Analysis of lesion localisation at colonoscopy: outcomes from a multi-centre U.K. study

Background Colonoscopy is currently the gold standard for detection of colorectal lesions, but may be limited in anatomically localising lesions. This audit aimed to determine the accuracy of colonoscopy lesion localisation, any subsequent changes in surgical management and any potentially influencing factors. Methods Patients undergoing colonoscopy prior to elective curative surgery for colorectal lesion/s were included from 8 registered U.K. sites (2012–2014). Three sets of data were recorded: patient factors (age, sex, BMI, screener vs. symptomatic, previous abdominal surgery); colonoscopy factors (caecal intubation, scope guide used, colonoscopist accreditation) and imaging modality. Lesion localisation was standardised with intra-operative location taken as the gold standard. Changes to surgical management were recorded. Results 364 cases were included; majority of lesions were colonic, solitary, malignant and in symptomatic referrals. 82% patients had their lesion/s correctly located at colonoscopy. Pre-operative CT visualised lesion/s in only 73% of cases with a reduction in screening patients (64 vs. 77%; p = 0.008). 5.2% incorrectly located cases at colonoscopy underwent altered surgical management, including conversion to open. Univariate analysis found colonoscopy accreditation, scope guide use, incomplete colonoscopy and previous abdominal surgery significantly influenced lesion localisation. On multi-variate analysis, caecal intubation and scope guide use remained significant (HR 0.35, 0.20–0.60 95% CI and 0.47; 0.25–0.88, respectively). Conclusion Lesion localisation at colonoscopy is incorrect in 18% of cases leading to potentially significant surgical management alterations. As part of accreditation, colonoscopists need lesion localisation training and awareness of when inaccuracies can occur.

surgical planning that is beneficial for both surgeon and patient. Recent work has suggested that the role of accurate colonoscopic lesion localisation has become increasingly important in the modern colorectal era for two reasons [2]. First, the establishment of the NHS Bowel Cancer Screening Programme (NHSBCSP) has led to the detection of earlier, and subsequently smaller, lesions that may not be visible on CT imaging [3]. Second, laparoscopic surgery with its reduced tactility, particularly with smaller lesions, is increasingly being offered to patients [4].
This study aimed to perform a large multi-centre audit across the U.K. to first assess the accuracy of colonoscopic lesion localisation and any subsequent operative consequences and second to determine potential influencing factors. The ALLaC study (Accurate Lesion Localisation at Colonoscopy) was registered with Clinical Effectiveness Unit, Greater Glasgow and Clyde. Each participating site registered locally as per local policies and procedures.

Methods and patients
Any patient undergoing elective curative surgical resection for a benign or malignant colorectal lesion/s that had undergone a colonoscopy prior to surgery was included. All patients had undergone colonoscopy either because of a positive faecal occult blood test through the NHS Bowel Cancer Screening Programme ('screener') or because they had been referred from primary care with colorectal symptoms ('symptomatic'). The local investigator at each participating centre identified patients from departmental multi-disciplinary colorectal cancer meetings and/or departmental operating lists, and the proforma was completed within 24 h of the patient undergoing surgery (''Appendix''). Patients who had undergone neoadjuvant chemo-radiotherapy as primary treatment or who underwent palliative surgery were excluded from this study.
The study proforma recorded patient, colonoscopic and imaging factors. Patient factors included: age, sex, BMI (body mass index), type of referral (screener or symptomatic) and previous abdominal surgery. Colonoscopic factors recorded were: caecal intubation [defined as 'passage of the scope beyond the ileocaecal valve into the caecal pole or terminal ileum (or anastomosis)' confirmed on electronic colonoscopy report] [21]; use of endoscopic scope guide; reasons for incomplete scope; experience/accreditation of the colonoscopist (Joint Advisory Group Accreditation awarded); tattoo placed and lesion localisation [22]. Imaging factors documented were: modality of imaging used and lesion localisation.
Lesion localisation was standardised into nine segments (caecum, ascending, hepatic flexure, transverse, splenic flexure, descending, sigmoid, rectum and 'other') with true lesion location defined as the intra-operative surgical location to allow comparison with colonoscopic and imaging localisation. Any changes to planned surgical management as a result of an alteration in lesion localisation intra-operatively were recorded.

Statistics
All categorical variables were analysed with the Chisquared test with 95% confidence intervals reported where appropriate. SPSS software (v22.0, Chicago, Illinois, USA) was used, and p values of less than 5% were taken as significant.
Results 365 forms were submitted for patients that had undergone colonoscopy followed by surgical resection in the eight centres across the U.K. On review of the forms, one case was excluded due to incomplete information in multiple sections. (Table 1) The mean age of the study population was 67.5 years (27-90 range). There were slightly more males (54%), with 75% being at least overweight and 39% recorded as obese. The majority of lesions found were in the colon, solitary, malignant and in patients referred with symptoms. 51% of the population had undergone previous abdominal surgery.

Colonoscopy factors (Table 2)
The majority of colonoscopists were accredited (60%), but only 36% used a scope guide. Only 36% tattooed a lesion's location. There was no clear pattern to why certain lesions were tattooed and others not, as the tattooing was performed throughout the nine segments of the colon. Caecal intubation was reported in 73% of cases, with reasons reported for failure as: obstructing lesion 68%, poor bowel preparation leading to lesion not seen 2%, looping sigmoid/colon preventing advancement 15% and not reported 15%. Except for the obstructing lesions, all incomplete colonoscopies went on to have further imaging by CT pneumocolon.
Colonoscopic lesion localisation (Table 3) Table 3 shows the true anatomical location of the colorectal lesions at surgery, with the majority being colonic.
The lesions are distributed throughout the nine segments but are mainly in the sigmoid (22.3%), rectum (28%) and right colon (32.7%). There were small non-significant variations between surgical and colonoscopic localisation in most of the segments. However, colonoscopy significantly underestimated the number of rectal lesions recorded (77 vs. 102) and significantly overestimated the number of sigmoid lesions (112 vs. 81).
Overall, two hundred and ninety-nine patients had their lesion/s located within the correct segment at colonoscopy (82%). (Table 4) All patients underwent a pre-operative CT (either abdomen/pelvis or CT colonography). Pre-operative CT visualised a colorectal lesion/s in only 73% of cases (265/361) with a significant reduction in visualising lesion/s in screening patients (64% screener vs. 77% in symptomatic; p = 0.008). Accurate localisation of lesion/s was reported in 213 cases, leading overall accuracy of CT to be 59%. Including only the cases where the lesion/s could be seen on CT, the accuracy increased to 80% (213/265).

Imaging lesion localisation and factors
Combining CT and colonoscopy localisation correctly localised the lesion in 87.1% of cases.
Only 21% underwent pre-operative MRI (n = 75) potentially reflecting the number of rectal lesions located at colonoscopy (n = 77). Of those cases, a lesion/s was localised in 65, with 64 cases being accurate. Overall lesion localisation for MRI was 88% (65/74 as one case localisation missing). Changes to intra-operative management Of the total number of lesion/s being incorrectly localised at colonoscopy (n = 65), 19 cases underwent a change in intra-operative management (5.2% of all cases) ( Table 5). The majority of these changes were a result of a lesion being incorrectly localised to the segment more proximal or distal at colonoscopy, with surgical management adapting accordingly. The majority of these cases were open procedures (n = 12) with 3 of the 7 laparoscopic cases being converted to open and 1 laparoscopic case performing an open rectal dissection, the later due to the tumour being more distal than thought pre-operatively. Further difficulties reported during these 19 altered surgical management cases included impalpable lesions or inappropriate/absent tattooing. Strategies employed to overcome these difficulties were: on-table colonoscopy with or without colorectal lavage; insertion of hand ports to allow palpation and accurate small lesion localisation; further port/s insertion to accommodate for alteration in surgical resection.
Factors influencing accurate lesion localisation at colonoscopy (Table 6) Analysis of patient and colonoscopic factors found that colonoscopy accreditation, use of the scope guide, caecal intubation and previous abdominal surgery all significantly influenced accurate lesion localisation. On multi-variate analysis, both caecal intubation and use of the scope guide

Discussion
This prospective multi-centred U.K. study has found colonoscopy to incorrectly localise colorectal lesions in 18% of cases, leading to altered surgical management in theatre in 5.2% of cases. Combining the locations of colonoscopy and CT increased correct lesion localisation to 87.1%. However, this study has confirmed that CT imaging does not visualise lesions in over a quarter of cases, particularly in the screening population. With earlier and smaller lesions expected to continue to be detected with NHSBCSP, the role of colonoscopy in optimal pre-operative surgical planning is likely to become increasingly important in the modern era. Previous publications have reported significant variability in the accuracy of colonoscopy to localise lesions, from 59.7 to 98.3% [5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]. In a previous publication from this group, Bryce et al. [2] reviewed the literature highlighting many of these studies have been retrospective, small in number, single centre or single endoscopist in design. With this heterogeneity, several influencing factors have been proposed to influence lesion localisation: increasing age; previous abdominal surgery and incomplete colonoscopy [5,8,14]. This is in comparison with this current work that has encompassed eight U.K. hospital sites with over fifty different colonoscopists, several of whom would have been blinded as the data were recorded perioperatively, not at the time of the colonoscopy. Furthermore, the larger number of patients in this study allows for further statistical analysis of influencing factors, with use of the scope guide, colonoscopic accreditation, caecal intubation and previous abdominal surgery all shown to be significant.
The scope guide is regarded by many colonoscopists as a teaching aid and is not always routinely available. Indeed the guidelines for JAG accreditation [22] state that as part of achieving accreditation, use of the scope guide is not mandatory. Results from this audit suggest that this approach needs to be changed as routine use of the scope guide could educate colonoscopists leading to improved lesion localisation and as a result should be available on all screening and symptomatic colonoscopy lists.
Experience of the colonoscopist was also significant with those not JAG accredited having greater inaccuracies of localising lesions. Currently, only JAG accredited colonoscopists can perform screening lists independently and perhaps this criterion should be applied to symptomatic referrals so that in the event of diagnosis of a lesion requiring surgical resection optimal localisation can occur [22]. In addition, it is worth highlighting that to achieve accreditation, colonoscopists must achieve competence in four domains: assessment; consent and communication; endoscopic skills and diagnostic and therapeutic ability [23]. Within these domains, focused training on localisation could be made mandatory. Education would be the first step as many, particularly non-surgeons, may not be aware of the difficulties for both patient and surgeon that incorrect localisation and/or inappropriate tattooing can cause. Included in this approach could be the proposal that all recto-sigmoid lesions undergo rigid proctoscopy to confirm tumour height as an incorrectly located rectal tumour that alters abdominal surgery to pelvic surgery has major implications for the patient (higher anastomotic leak risk and urinary and sexual dysfunction). These changes could be supported by the introduction of specific and mandatory questions on the electronic colonoscopy record about lesion localisation: tattoo yes/no; tattoo sites (with proximal and distal options suggested depending on lesion site); rigid proctoscopy performed yes/no. Currently, it is optional for the colonoscopist to include these pieces of information which may partly explain the low number of lesions tattooed in this study (36%).
Incomplete colonoscopy means that the colonoscopist has less visual exposure to the landmarks that they are trained to recognise and the complete colonoscopy allows two views (insertion and withdrawal) to increase the probability of correctly localising a lesion. The two main reasons for incomplete colonoscopy were obstructing lesions and sigmoid loops and, especially if the scope guide is not used, one can see why the colonoscopist could become disorientated in this situation.
Previous abdominal surgery included all types, not just colorectal resections, and the experienced colonoscopist is aware that previous surgery can make gentle passage of the scope difficult, particularly in women who have undergone a hysterectomy. However, colonoscopists and surgeons must now be aware that in this same group of patients, localisation can also be difficult. Only a small number of patients in this audit had undergone a resection of the colon or rectum, making conclusions about lesion localisation in this specific population limited; however, one extreme case is presented in this study where an anastomotic recurrence was not correctly reported leading to a surgical alteration in management. There was only a 5.2% change in on-table management due to inaccurate lesion localisation at colonoscopy. This figure would have been higher if not for CT in combination with colonoscopy, increasing the number of correctly localised lesions (to 87.1%). These incorrect localisations mainly occurred at a more proximal or distal segment with no change in management. However, when a change in management occurred (29% of inaccurate cases), it was significant with 3 laparoscopic cases converted to open and one requiring an unexpected open TME dissection. This raises the possibility that incorrect lesion localisation could make laparoscopic surgery vulnerable to ontable alterations.
From a surgical viewpoint, these results demonstrate key areas where incorrectly pre-operative localisation can have significant operative changes: hepatic and transverse colon (decision to take middle colics and further dissection for an extended right hemicolectomy); splenic flexure (extended right hemicolectomy vs. left hemicolectomy); sigmoid versus rectum (to perform a total mesorectal excision/covering ileostomy and consideration to taking down the splenic flexure). It is beyond the remit of this work to document what the short-and long-term implications for the patients with altered management were; however, with the additions of on-table colonoscopy/lavage, further dissection and extra ports, it is unsurprising that many proformas made the comment 'increased operating time'.

Limitations
This is not a consecutive series of patients, so case selection bias may be present. Identification of the patients was left to the local investigator who may have included or excluded patients for various reasons. Excluding the authors, the majority of the endoscopists were blinded to this study. 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
Surgical planning pre-operatively is becoming increasingly reliant on accurate lesion localisation at colonoscopy. Colonoscopists need to be educated of the key anatomical areas where inaccuracies occur and have increased vigilance where caecal intubation has been unsuccessful. Routine clinical use of the scope guide could potentially increase correct localisation, minimising on-table alterations in surgical management and optimising outcomes for surgeons and patients.