mulhall et al (2005) did not report per polyp sensitivities as it was considered that the per patient outcomes were more important to know for the accuracy of CT Colonography in diagnosis and screening.

Per Patient Sensitivity and Specificity

From one systematic review (Chaparro et al, 2009), the per patient sensitivity for CT colonography ranged from 24%–100% across the individual studies and the overall pooled sensitivity was 69% (95% CI, 66%–72%).

Mulhall et al, 2005 reported that per patient sensitivity ranged from 21% to 96% with an overall pooled sensitivity of 70% (95% CI, 53% to 87%).

Sensitivity again increased with increasing polyp size with a pooled sensitivity of 60% (95% CI 56%–65%) for patients with polyps 6–9mm (range 20%–91%) and 83% (95% CI, 70%–85%) for patients with polyps >9mm (range 46%–100%) (Chaparro et al, 2009).

Again there was significant between studies heterogeneity for each of the analyses groups.

Halligan et al (2005) reported an average per patient sensitivity of 93% (95% CI, 73%–98%) for large polyps (≥1cm), 86% (95% CI 75%–93%) for medium polyps (6–9mm) and did not report an average sensitivity for small polyps (<6mm) due to the heterogeneity of the data across studies.

Sensitivity progressively increased as polyp size increased with sensitivity of 48% (95% CI, 25%–70%, range, 14%–86%) for the detection of polyps <6mm, 70% (95% CI, 55%–84%, range, 30%–95%) for polyps 6–9mm and 85% (95% CI 79%–91%, range, 48%–100%) for polyps >9mm.

Significant statistical heterogeneity was observed for each of these analyses (p<0.001 for each group) with most of the variance attributed to between-study heterogeneity.

The overall specificity of CT colonography was reported to be 83% (95% CI, 81%–84%, I²=89%) with specificity improving with increasing polyp size; specificity was 90% (95% CI, 89%–91%, I²=21%) for patients with polyps 6–9mm in size and increased to 92% (95% CI, 92%–93%, I²=62%) for polyps >9mm (Chaparro et al, 2009).

Halligan et al (2005) also reported improved specificity with larger polyp size with an average sensitivity of 70% (95% CI, 63%–76%) for medium polyps (6–9mm) increasing to 77% (95% CI, 70%–83%) for large polyps (≥1cm).

Mulhall et al (2005) reported a consistent per patient specificity across polyp sizes though there was significant heterogeneity; overall specificity was reported as being 86% (95% CI, 84%–88%, I²=92.6%, p=0.001).

When examining specificity according to polyp size no heterogeneity was observed within the groups (though the I² statistic was still around 50% for all groups) and specificity improved as polyp size increased; for polyps <6mm pooled specificity was 91% (95% CI, 89%–95%, I²=47.1%, p=0.15), for polyps 6–9mm in size, pooled specificity was 93% (95% CI, 91%–95%, I²=50%, p=0.07) and for polyps >9mm, pooled specificity was 97% (95% CI, 96%–97%, I²=41.8%, p>0.2).

Subgroup Analysis

Two systematic reviews (Chaparro et al, 2009 and Mulhall et al, 2005) examined a number of variables in an effort to explain some of the heterogeneity, the results of which are outlined in table 2.1. Variables investigated included colonic preparation, use of contrast, use of faecal tagging, collimation width, scanner type, width of reconstruction interval, imaging (2D with 3D confirmation or 3D only), high risk versus average risk patients, study quality, year of publication and type of scanner.

Table 2.1

Subgroup analysis for possible variables contributing to heterogeneity.

Likelihood Ratios

Chaparro et al (2009) reported overall positive likelihood ratio was 2.9 (1.8–4) and the overall negative likelihood ratio was 0.38 (0.27–0.53).

For polyps between 6–9mm, the positive likelihood ratio was 3.8 (2.5–5.7) and the negative likelihood ratio was 0.4 (0.27–0.59).

For polyps >9mm, the positive likelihood ratio was 12.3 (7.7–19.4) and the negative likelihood ratio was 0.19 (0.12–0.3).

Likelihood ratios were not reported in either of the other systematic reviews.

Risks and Safety of CT Colonography

Two retrospective case series studies reported on the potential adverse events related to CT colonography (Burling et al (2006) & Sosna et al (2006).

Burling et al (2006) reported that 17,067 CT Colonographic examinations had been performed across 50 centres; which had performed a total of 100 examinations or more.

No deaths were reported and 13 patients (0.08%; 1 in 1313 patients) had experienced potentially serious adverse events believed to be related to CT colonography, 9 of which were luminal perforations giving a perforation rate of 0.05% (1 in 1896 patients). The symptomatic perforation rate was 0.03% (1 in 3413 patients).

8/9 patients with perforation were treated conservatively either as inpatients or outpatients and to the knowledge of the respondents, all patients were alive and well at the time of the survey.

At 29 centres (58%) an inflated balloon catheter was never used, at 7 centres (14%) one was used occasionally and at 14 centres (28%) one was always used.

Overall, 9378 CT Colonographic examinations were performed using an inflated balloon in the rectum and among these there were 6 perforations; 7689 CT colonographic examinations were performed without an inflated balloon with 2 perforations.

At 6 centres (12%) an automated insufflation device was used with 2 perforations associated.

There was no significant difference in the proportion of perforations associated with and without rectal balloon inflation (p=0.3)

Sosna et al (2006) reported 7 colonic perforations at 5 centres for a perforation risk rate of 0.059% (95% CI 0.02%–0.1%), translating to an event occurrence of 1/1695 studies (95% CI 1/974 – 971/6537).

6/7 cases of perforations were in symptomatic patients at high risk of colorectal neoplasia and only 1 occurred in an asymptomatic patient with average risk who underwent screening.

4 cases of perforation were in patients undergoing CT Colonography as completion studies following incomplete conventional colonoscopy.

There were 5 cases of perforation in the sigmoid colon and 2 in the rectum.

6 cases of perforation occurred in patients in whom a rectal tube was inserted and in 5/6 cases the balloon was inflated. In the remaining patient a 16-F Foley catheter was inserted and 5ml of saline was inflated into the balloon.

4/7 patients with perforation required surgical treatment with a one-stage procedure performed in 3 patients and a two-stage procedure performed in 1.

The incidence of surgical intervention was 1/2968 patients (95% CI 1.5 of 10,000 – 14.7 of 10,000).

The remaining 3 patients had multiple comorbidities and were at high risk for surgery and so received conservative treatment without any complications.

No deaths were recorded.

The physicians performing the air insufflation in 2 cases of perforation did not have any experience in the performance of CT colonography at the time of examination with neither having performed unsupervised air insufflation previously nor read images from CT colonographic studies on a regular basis.

Requirement for Alternative Procedures

From one randomised trial, there was a significant difference between the arms in relation to the proportion of patient’s recommended alternative lower GI procedures (p≤0.0001)

53/168 (32%) patients in the flexible sigmoidoscopy group were referred for subsequent colonoscopy due to inadequate study (n=11), for polypectomy (n=38) and for biopsies on lesions outside the reach of flexible sigmoidoscopy.

13/164 (8%) patients in the colonoscopy arm were referred for ACBE because of difficulty advancing the colonoscope to the cecum (Rex et al, 1990).

While in the second trial (Rex et al, 1995) patients undergoing flexible sigmoidoscopy were more likely to require an alternative intervention such as colonoscopy than were patients undergoing colonoscopy to require air contrast barium enema (OR=2.07, 95% CI, 1.47–16.4)..

Complications and Risks

No significant difference between the two groups in relation to procedural complications. Phlebitis occurred in 7 patients in the colonoscopy group versus 4 patients in the flexible sigmoidoscopy +ACBE group, this difference was not statistically significant, however the authors state that the study did not have sufficient power to detect a true difference in the incidence of phlebitis of this magnitude (Rex et al, 1990).

No deaths, transfusions, hospitalisations, or prolonged hospital stays were reported in either patient group from either study (Rex et al, 1995 & Rex et al, 1990).

Updated Evidence

A single systematic review (Dighe et al, 2010) investigated the accuracy and limitations of CT in identifying poor prognostic factors in colon cancer as well as investigating which CT technique achieved the best results.

The comprehensive review included 19 studies from which relevant data could be extracted and specifically examined the ability of CT to detect muscularispropria invasion enabling the differentiation between T1/T2 and T3/T4 tumours and the detection of lymph node metastases.

For the detection of muscularispropria invasion, sensitivity and specificity measures could be obtained from 17 studies, while for lymph node involvement data could be obtained from 15 studies.

Funnel plots for publication bias showed some evidence that the smaller studies included in the review were associated with a larger diagnostic odds ratio for both tumour invasion and lymph node detection and therefore the evidence provided from smaller studies alone potentially over-estimates the true effect; though this was not statistically significant (p=0.07).

From the systematic review, a significant number of false negatives for muscularispropria invasion resulted in understaging of T3/T4 tumours in 4 of the included studies however the three of the four studies were older and CT was performed without the benefit of spiral or MDCT and with a section thickness of 10mm which may be a factor in the failure to detect small amount of tumour invasion. In the fourth study, the authors of the systematic review reported that there did not appear to be any reason for the high false negative rate other than the possibility that the study population included many patients with microscopic invasion beyond the muscularispropria.

The false positive rate was low in all included studies suggesting that CT can reliably identify T3/T4 tumours.

For nodal involvement, earlier studies showed poor results for similar reasons to those outlined for muscularispropria invasion.

Digital Rectal Exam

Four studies provided information on digital rectal examination (Beynon et al, 1986; The Mercury Study Group (2006); Brown et al (2004) and Rafaelson et al).

From Beynon et al (1986), surgeons were asked to allocate palpable tumours to one of four grades. Digital rectal exam was performed in 35 patients and the study reported an accuracy of 68%, sensitivity of 68% and specificity of 83% in the ability of DRE to preoperatively stage rectal cancer (histology was used as the reference standard).

The Mercury Study group is primarily concerned with investigating the accuracy of MRI to preoperatively stage rectal cancer through the prediction of circumferential resection margins. Patients participating in the study were also required to undergo a clinical exam which, included a DRE and the study reported that DRE resulted in an accuracy of 70% for the prediction of circumferential resection margins and that when DRE showed fixed or tethered tumours, this corresponded to an involved margin in only 15% of cases. Sensitivity and specificity for DRE were 38% and 74% respectively.

Brown et al (2004) evaluated the accuracy of CRE in the identification of favourable, unfavourable and locally advanced rectal carcinoma in 98 patients in order to determine which patients should be offered pre-operative short course or long course radiotherapy or surgery alone.

Compared with pathological findings, DRE correctly identified 71% of patients with favourable prognosis tumours, 36% of patients with unfavourable prognosis tumours and 11% of patients with features indicative of locally advanced tumours. Based on the results of DRE, Brown et al (2004). concluded that 51 patients would have undergone surgery alone, 39 patients would have been offered short-course radiotherapy and 8 patients would have been offered long-course radiotherapy compared with 22, 14, and 3 patients in each group if basing decision on results of histopathology.

Rafaelson et al (1994) aimed to pre-operatively stage rectal cancer by DRE. A total of 107 patients were included in the study though in 13 patients, tumour was beyond the reach of the examining finger. DRE underestimated depth of rectal wall penetration in 28% of patients and overestimated depth of penetration in 26% of cases. Overestimation appeared to occur more often with small tumours versus large tumours and a significant difference in overestimation was observed when comparing tumours located in a single quadrant compared with tumours located in more than one quadrant (p=0.01).

Underestimation of penetration depth on DRE was significantly higher in large tumours versus smaller tumours (p=0.006).

Complete clinical and pathological data were available for 53 patients and palpable lymph nodes were found in one patient on DRE though no metastases were found in the resected specimen. Overall, this study reported that 45% of patients were correctly staged by DRE.

Computed Tomography

A total of 11 studies investigating the use of computed tomography (CT) for the preoperative staging of rectal cancer were identified (Kwok et al (2000), Bipat et al (2004), Filipone et al (2004) Kantarova et al (2003), Kulinna et al (2004a), Kulinna et al (2004b) Mainenti et al (2006), Llamas-Elvira et al (2007), Beynon et al (1986), Kim et al (2007), Dirisamer et al (2010), Nicholls et al (1982). There was variation across the studies in relation to the factors examined and the type of CT used.

Features investigated by individual studies included depth of rectal wall penetration, nodal involvement, muscularispropria invasion, perirectal tissue invasion, adjacent organ invasion, T stage and presence of liver metastases.

Methods of CT reported across individual studies included CT colonography with transverse images alone or in combination with multi-planar reconstructions (MPRs), multislice CT, axial slice CT with and without coronal and saggital MPRs.

Two good quality systematic reviews (Kwok et al (2000) and Bipat et al (2004)) evaluated the use of CT as a method for the preoperative staging of rectal cancer.

From Kwok et al (2000) 23 studies with a total of 1116 patients, were reported to have used CT in the pre-operative assessment of local tumour penetration (defined as ‘through wall’ i.e. invading muscularispropria or ‘not through wall’). The pooled sensitivity was 78%, pooled sensitivity was 63% and pooled accuracy was 73%.

Of these, 4 studies (n=135 patients) classified wall penetration according to TNM notation and of these 80% were correctly staged, 11% were over-staged and 7% were understaged.

From Bipat et al (2004) depth of tumour pentetration was investigated as three specific subgroups; muscularispropria invasion, perirectal tissue invasion and adjacent organ invasion. There were not enough data available to determine sensitivity and specificity of CT in determining muscularispropria invasion. For perirectal tissure invasion the pooled sensitivity was 72% (95% CI, 64%–79%) and the pooled specificity was 78% (95% CI, 73%–83%). For adjacent organ invasion the pooled sensitivity was 72% (95% CI, 64%–79%) and the pooled specificity was 96% (95% CI, 95%–97%).

For nodal involvement, Kwok et al (2000) reported on data that were drawn from 18 studies (n=945 patients) and the pooled sensitivity was 52%, pooled specificity was 78% and pooled accuracy was 66%.

Bipat et al (2004) reported a pooled sensitivity of 55% (95% CI, 43%–67%) and pooled specificity of (74% (67%–80%) for the detection of lymph node involvement.

Endoluminal Ultrasound (EUS)

A total of 9 studies reported on the use of endoluminal ultrasound (EUS) in the pre-operative staging of rectal cancer including 2 good quality systematic reviews (Kwok et al (2000) and Bipat et al (2004)). Features investigated in order to stage rectal cancer, again varied across the individual studies in relation to the subgroups identified and investigated but again primarily included wall penetration, nodal involvement, presence/absence of liver metastases.

From 53 studies (n=2915 patients) Kwok et al (2000) reported a pooled sensitivity of 93%, pooled specificity of 78% and pooled accuracy of 87% for the detection of wall penetration according to the TNM classification; of these, 84% were correctly staged, 11% were over-staged and 5% were understaged.

Bipat et al (2004) reported a pooled sensitivity of 94% (90%–97%), and pooled specificity of 86% (95% CI, 80%–90%) for muscularispropria invasion; a pooled sensitivity of 90% (95% CI, 88%–92%) and a pooled specificity of 75% (95% CI, 69%–81%) for perirectal tissue invasion and a pooled sensitivity of 70% (95% CI, 62%–77%) and pooled specificity of 97% (95% CI, 96%–98%) for adjacent organ invasion.

In relation to nodal involvement Kwok et al (2000) reported a pooled sensitivity of 71%, pooled specificity of 76% and an accuracy of 74%(36 studies with a total of 2032 patients) while Bipat et al (2004) reported a pooled sensitivity of 67% (95% CI, 60%–73%) and a pooled specificity of 78% (95% CI, 71%–84%).

Magnetic Resonance Imaging (MRI)

MRI as a method of pre-operatively staging rectal cancer was investigated in 18 studies including two good quality systematic reviews (Kwok et al (2000), Bipat et al (2004) and one large multicentre, UK study (Mercury Study Group, 2006 and 2007).

The method of MRI varied across the studies identified and included studies which investigated all types of MRI and studies which investigated subgroups of MRI such as MRI with endorectal coil, MRI with body coil, MRI with and without contrast material and phased array MRI.

From Kwok et al (2000) with a total of 18 studies (n=521 patients and 546 MRI scans) the pooled sensitivity was 86%, pooled specificity was 77% and pooled accuracy was 82% for wall penetration.

Eight studies included in the review reported results using TNM notation (246 patients) and the pooled sensitivity, specificity and accuracy for these studies was 89%, 79% and 84% respectively.

In a subgroup analysis of patients using endorectal surface coil (6 studies; 169 patients) resulted in a pooled sensitivity, specificity and accuracy of 89%, 79% and 84% respectively.

Four studies (124 patients) reported the results according to TNM notation, of these 81% were correctly staged, 12% were overstaged and 6% were understaged.

For muscularispropria invasion, Bipat et al (2004) reported a pooled sensitivity of 90% (95% CI, 89%–97%) and a pooled specificity of 69% (95% CI (52%–82%); for perirectal tissue invasion the pooled sensitivity was 82% (95% CI, 74%–87%) and pooled specificity was 76% (95% CI, 65%–84% and for adjacent organ invasion the pooled sensitivity was 74% (95% CI, 63–83%) and pooled specificity was 96% (95% CI, 95%–97%).

A total of 15 studies (14 patients) with a total of 436 MRI scans assessed local nodal involvement by MRI. The pooled sensitivity, specificity and accuracy were 65%, 80% and 74% respectively.

A total of 181 patients (6 studies) received MRI with endorectal surface coil; the pooled sensitivity, specificity and accuracy for this subgroup were 82%, 83% and 82% respectively (Kwok et al, 2000).

Pooled sensitivity was 66% (95% CI, 54%–76%) and pooled specificity was 76% (95% CI, 59%–87%) for lymph node involvement (Bipat et al, 2004).

Specific UK evidence was provided from the Mercury Study group, (Mercury Study Group 2006 and 2007) investigating MRI in the staging of rectal cancer.

The accuracy of MRI for predicting the status of circumferential resection margin (presence/absence of tumour) by initial imaging or imaging after pre-operative treatment was 88% (95% CI, 85%–91%), sensitivity was 59% (95% CI, 46%–72%) and specificity was 92% (95% CI, 90%–95%).

For patients undergoing primary surgery with no pre-operative treatment (n=311), accuracy of prediction of a clear margin was 91% (95% CI, 88%–94%), sensitivity of 42% and specificity of 98%.

For patients undergoing pre-operative chemoradiotherapy or long-course radiotherapy the accuracy of prediction of clear margins on MRI was 77% (95% CI, 69%–86%), sensitivity was 94% and specificity was 73%.

Histopathology results showed 58 patients with affected margins, of which MRI correctly identified 32.

A second publication by the same study group (Mercury Study Group, 2007) evaluated the accuracy of MRI in depicting the extramural depth of invasion in patients with rectal cancer with the primary outcome being equivalence between MRI and histopathology in the measurement of extramural depth of tumour invasion.

Information on the depth of extramural tumour invasion was available for both histopathology and MRI in 295 patients. Mean extramural depths of invasion at MRI was 2.8mm (SD±4.6mm) and for histopathology was 2.81mm (SD±4.28mm). The mean difference between MRI and histopathologic analysis was 0.05mm±3.85 (95% CI, −0.49mm−0.4mm) resulting more than 95% certainty that the assessments were equivalent (i.e. MRI was as good as histopathology for the measurement of the depth of extramural invasion). Overall, MRI depicted depth of tumour spread in 92.5% of patients to within 5mm of histopathology and in 7.25% of patients MRI resulted in overestimation of depth of tumour spread by more than 5mm which would have resulted in patients being assigned to the wrong prognostic group.

MRI led to underestimation of tumour depth in 13 patients of which 5 were deemed to be interpretation errors due to movement artefact.

FDG-PET

Three studies investigated the use of FDG-PET in the pre-operative staging of rectal cancer (Kantarova et al, 2003, Llamas-Elvira et al 2007, and Dirisamer et al 2010). All three studies were retrospective case series of poor quality with small numbers of patients and little information on methodology and outcomes provided.

Kantarova et al (2003) reported that FDG-PET correctly detected 95% of primary tumours. For the detection of lymph nodes accuracy was 75%, sensitivity was 29% and specificity was 88%. Liver metastases were present in 9 patients and FDG-PET had an accuracy of 91%, sensitivity of 78% and specificity of 96%.

Llamas-Elvira et al (2007) evaluated FDG-PET in the initial staging of colorectal cancer and reported an accuracy of 56%, sensitivity of 21% and specificity of 95% for N0/N+ staging and an accuracy of 92%, sensitivity of 89% and specificity of 93% for M0/M+ staging.

Dirisamer et al (2010) evaluated the diagnostic role of FDG-PET in the staging and restaging of colorectal cancer and reported an overall accuracy of 84%, sensitivity of 85% and specificity of 70%.

Tumour Penetration

Tumour penetration was reported in some form in a total of 6 studies (Kwok et al (2000), Bipat et al (2004), Rafaelson et al (1994), Chun et al (2006), Fuchsjager et al (2003) and The Mercury Study Group (2007)) with some reporting wall penetration as a single outcome and some studies reporting subgroups of penetration including; muscularispropria invasion, perirectal tissue invasion and adjacent organ invasion.

From one systematic review study using data from a number of studies (Kwok et al. 2000) EUS had the highest sensitivity (93%) specificity (78%) and accuracy (87%) for wall penetration when compared with CT and MRI, though MRI with endorectal coil was quite similar.

Reported sensitivities for all types of penetration ranged from 72%–79% for CT; 79%–100% for MRI and 70%–94% for EUS. No sensitivities or specificities were reported for either DRE or PET though one study (Rafaelson et al) reported that DRE correctly identified tumour penetration in 73% of cases examined.

Specifically for muscularispropria invasion; from two studies (Bipat et al. 2004, Chun et al. 2006), endoluminal ultrasound/endorectalsonography had the highest sensitivity (100%) and specificity (86%, range: 61.1%–86%) for muscularispropria invasion. Accuracy for endorectalsonography was 90.3%, similar to that of 3-T MRI (91.7%) but this was only reported in one study (Chun et al. 2006).

For perirectal tissue invasion; from two studies (Bipat et al. 2004, Chun et al. 2006) endorectalsonography had the highest sensitivity (100%; range 89%–100%) and accuracy (91.7%), whereas MRI had the highest specificity (92.6%; Range 71%–92.6%) compared to endorectalsonography/endoluminal ultrasound (81.5%; Range 75%–81.5%)

Adjacent organ involvement was specifically reported in 1 study and reported sensitivities and specificities ranged from 70% to 74% and 96%–97% respectively. MRI showed the highest sensitivity at 74%.

Mesorectal Fascia Involvement

Three studies reported on circumferential resection margin or mesorectal fascia involvement (Mercury Study Group, 2006, Rao et al, 2007, and Salerno et al, 2009); all studied used MRI and the Mercury Study Group also reported on DRE.

Salerno et al (2009) reported a significant higher rate of positive resection margins in patients with MRI stage T3/T4 tumours compared with patients with MRI stage T1/T2 tumours (36.7% versus 5.6%, p<0.001). Multivariate analysis showed MRI to be a significant predictor of positive margins (OR for stages T3/T4=15.2, p=0.002). The Mercury Study Group (2006) reported a sensitivity of 42%, 98% and accuracy of 92% for MRI in predicting circumferential margin involvement versus a sensitivity of 38%, specificity of 74% and accuracy of 70% for DRE.

Rao et al (2007) reported that mesorectal fascia was observed in all patients on MRI and found to be involved in 15/67 patients. The reported overall accuracy of predicting mesorectal fascia involvement was 88%, sensitivity was 80% and specificity was 90.4%.

T Stage

T-stage was reported in a number of papers (Kulinna et al. 2004; Bianchi et al. 2005; Akin et al. 2004; Fuchsjager et al. 2003; Halefoglu et al. 2008; Tatli et al 2006, Kim et al. 2006; Mainenti et al 2006, Fillipone et al. 2004; Rao et al. 2007), with some reporting results of comparisons for T-stage as a whole and some reporting results of comparisons for specific T stage. For overall T-stage, from five studies MRI had the highest sensitivity (93%; Range 55%–93%), specificity (9.14%; Range 63%–100%) and accuracy (89.7%; Range 43%–89.7%).

N Stage

N-stage was reported in 6 studies (Kulinna et al. 2004; Bianchi et al 2005; Halefoglu et al. 2008; Low et al. 2003; Tatli et al. 2006; Kim et al. 2006) and nodal involvement was reported in four studies (Kwok et al. 2000; Bipat et al. 2004; Chun et al. 2006; Kantorova et al. 2003). MRI has the highest sensitivity (85%; Range 62%–85%), specificity (98%; Range 69%–98%) and accuracy (95%; Range 64%–95%).

Nodal Involvement

Four studies reported nodal involvement (Kwok et al. 2000; Bipat et al. 2004; Chun et al. 2006; Kantorova et al. 2003) and MRI had the highest sensitivity (82%; Range 65%–82%), specificity (92.3%; Range 80%–92.3%) and accuracy (82%; Range 74%–79.2%).

References

• Akin O, Nessar G, Agildere AM, Aydog G. Preoperative staging of rectal cancer with endorectal MR imaging: Comparison with histopathologic findings. Journal of Clinical Imaging. 2004;28:432–438. [PubMed: 15531145]
• Beets-Tan RGH, Beets GL, Vliegen RFA, Kessels AGH, Van Boven H, De Bruine A, von Meyenfeldt MF, Baeten CGMI, van Engelshoven JMA. Accuracy of magnetic resonance imaging in prediction of tumour-free resection margin in rectal cancer surgery. The Lancet. 2001;357:497–504. [PubMed: 11229667]
• Beynon J, Mortensen NJ, Foy DMA, Channer JL, Virjee J, Goddard P. Preoperative assessment of local invasion in rectal cancer: digital examination, endoluminal sonography or omputed tomography. British Journal of Surgery. 1986;73:1015–1017. [PubMed: 3539255]
• Bianchi P, Ceriami C, Rottoli M, Torzilli G, Pompili G, Malesci A, Ferraroni M, Montorsi M. Endoscopic Ultrasonography and Magnetic Resonance in Preoperative Staging of Rectal Cancer: Comparison with Histological Findings. Journal of Gastrointestinal Surgery. 2005;9(9):1222–1227. [PubMed: 16332477]
• Bipat S, Glas AS, Slors FJM, Zwinderman AH, Bossuyt PMM, Stoker J. Rectal Cancer: Local Staging and Assessment of Lymph Node Involvement with Endoluminal US, CT and MR imaging – A Meta-Analysis. Radiology. 2004;232:773–783. [PubMed: 15273331]
• Brown G, Davies S, Williams, et al. Effectiveness of preoperative staging in rectal cancer: digital rectal examination, endoluminal ultrasound or magnetic resonance imaging? British Journal of Cancer. 2004;91:23–29. [PMC free article: PMC2364763] [PubMed: 15188013]
• Brown G, Richards C, Bourne A, Newcombe R, Radcliffe A, Dallimore N, Williams G. Morphological predictors of lymph node status in rectal cancer with the use of high-spatial resolution MR imaging with histopathological comparison. Radiology. 2003;227:371–377. [PubMed: 12732695]
• Brown G, Richard C, Newcombe R, et al. Rectal Carcinoma: Thin Section MR Imaging for staging in 28 patients. Radiology. 1999;211:215–222. [PubMed: 10189474]
• Chun HK, Choi D, Kim MJ, Lee J, Yun SH, Kim SH, Lee SJ, Kim CK. Preoperative Staging of rectal Cancer: Comparison of 3-T High Field MRI and Endorectal Sonography. American Journal of Roentgenology. 2006;187(6):1557–1562. [PubMed: 17114550]
• Dighe S, Purkayastha S, Swift I, et al. Diagnostic precision of CT in local staging of colon cancers: a meta-analysis. Clinical Radiology. 2010;65:708–719. [PubMed: 20696298]
• Dirisamer A, Halpern B, Flory D, et al. Performance of integrated FDG-PET/contrast enhanced CT in the staging and restaging of colorectal cancer: Comparison with PET and enhanced CT. European Journal of Radiology. 2010;73:324–328. [PubMed: 19200683]
• Fillipone A, Ambrosini R, Fushi M, Marinelli T, Genovesi D, Bonomo L. Preoperative T and N staging of colorectal cancer: Accuracy of Contrast-enhanced Multi-Detector Row CT Colonography – Initial Experience. Radiology. 2004;231:83–90. [PubMed: 14990815]
• Fuchsjager M, Maier A, Schima W, Zebedin E, Herbst F, Mittlbock M, Wrba F, Lechner G. Comparison of transrectal sonography and double-contrast MR imaging when staging rectal cancer. American Journal of Roentgenology. 2003;181(2):421–427. [PubMed: 12876020]
• Halefoglu A, Yildirim S, Avlanmis O, Sakiz D, Baykan A. Endorectal ultrasonography versus phased array magnetic resonance imaging for preoperative staging of rectal cancer. World Journal of Gastroenterology. 2008;14(22):3504–3510. [PMC free article: PMC2716612] [PubMed: 18567078]
• Kantorova I, Lipska L, Belohlavek O, Visokai V, Trubac M, Schneiderova M. Routine ¹⁸F-FDG PET Preoperative staging of colorectal cancer: comparison with conventional staging and its impact on treatment decision making. Journal of Nuclear Medicine. 2003;44(11):1784–1788. [PubMed: 14602860]
• Kim CK, Kim SH, Choi D, Kim MJ, Chun HK, Lee SJ, Lee JM. Comparison between 3-T Magnetic Resonance Imaging and Multi-Detector Row Computed Tomography for the Preoperative Evaluation of Rectal Cancer. Journal of Computer Assissted Tomography. 2007;31:853–859. [PubMed: 18043346]
• Kim CK, Kim SH, Chun HK, Lee WY, Yun SH, Song SY, Choi D, Lim HK, Kim MJ, Lee J, Lee SJ. Preoperative staging of rectal cancer: accuracy of 3-Tesla magnetic resonance imaging. European Radiology. 2006;16(5):972–980. [PubMed: 16416276]
• Kulinna C, Eibel R, Matzek W, et al. Staging of rectal cancer: diagnostic potential of multi-planar reformatting with multidetector CT. AJR. 2004;183:421–427. [PubMed: 15269036]
• Kulinna C, Scheidler J, Strauss T, Bonel H, Herrmann K, Aust D, Reiser M. Local staging of rectal cancer: assessment with double contrast multislice computed tomography and transrectal unltrasound. Journal of Computer Assisted Tomography. 2004 [PubMed: 14716245]
• Kwok H, Bisset IP, Hill GL. Preoperative Staging of Rectal Cancer. International Journal of Colorectal Disease. 2000;15(1):9–20. [PubMed: 10766086]
• Llamas-Elvira JM, Rodriguez-Fernandez A, Gutierrez Sainz J, Gomez-Rio M, Bellon-Guardia M, Ramos Font C, Rebollo Aguirre AC, Cabello Garcia D, Ferron Orihuela A. Fluorine-18 fluorodeoxyglucose PET in the preoperative staging of colorectal cancer. European Journal of Nuclear Medicine and Molecular Imaging. 2007;34(6):859–867. [PubMed: 17195075]
• Low RN, McCue M, Barone R, Saleh F, Song MR staging of primary colorectal carcinoma: comparison with surgical and histopathological findings. Abdominal Imaging. 2003;28(6):784–793. [PubMed: 14753591]
• Mainenti PP, Cirillo LC, Camera L, Perscio F, Cantalupo T, Pace L, De Palma GD, Persico G, Alvatore M. Accuracy of single phase contrast enhanced multidetector CT colonography in the preoperative staging of colorectal cancer. European Journal of Radiology. 2006;60:453–459. [PubMed: 16965883]
• Mercury Study Group. Extramural Depth of tumour invasion at thin section MR in Patients with rectal cancer: Results of the Mercury Study. Radiology. 2007;243(1):132–139. [PubMed: 17329685]
• Mercury Study Group. Diagnostic accuracy of preoperative magnetic resonance imaging in predicting curative resection of rectal cancer: prospective observational study. 2006. [PMC free article: PMC1602032] [PubMed: 16984925]
• Nicholls R, York Mason A, Morson B, et al. The clinical staging of rectal cancer. British Journal of Surgery. 1982;69:404–409. [PubMed: 7104612]
• Rafaelsen S, Kronborg O, Fenger C. Digital rectal examination and transrectal ultrasonography in staging of rectal cancer. Acta Radiology. 1994;35(3):300–304. [PubMed: 8192972]
• Rao SX, Zeng MS, Xu JM, QXU, Chen CZ, Li RC, Hou YY. Assessment of T-staging and mesorectal fascia status using high-resolution MRI in rectal cancer with rectal distention. World Journal of Gastroenterology. 2007;13(30):4141–4146. [PMC free article: PMC4205321] [PubMed: 17696238]
• Salerno G, Daniels I, Moran B, et al. Magnetic Resonance Imaging Prediction of an Involved Surgical Resection Margin in Low Rectal Cancer. Diseases of the Colon and Rectum. 2009;52(4):632–639. [PubMed: 19404067]
• Tatli S, Mortele K, Breen E, Bleday R, Silverman S. Local staging of rectal cncer using combined pelvic phased array and endorectal coil MRI. Journal of Magnetic Resonance Imaging. 2006;23(4):534–540. [PubMed: 16523466]
• Tateishi U, Maeda T, Morimoto T, Miyake M, Arai Y, Kim E. Non-enhanced CT versus contrast enhanced CT in integrated PET/CT studies for nodal staging of rectal cancer. European Journal of Nuclear Medicine and Molecular Imaging. 2007;34(10):1627–1634. [PubMed: 17530248]