The accuracy of colonoscopic localisation of colorectal tumours: a prospective,multi-centred observational study

Abstract

Background and aims

Colonoscopy is essential for accurate pre-operative colorectal tumour localisation, but its accuracy for localisation remains undetermined due to limitations of previous work. This study aimed to establish the accuracy of colonoscopic localisation and to determine how frequently inaccuracy results in altered surgical management.

Method

A prospective, multi-centred, powered observational study recruited 79 patients with colorectal tumours that underwent curative surgical resection. Patient and colonoscopic factors were recorded. Pre-operative colonoscopic and radiological lesion localisations were compared to intra-operative localisation using pre-defined anatomical bowel segments to determine accuracy, with changes in planned surgical management documented.

Results

Colonoscopy accurately located the colorectal tumour in 64/79 patients (81%). Five out of 15 inaccurately located patients required on-table alteration in planned surgical management. Pre-operative imaging was unable to visualise the primary tumour in 23.1% of cases, a finding that was more prevalent amongst bowel screener patients compared to symptomatic patients (45.8% vs. 13%; p = 0.003).

Conclusion

Colonoscopic lesion localisation is inaccurate in 19.0% of cases and occurred throughout the colon with a change in on-table surgical management in 6.3%. With CT unable to visualise lesions in just under a quarter of cases, particularly in the screening population, preoperative localisation is heavily reliant on colonoscopy.

Keywords

Colonoscopy lesion localization accuracy

Introduction

Colonoscopy is the current optimal investigation for the detection of colorectal cancer and forms the basis of pre-operative lesion localisation alongside radiological imaging. Accurate pre-operative localisation of colorectal tumours is becoming increasingly important in modern colorectal surgery with the detection of smaller lesions as a result of the NHS Bowel Cancer Screening Programme¹ and the reduced tactility associated with increasingly performed laparoscopic surgery.² The potential adverse outcomes of inaccurately locating a tumour pre-operatively include requirement for on-table alteration in planned surgical resection, on-table colonoscopy to aid localisation, insertion of further laparoscopic ports to accommodate an unexpected tumour position and conversion from a laparoscopic to open procedure; all of which have the potential to increase anaesthetic time, alter post-operative recovery and modify patients’ expectations. Indeed, one study by the American Society of Colon and Rectal Surgeons found that 6.5% of surgeons who routinely performed laparoscopic resections admitted to removing the incorrect piece of bowel on at least one occasion.³

Previous work^4–14 has attempted to document the accuracy of lesion localisation by colonoscopy with wider ranging inaccuracies documented from 4% to 40.5%. In addition to the majority of data collection for these studies being pre-2006 and out-with the laparoscopic and bowel screening era, all are retrospective in design with only one study being multi-centred. Furthermore, some of these studies excluded certain tumour locations or only analysed one location (e.g. caecum). As a result, there is no widely accepted figure for colonoscopic localisation accuracy.

The primary aim of this study was to determine the accuracy of colonoscopic localisation of colorectal tumours when compared to intra-operative lesion localisation. Secondary aims were to determine what proportion of any inaccurate localisations required on-table alteration in planned surgical management and to calculate the accuracy of pre-operative imaging in localising colorectal tumours for comparison.

Patients and methods

Study design

This was a prospective, multi-centre observational study that recruited patients from four hospitals in the West of Scotland over a 6-month period (October 2011–March 2012). The study was registered with the Clinical Effectiveness Department for Greater Glasgow and Clyde. All patients underwent a pre-arranged colonoscopy as a result of either having colorectal symptoms (symptomatic patients) or following a positive faecal occult blood test as part of the NHS Bowel Cancer Screening Programme (screening patients). Patients were included if they had a colorectal tumour (benign or malignant) identified at colonoscopy for which they underwent CT and/or MRI imaging and subsequent elective curative surgical resection (open or laparoscopic).

Patients were excluded if the patient had more than one CT/MRI scans (i.e. patients with rectal cancers that were undergoing neo-adjuvant chemo-radiotherapy) or if surgical resection was abandoned intra-operatively due to the presence of unexpected metastases making intra-operative lesion localisation difficult.

Colonoscopy, CT/MRI imaging and surgical resection

Colonoscopy was performed according to the individual colonoscopist’s practice. Recorded patient factors were age, sex, symptomatic or screening patients. Colonoscopic factors were incomplete colonoscopy and colonoscopic tumour location. To standardise reporting of tumour localisation, the bowel was divided into eight sections (rectum, sigmoid colon descending colon, splenic flexure, transverse colon, hepatic flexure, ascending colon and caecum).

Radiological imaging was performed as part of the patients’ clinical pathway and was directed by the responsible clinician. At the time of surgical resection, the operating surgeon recorded intra-operative tumour localisation, according to the same pre-defined anatomical categories, and documented any change in the planned surgical resection with explanation for this change.

Statistical analysis

To analyse the accuracy of colonoscopic lesion localisation, intra-operative lesion localisation was defined as the true location of the lesion. A one-proportion power calculation found that 79 patients were required to detect a significant difference between a hypothesised best possible accuracy of 93.75% (based on the highest level of accuracy that could be found in the literature) and a clinically significant inaccuracy threshold of 15%, at a two-tailed significance level of 0.05 and 80% power. To minimise the selection bias, all patients meeting the inclusion criteria were entered into the study in the chronological order they underwent colonoscopy until the required sample size of 79 patients was achieved.

To determine the role of pre-operative imaging in lesion localisation, location of the lesion on CT and/or MRI was recorded using the same anatomical divisions as colonoscopy, allowing accuracy to be calculated and compared. All statistical analyses were performed using SPSS (version 18.0, Chicago, Illinois, USA).

Results

Participants

A total of 94 consecutive patients were identified for inclusion in this study, with 15 patients subsequently excluded, leaving 79 patients for the final formal analysis (Figure 1). All patients had one colorectal tumour identified at colonoscopy; no patient had synchronous tumours. Mean age of patients was 67.3 years (SD: 10.2; range, 43–89) with a male to female ratio of 1.5:1. The majority of patients were symptomatic referrals (n = 55) with the remaining patients (n = 24) presenting as part of the screening programme.

Figure 1.

Flow chart of patient recruitment and patient baseline characteristics.

Colonoscopic outcomes and lesion localisation accuracy

Colonoscopies were performed by 39 different colonoscopists (of which 37 were JAG accredited): consultants in 63 (79.7%) cases, specialist nurse practitioners in 14 (17.7%) and surgical trainees (grades ST 6) in 2 (2.5%). The magnetic endoscopic imaging ‘scope guide’ was used in 25 (31.6%) of colonoscopies. Complete colonoscopy was achieved in 63 (79.7%) cases with the majority of incomplete colonoscopies resulting from obstructing/stenosing tumours (n = 13). The remaining three incomplete colonoscopies were due to poor bowel preparation (n = 1), patient became clinically unstable during colonoscopy (n = 1) and lesion localisation already defined by CT pneumocolon so colonoscopy not advanced beyond lesion (n = 1). The tumours were distributed throughout the colon and rectum with the sigmoid colon (n = 29, 36.7%) being the commonest tumour site (Figure 2).

Figure 2.

True anatomical distribution of all colorectal tumours defined intra-operatively.

Of the 79 lesions localised at colonoscopy, 64 (81%) were accurately located in the same position intra-operatively. The 15 lesions that were inaccurately localised were distributed throughout the colon: rectum (n = 3), sigmoid (n = 3), descending colon (n = 3), splenic flexure (n = 2), transverse colon (n = 1), ascending colon (n = 2) and caecum (n = 1) (Figure 2). Of these 15 lesions, 10 patients required no change in surgical management due to the actual intra-operative lesion location still being included within the planned surgical resection.

The remaining five (6.3% of all 79 patients) patients did require an on-table alteration in planned surgical procedure (Table 1). Out of these five, three lesions related to the descending colon/sigmoid colon/rectum and one related to the hepatic flexure/transverse colon. The remaining inaccurately localised lesion occurred in a patient that had undergone a previous right hemicolectomy and was thought to have recurrence at the anastomosis at colonoscopy. Intra-operatively, the lesion was found to be in the sigmoid, and the surgical management changed from resection of the anastomosis to a subtotal colectomy.

Table 1.

Inaccurate colonoscopic lesion localisation that resulted in an alteration in surgical resection (n = 5).

Number of patients	Colonoscopy lesion localisation	Intra-operative lesion localisation	Planned surgical resection	Actual surgical resection/management
1	Sigmoid	Mid-rectum	Anterior resection	Surgery abandoned for neo-adjuvant therapy.
2	Sigmoid	Descending colon	Sigmoid colectomy	Left hemicolectomy
1	Hepatic flexure	Transverse colon	Right hemicolectomy	Extended right hemicolectomy
1	Ileo-colic anastomosis	Sigmoid colon	Resection of ileocolic anastomosis	Subtotal colectomy

Accuracy of pre-operative radiological imaging in lesion localisation

Of the 79 patients, 78 underwent at least one pre-operative imaging modality (CT or both CT and MRI in the case of rectal tumours). The remaining patient underwent a CT scan within the first week after surgery.

Pre-operative imaging accurately located the lesion in only 67.1% of all 78 cases. This is due to pre-operative imaging being unable to identify the primary tumour in 18 of 78 (23.1%) patients. In addition, screening patients were significantly more likely to have unidentified tumours on pre-operative imaging when compared to symptomatic patients (45.8% vs. 13%; p = 0.003). Of the 60 patients where the lesion could be visualised by pre-operative imaging, the lesion was correctly localised in 53 (88.3%) patients.

The addition of imaging localisation to the 15 inaccurately localised colonoscopic lesions correctly localised seven of these patients resulting in a combined accuracy of 91.0% (71/78 patients). In the remaining eight patients, imaging incorrectly localised the lesion (n = 5), could not visualise the primary tumour (n = 2) and was not performed pre-operatively (n = 1). The proportion of patients with inaccurate imaging localisation was significantly higher for those with inaccurate colonoscopic localisation compared to those with accurate colonoscopic localisation (41.7% vs. 4.3%; p = 0.003). In all five cases where both imaging and colonoscopy were inaccurate, each modality concurred on the same wrong localisation.

Discussion and conclusions

This pragmatic study has prospectively assessed lesion localisation at colonoscopy, finding it to be inaccurate in 19% of cases, resulting in an on-table alteration in planned surgical resection in 6.3% of cases. This study provides a true reflection of the current accuracy of lesion localisation at colonoscopy in the era of the NHS Bowel Screening Programme.

A literature review identified 12 studies that stated as their primary aim, the accuracy of colonoscopic localisation where colonoscopic localisation accuracy ranged from 59.5% to 96.0%.^4–15 As stated in the introduction, these studies have varying methodologies, making comparisons with each other and this study difficult. To the authors’ knowledge, the current study is the first to combine a prospective, multi-centred and powered approach to investigating colonoscopic localisation accuracy.

One of the larger retrospective studies assessing colonoscopic lesion localisation analysed 374 patients over a 17-year period, finding colonoscopic lesion localisation inaccurate in only 4% of patients, with a change in management in 3% of all patients.⁴ The majority of inaccurate lesions were in the descending/sigmoid/rectum, divisions that can be a difficult area for landmarks at colonoscopy, a finding that is supported by the results of this study. However, this work is limited as all the colonoscopies were performed by the senior author and that the anatomical bowel divisions used were fewer than the current study, which will bias towards improved accuracy.

Similar colonoscopic lesion localisation inaccuracies to the current study have been found. For example, an older study by Piscatelli et al.⁷ retrospectively analysed a series of 236 colorectal cancer patients finding inaccurate colonoscopic lesion localisation in 49 patients (21%) when compared to intra-operative localisation. Furthermore, 31 patients (13%) required on-table alterations in planned surgical management.⁷ Similarly, Lam et al.¹⁰ found an inaccuracy of 19% in their retrospective analysis of 84 colorectal cancer patients with the highest accuracy in the descending colon and splenic flexure (100%) and lowest accuracy in the transverse colon (50%). It would be reasonable to assume that some anatomical areas provide more reliable landmarks for the colonoscopist (ileocaecal valve and hepatic flexure as examples). However, what this study and others are showing is that all areas in the colon and rectum are vulnerable to inaccurate localisation, a factor that any colonoscopist should be made aware of.

The five previous studies that assessed alterations in surgical management following inaccurate colonoscopic localisation found a wide range from 0.4% to 16% of cases that are likely to reflect the heterogeneity of the study methodology.^4,7,12,13,15 In the current study, 6.3% (n = 5) of all study participants required altered surgical management, which seems to suggest that the majority of pre-operative localisation errors do not change the planned surgical resection. For example, on one occasion, a lesion documented at colonoscopy to be in the caecum, which was actually in the ascending colon, was categorised as inaccurate, but did not alter the planned right hemicolectomy.

In the remaining five patients with inaccurate colonoscopic localisation, the change in surgical management varied. In three of these patients, the lesion was found in an adjacent anatomical division. Although the different location did not result in any differences that the patient would be aware of, the surgery and anaesthetic were prolonged in all cases to ensure accurate localisation of the lesion followed by more extensive colonic mobilisation and resection than was planned. Fortunately, in all three cases, the surgery was open (reflecting surgeon choice) allowing direct palpation when the lesion was not in the expected area. Therefore, reliance on tattooing and on-table colonoscopy was not required nor was there any change in laparoscopic port positions, conversion to open or consideration to patient position to mobilise the splenic flexure that could have occurred if laparoscopic surgery was performed.²

The two remaining patients requiring on-table alteration in surgical management had completely unexpected alterations. In one, pre-operative colonoscopy and CT had both diagnosed a low-sigmoid tumour that was suitable for surgical resection. In theatre, the tumour was found to be in the mid-rectum, bulky and fixed posteriorly with surgery abandoned to allow neo-adjuvant chemo-radiotherapy. In the second, the patient had undergone a right hemicolectomy for a caecal cancer previously, and the colonoscopist documented a recurrence at the ileo-colic anastomosis, which was found to be more distal in the sigmoid, at laparotomy. With metachronous tumours diagnosed, the surgical management was altered to a subtotal colectomy with a subsequent increase in surgical and anaesthetic duration in addition to implications for the patient’s long-term bowel habit. One previous author has highlighted the potential difficulties in colonoscopically localising lesions in patients with previous colorectal resection,⁷ but as the patient in the current study was only one of three patients to have had a previous colorectal resection, further work needs to be performed before making conclusions.

As part of the current clinical practice, radiological imaging works alongside colonoscopy in pre-operative lesion localisation and assessment of metastatic spread. In this study, imaging could not visualise and subsequently not localise nearly a quarter of colorectal lesions (23.1%). With radiological imaging continuing to evolve, this figure is unexpectedly high but is supported by a recent study that found that 28/237 (11.8%) tumours were not visible on CT.¹³ It is likely that the recruitment of screening patients in this study accounted for this low-radiological visualisation of colorectal tumours as a higher proportion of imaging-invisible lesions were present in screening patients when compared to symptomatic patients. Surgeons should be aware that up to a quarter of their patients, especially those referred through the bowel cancer screening programme, may not have an imaging-estimated lesion localisation available and therefore will have to plan surgery based on colonoscopy alone.

However, the role of radiology is quite different when imaging does visualise the colorectal lesions: pre-operative CT and/or MRI achieved a higher accuracy of colorectal lesion localisation than colonoscopy (88.3% vs. 81.0%). Additionally, radiological imaging correctly localised 7 of the 15 lesions inaccurately located by colonoscopy giving overall the highest accuracy of 91% when colonoscopy is combined with CT imaging. This suggests that current clinical practice should take both colonoscopic and imaging localisation into account to optimise surgical planning.

Limitations

The main limitation of this study is that the surgeon and radiologist were not blinded to the result of the lesion localisation at colonoscopy, leading to verification bias. This limitation, however, reflects the pragmatic nature of the study and current clinical practice.

Conclusion

With the introduction of the NHS bowel screening programme screening and the increasing use of laparoscopic colorectal surgery, preoperative lesion localisation is increasingly important for optimal surgical management. Colonoscopic lesion localisation is inaccurate in 19.0% of cases and occurs throughout the colon with a change in on-table surgical management in 6.3%. With CT unable to visualise lesions in just under a quarter of cases, particularly in the screening population, preoperative localisation is heavily reliant on colonoscopy. Future work is required to determine factors that optimise lesion localisation at colonoscopy.

Footnotes

Authorship and contributions

MS Johnstone – Design of study, acquisition of data, analysis and interpretation of data, drafting article.

SJ Moug – Conception, design of study, acquisition of data, analysis and interpretation of data, drafting article, revising article critically for important intellectual content and final approval of version to be published.

Acknowledgements

The study was registered with the Clinical Effectiveness Department of Greater Glasgow and Clyde.

Declaration of conflicting interests

None declared.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

*

Collaborators: A Renwick, M Vella, C Porteous, L Chisholm, PS Chong, RG Molloy, M Duff, D Galloway, G Mackay, B Page, RF McKee, JH Anderson, D Chong, A McMahon, PG Horgan; T McKenna, A Sutherland, M McCormack and I MacCallum. This study adhered to STROBE Guidelines ().

References

NHS Bowel Cancer Screening Programme, www.cancerscreening.nhs.uk/bowel (accessed 27 July 2013).

Hospital Episode Statistics. Proportion of colorectal resections undertaken laparoscopically in England, www.lapco.nhs.uk/activity-latest-HES-data.php (2012, accessed 27 June 2013).

Wexner

Cohen

Ulrich

. Laparoscopic colorectal surgery – are we being honest with our patients? Disease Colon Rectum 1995; 38: 723–727.

Vaziri

Choxi

Orkin

. Accuracy of colonoscopic localization. Surg Endosc 2010; 24: 2502–2505.

Solon

Al-Azawi

Hill

. Colonoscopy and computerized tomography scan are not sufficient to localize right-sided colonic lesions accurately. Colorectal Disease: Off J Assoc Coloproctol Great Britain Ireland 2010; 12: e267–e272.

Cho

Lee

Yun

. Tumor localization for laparoscopic colorectal surgery. World J Surg 2007; 31: 1491–1495.

Piscatelli

Hyman

Osler

. Localizing colorectal cancer by colonoscopy. Arch Surg 2005; 140: 932–935.

Feuerlein

Grimm

Davenport

. Can the localization of primary colonic tumors be improved by staging CT without specific bowel preparation compared to optical colonoscopy? Eur J Radiol 2011; 81: 2538–2542.

Ellul

Fogden

Simpson

. Colonic tumour localization using an endoscope positioning device. Eur J Gastroenterol Hepatol 2011; 23: 488–491.

10.

Lam

Kwong

Lam

. How useful is colonoscopy in locating colorectal lesions? Surg Endosc 1998; 12: 839–841.

11.

Vignati

Welch

Cohen

. Endoscopic localization of colon cancers. Surg Endosc 1994; 8: 1085–1087.

12.

Louis

Nandipati

Astorga

. Correlation between preoperative endoscopic and intraoperative findings in localizing colorectal lesions. World J Surg 2010; 34: 1587–1591.

13.

Borda

Jimenez

Borda

. Endoscopic localization of colorectal cancer: study of its accuracy and possible error factors. Rev Esp Enferm Dig 2012; 104: 512–517.

14.

Lee

Voytovich

Pennoyer

. Accuracy of colon tumor localization: computed tomography scanning as a complement to colonoscopy. World J Gastrointest Surg 2010; 2: 22–25.

15.

Hancock

Talbot

. Accuracy of colonoscopy in localisation of colorectal cancer. Int J Colorectal Dis 1995; 10: 140–141.