SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

Summary

Background

The cancer risk of low-grade dysplasia (LGD) in chronic ulcerative colitis is variable and its management remain contentious.

Aim

To determine the risk of cancer or any advanced lesion once LGD is diagnosed.

Methods

A MEDLINE, EMBASE and Pub Med search was conducted using the key words ‘surveillance’, ‘colorectal cancer’, ‘low-grade dysplasia’ and ‘ulcerative colitis’. A random effects model of meta-analysis was used.

Results

Twenty surveillance studies had 508 flat LGD or LGD with dysplasia-associated lesion or mass. An average of 4.3 colonoscopies was performed/patient post-LGD diagnosis (range: 3–7.6). An average of 18 biopsies taken per colonoscopy (range: 9–24) detected 73 advanced lesions (cancer or high-grade dysplasia) pre-operatively. The cancer incidence was 14 of 1000 (95% CI: 5.0–34) person years duration (pyd) and the incidence of any advanced lesion was 30 of 1000 pyd (95% CI: 12–76). When LGD is detected on surveillance there is a ninefold risk of developing cancer (OR: 9.0, 95% CI: 4.0–20.5) and 12-fold risk of developing any advanced lesion (OR: 11.9, 95% CI: 5.2–27).

Conclusions

The risk of developing cancer in patients with LGD is high. These estimates are valuable for decision-making when LGD is encountered on surveillance.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

Ulcerative colitis (UC) is a chronic relapsing inflammatory bowel disease of uncertain etiology. Although prevalent world wide it predominantly affects North Americans and Europeans.1 Chronic colitis predisposes to colorectal cancer and the risk increases with time.2 Progression from inflammation to colorectal cancer is supposed to follow a sequence from colitis without dysplasia, low-grade dysplasia (LGD), high-grade dysplasia and eventually carcinoma.3 However, recent reports have suggested that LGD can progress to colorectal cancer without the intermediate stage of high-grade dysplasia.4 Therefore, LGD can be considered as the definitive interventional point where prophylactic colectomy for colorectal cancer in chronic UC can be instituted.

The optimum management of LGD in patients with UC remains contentious. Some advocate intensive colonoscopic surveillance while others recommend early colectomy. Patients with sporadic adenomas or adenoma like mass (ALM: raised or polypoidal dysplasia without surrounding flat dysplasia) can be successfully treated with endoscopic polypectomy or endomucosal resection. Surveys have shown that adoption of any particular approach may be influenced by a physician’s individual perception and lacunae in our current knowledge of the disease.5, 6 This conundrum is due to many factors: variable and uncertain natural history of the disease, difficulty in histological interpretation during the active phase of colitis and wide inter-observer variability in the histological diagnosis of dysplasia even among expert pathologists.7, 8 Data on the natural history of the disease are largely based on prospective and retrospective cohort studies. The variable results in cancer incidence rates reflect the heterogeneity between studies in terms of different patient populations, extent of colitis, number of surveillance biopsies taken and duration of surveillance. Based on these studies some centres have adopted a conservative approach9 where as others advocate colectomy.10 Guidelines published by the British and American Societies of Gastroenterology address these problems to a certain extent.11, 12 A randomised prospective multicentre study comparing continuing surveillance vs. colectomy at LGD diagnosis with colorectal cancer related deaths, quality of life and death from all causes, as primary outcome measures would be ideal to resolve this debate. Lennard-Jones et al. points out the overwhelming problems associated with such a study in terms of ethical and practical difficulties.13 An acceptable compromise would be a meta-analysis of all surveillance studies to analyse the incidence and the relative risk of LGD progressing to an advanced lesion, i.e. high-grade dysplasia or colorectal cancer.

Aims and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

The aim of this meta-analysis was to determine the incidence of cancer and the relative risk of developing cancer in patients with LGD in chronic UC undergoing surveillance. We also aimed to determine the incidence of LGD in chronic ulcerative colitis and the positive predictive value of LGD for concurrent cancer and progression to cancer. The guidelines produced by the NHS Centre for Reviews and dissemination York University were followed for the meta-analysis.14

Identification of studies

MEDLINE (1966 to July 2005), EMBASE (1986 to July 2005) and a PubMed were searched using the keywords ‘ulcerative colitis, ‘dysplasia’, ‘colorectal cancer’, and ‘surveillance’. In addition, the references lists of the original articles and relevant review articles were scrutinized to identify studies missed on the initial search. The studies to be included in the analysis were reviewed un-blinded by two authors (TT and RJR). Studies were included based on a predefined inclusion criterion.

The following data were recorded on a proforma (Appendix A): type and year of study, number of patients with ulcerative colitis, number of patients with low-grade dysplasia and dysplasia-associated lesion or mass (DALM), duration of study, duration of follow-up in patient years, number of colonoscopies per patient, mean number of surveillance biopsies per colonoscopy, number of patients progressing to an advanced lesion, i.e. colorectal cancer (CRC) or high-grade dysplasia or DALM, cancer stage, number of deaths as a result of CRC and time taken to progress form LGD to CRC. In studies published prior to 1983 (classification of dysplasia was changed in 1983)15 where dysplasia was classified as mild, moderate and severe, only cases with mild dysplasia were considered as low-grade and included in the analysis. Studies where more than one pathologist reviewed histology, we included the reviewed results in the analysis.

Inclusion criteria

Studies in the English language where the study population (histologically proven UC and LGD), interventions (surveillance colonoscopy or colectomy) and outcomes (CRC, high-grade dysplasia or DALM at colectomy and/or surveillance) were clearly defined and documented were included in the analysis.

Exclusion criteria

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

The following types of studies were excluded from the analysis:

  • (i)
    Studies with data only on colorectal cancer in colitis.
  • (ii)
    Studies involving only high-grade dysplasia.
  • (iii)
    Studies which included predominantly indeterminate and Crohn’s colitis in a combined analysis.
  • (iv)
    Studies using magnifying chromendoscopy for detecting dysplasia.
  • (v)
    Studies with inadequate follow-up data (outcome data in the subgroup of patients with LGD in terms of progression to DALM, high-grade dysplasia or cancer was not documented in the studies).

Data extraction

One author (TT) read each paper and extracted data using a predefined review form (Appendix A).

Definitions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices
  • (i)
    Prevalent cancer or LGD: Cancer or LGD found within 6 months of initiating surveillance either on colonoscopy or colectomy.
  • (ii)
    Incident cancer or LGD: Cancer or LGD found at colectomy or colonoscopy after 6 months of initiating surveillance. LGD found at colectomy in patients with no dysplasia or indefinite for dysplasia was also considered incident LGD.
  • (iii)
    Advanced lesions: Flat LGD was considered to progress to an advanced lesion if high-grade dysplasia, CRC or DALM was detected on surveillance. LGD in the presence of DALM was considered to progress to an advance lesion if HGD or CRC was detected on surveillance.

Statistical analysis

All statistical analyses were done using STATA (Statacorp, Texas, Version 9.0, 1997) meta-analysis programmes.16, 17 The heterogeneity between studies was calculated using the Chi-square test for homogeneity (Q-value) and a value of < 0.05 was considered statistically significant.18 As there were significant heterogeneity between the studies a random model of meta-analysis was used as described by DeSimion and Laird.19 Overall the pooled estimate, with the 95% confidence intervals of the incidence of cancer, any advanced lesion and LGD was obtained using the random effects model of meta-analysis and the logs odds or log incidence scales were used as appropriate. Changes in log incidence over time were calculated using the mixed effects meta-regression technique.20 The size of the circles in are inversely proportional to the variance associated with the log incidence rates in each study (the larger the circle the smaller the study) and the regression line was estimated using the mixed effect meta-regression techniques. Follow-up in patient years duration (pyd) was extracted from the articles. When the follow-up in pyd was not specifically documented it was calculated using the mean/median follow-up duration and the total number of patients with LGD. Begg’s test21 and Egger’s test22 was used to determine publication bias.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

From an explosion of abstracts, 776 relevant abstracts were reviewed. Of the 776 abstracts 52 articles were scrutinized in depth. Thirty-two studies did not full fill the predefined exclusion criteria. Hence 20 studies3, 4, 8–10, 23–37 were included in the final analysis (Table 1).

Table 1.   Table showing the various studies with outcome data on patients with LGD detected on surveillance for chronic ulcerative colitis
AuthorYearfLGD + DALM (n)Follow-up in pyd for LGDAL on Col or SurAL on Col at initial diagnosis or for NPLGD
  1. pyd, patient years duration; AL, advanced lesions; LGD, low-grade dysplasia; DALM, dysplasia-associated lesion or mass; NPLGD, nonprogressive flat LGD or DALM; Col, colectomy; Sur, surveillance colonoscopy. * Follow-up in pyd in 38 patients. ** Two patients had colectomy for refractory disease.

Befritis20026050450
Biasco19947No data40
Blackstone198225No data04
Brostrom19866No data20
Connell199412No data80
Hata20039No data00
Jonsson199420No data40
Lashner198920No data60
Leidenius19918No data00
Lennard J*199076266.698
Lim20032929040
Lindberg199637No data013
Lofberg19907No data30
Lynch19944022310
Nugent199117No data42
Rosenstock198541No data01
Rozen1995831.220
Ullman T**20021848.693
Ullman T20034657.572
Woolrich199222No data51

Overall analysis

The study analysed 508 patients with LGD (477 patients with fLGD and 31 patients with LGD in the presence of DALM) in more than 2677 patients with chronic ulcerative colitis (the total number of patients with UC was not specified in 3 studies4, 10, 25) who were enrolled into a colorectal cancer surveillance programme (average duration of colitis prior to entry into the study was 12 years). Overall an average of 3.6 colonoscopies were done per patient (range: 1.5–8.4) in 17 studies3, 8, 9, 23, 24, 26–37 over a mean surveillance period of 12 years (9713 colonoscopies for 2677 patients) and an average of 18 biopsies (range: 9–24) were taken per colonoscopy. The average age of patients with LGD from 11 studies3, 4, 8–10, 24, 25, 27, 29, 30, 35 was 42 years (range, 17–58.5). The average duration of colitis at the diagnosis of LGD from 16 studies3, 4, 8–10, 23–30, 32, 35, 36 was 17 years (range: 11–23).

PostLGD diagnosis an average of 4.3 colonoscopies (range: 3–7.6) were performed/patient from 84, 8–10, 25–27, 30 studies (1225 colonoscopies in 288 patients) to detect seventy-three advanced lesions preoperatively during surveillance (HGD: 47, colorectal cancers: 18, DALM: 8). Thirteen of the 47 patients with pre-operative high-grade dysplasia had cancer on operative histology, three had low-grade dysplasia, three had no dysplasia, data were not available on three patients and the remaining 25 had persistent high- grade dysplasia. Overall 31 CRCs (Appendix B) were detected of which 55% were early cancers (Dukes A or B). In 98 cases, colectomy was performed within 6 months of initial diagnosis or for nonprogressive LGD. In 25 of 98 (25.5%) of cases colorectal cancer was detected on operative histology (concurrent CRC) and in addition 11 of 98 (11%) had concurrent high-grade dysplasia. Thus, overall 36 of 98 (37%) of patients with LGD on surveillance had a co-existing advanced lesion.

Incidence of colorectal cancer and advanced lesions in LGD

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

Seven of the 20 studies4, 8–10, 25, 27, 30 reported the duration of follow-up (mean duration or follow-up in patient years) postLGD (including DALM) diagnosis (involving more than 1000 patients with ulcerative colitis undergoing surveillance and 239 patients with LGD followed up for 1520.7 pyd). Figure 1 shows the Forrest plot of the cancer incidence rates from these studies. The overall cancer incidence rate from was 14 of 1000 pyd (95% CI: five of 1000 to 34 of 1000) and the overall incidence rate of any advanced lesion (HGD, CRC or DALM) was 30 of 1000 pyd (95% CI: 12 of 1000–76 of 1000) (Figure 2). Two8, 27 of the seven studies included are from the same centre and 29 patients have been double reported. However, excluding the earlier study27 from this centre does not significantly alter the incidence rate for cancer (cancer incidence rate 16 of 1000 pyd, 95% CI: six of 1000 to 42 of 1000).

image

Figure 1.  Forrest plot showing the incidence of colorectal cancer in patients with low-grade dysplasia undergoing surveillance. The numbers 0.0002, 0.002, 0.02, and 0.2 on the x-axis denotes 0.2 of 1000, 2 of 1000, 20 of 1000 and 200 of 1000, respectively.

Download figure to PowerPoint

image

Figure 2.  Forrest plot showing the incidence of any advance lesion (colorectal cancer, high-grade dysplasia or dysplasia-associated lesion or mass) in patients with LGD undergoing surveillance. (0.0006, 0.006, 0.06 on the x-axis denotes 0.6 of 1000, 6 of 1000 and 60 of 1000, respectively).

Download figure to PowerPoint

A mixed effect meta-analysis regression showed that only the number of biopsies taken per colonoscopy had a statistically significant influence on the incidence rates of advanced lesions (P = 0.002) whereas the duration of colitis prior to the diagnosis of LGD (P = 0.27), number of pathologists reviewing histology (P = 0.72) and the number of surveillance colonoscopies done per patient (P = 0.54) did not have a statistically significant influence on the cancer incidence rates. When the overall risk from studies was plotted against the midpoint of each study (Figures 5 and 6) there was a significantly increased trend in the incidence rates of cancer (P = 0.002) and advanced lesions from 1987 to the present date (P = 0.05). Cancer incidence and the incidence of advanced lesions increased by 13% and 11% per year of study, respectively.

image

Figure 5.  Meta-regression plot showing the trend of cancer incidence rates and the incidence rates of advanced lesions with time. The cancer incidence rate is plotted on the log scale against the study mid points on the x-axis.

Download figure to PowerPoint

image

Figure 6.  Meta-regression plot showing the trend of cancer incidence rates and the incidence rates of advanced lesions with time. The cancer incidence rate is plotted on the log scale against the study mid points on the x-axis.

Download figure to PowerPoint

Using overall incidence rates, Begg’s test was used to determine if the results could be explained by publication bias. Overall publication bias was present (paucity of small studies showing low cancer incidence causing a gap in the base of the graph – upper right hand corner in Figure 7) but not statistically significant (Begg’s test p = 0.24)

image

Figure 7.  Funnel plot showing no significant publication bias in surveillance studies in patients with LGD in chronic UC.

Download figure to PowerPoint

Risk of developing crc/high-grade dysplasia in LGD

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

The relative risk of colorectal cancer in patients with LGD compared to patients with no dysplasia on surveillance was determined from 78, 9, 26, 29, 33, 36, 37 of the 20 studies (Figure 3). There was a ninefold higher risk of developing CRC once LGD is diagnosed compared with patients having no dysplasia (OR: 9.0, 95% CI: 4.0–20.5). Data from eight8, 9, 26, 29, 32, 33, 36, 37 of the 20 studies were used to estimate the relative risk of developing an advanced lesion in patients with LGD (Figure 4). There was a 12-fold higher risk of developing an advanced lesion (HGD or CRC) once LGD is diagnosed compared to no dysplasia (OR: 11.9, 95% CI: 5.2–27). In terms of absolute risk reduction this equates to 12% and 15% for CRC and advanced lesions respectively. Therefore, the number needed to colonoscope (NNC) was 8 (95% CI 5–25) to detect one CRC and 6 (95% CI: 5.0–27) to detect one advanced lesion over an average duration of 5.2 years. Meta-regression analysis showed that the duration of colitis prior to the first diagnosis of LGD during surveillance had no significant influence on the cancer risk (p = 0.57).

image

Figure 3.  Forrest plot showing the odds ratio for cancer in patients with low-grade dysplasia in chronic ulcerative colitis undergoing surveillance.

Download figure to PowerPoint

image

Figure 4.  Forrest plot showing odds ratio of any advanced lesion in patients with low-grade dysplasia in chronic ulcerative colitis undergoing surveillance.

Download figure to PowerPoint

Incidence and prevalence of LGD in chronic UC

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

The overall prevalence of both flat low-grade dysplasia (fLGD) and LGD in the presence of DALM from 133, 9, 23–26, 29, 30, 32, 33, 35–37 studies was 9.4% (95% CI: 1.1–51). The overall incidence of LGD (including DALM’s) was 21 of 1000 pyd (95% CI: 15 of 1000 to 30 of 1000) from 16 studies.3, 8, 9, 23, 24, 26–30, 32–37 Of the 16 studies, four were from the UK, 5 from the USA and 6 from other European Union countries and 1 from Israel. The overall incidence rates for LGD in the UK was 15 of 1000 pyd (95% CI: eight of 1000 to 32 of 1000), in USA, 43 of 1000 pyd (95% CI: 27 of 1000 to 67 of 1000) and in other European Union countries 15 of 1000 pyd (95% CI 10 of 1000 to 21 of 1000).

Mixed effects meta-regression techniques was used to check if the number of biopsies taken per colonoscopy or the number of colonoscopies done per patient during surveillance had a significant influence on the geographical variation in incidence rates. The number of colonoscopies (P = 0.57), number of biopsies done per colonoscopy (P = 0.09) had no significant influence on the incidence rates of LGD.

LGD as a predictor of advanced lesions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

The positive predictive value (PPV) of LGD and LGD in the presence of DALM for concurrent advanced lesions and progression to advanced lesions/cancer from 21 studies are given in Table 2. Overall the positive predictive value for LGD including DALM (number of cases of advanced lesions on colectomy at initial diagnosis or for nonprogressive lesions/total number of cases who had colectomy at initial diagnosis or nonprogressive lesions) for concurrent advanced lesions was 37% (36 of 98) and the overall PPV for LGD for progression to an advanced lesion preoperatively was 18% (73 of 410). When only cancers are considered the overall PPV for concurrent cancer was 25.5% (25 of 98) and PPV for progression to cancer was 7.6% (31 of 410). When only fLGD are considered 18 of the 81 patients whom had colectomy either at initial diagnosis or for nonprogressive LGD had cancer at colectomy giving a PPV for fLGD of 22%. In 58 of the remaining 396 cases of fLGD progressed to an advanced lesion (CRC, HGD, or DALM) giving a PPV of 14.6% for LGD for progression to an advanced lesion. With respect to LGD in the presence of DALM, 7 of the 17 patients in 15 studies had an advanced lesion on colectomy either at initial diagnosis or for nonprogrerssive disease giving a PPV of 41% for LGD in the presence of DALM for concurrent advanced lesions and 9 of the remaining 14 patients progressed CRC or HGD giving a 64% PPV for LGD in the presence of DALM for progression to an advanced lesion.

Table 2.   Table showing the positive predictive value of LGD (fLGD and LGD in the presence of DALM) for concurrent cancer at colectomy and progression to cancer and advanced lesions preoperatively
 Concurrent CRC only (%)Concurrent HGD±CRC (%)Progression to CRC only (%)Progression to any advanced lesion (%)
  1. * The positive predictive value for fLGD for progression to only CRC cannot be accurately determined.

Overall25.5377.618
Only fLGD*223614.6
LGD+DALM41415064

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

This is the first meta-analysis of cancer incidence rates when LGD is found on surveillance. This also the first comprehensive systematic review in the last 10 years since Bernstein et al. reported the cancer rate in any dysplasia from 10 surveillance studies in patients with chronic UC.38 This relatively small meta-analysis, in the absence of large randomised multi-centre trials provides useful estimates of cancer risk and the risk of any advanced lesion in patients diagnosed with LGD on surveillance. The pooled estimates take into account the study heterogeneity as they are based on the random effects model. The methods used have made assumptions, and must be interpreted with caution. We accept the pitfalls of such an analysis.

The incidence rate of any advanced lesion was estimated at 30 of 1000 pyd and that of cancer at 14 of 1000 pyd from seven studies. Compared with Eaden et al. meta-analysis2 there is a fourfold increase in the cancer incidence rate and a 10-fold rise in the incidence rate of any advanced lesion once LGD is diagnosed on surveillance. The estimated cancer risk may be an underestimate for the following reasons: Firstly, the number of surveillance biopsies postLGD diagnosis is an important (especially for flat/depressed lesions) determinant of the cancer incidence rates. Rubin et al. has estimated that 33 surveillance biopsies are required to detect dysplasia with 90% confidence.39 However, questionnaire surveys in UK have shown that surveillance practice is not uniform and only a few gastroenterologists performing more than 20 biopsies during surveillance.40 Therefore, sampling bias has a significant influence on the cancer incidence rates as confirmed later in our meta-regression analysis. Secondly, the operator’s ability to detect dysplasia is another important variable. The drawbacks of conventional surveillance compared with surveillance following dye spraying and magnification has been recently reported. Rutter et al. has shown that targeted biopsies following dye spraying significantly increases the chances of detecting dysplasia when compared to conventional surveillance. In the same study dysplasia was not detected in 2900 non-targeted biopsies.41 High magnification colonoscopy in addition to dye spraying has a high predictive value in detecting both non-neoplastic and neoplastic lesions.42 Chromendoscopy detects new neoplasia missed on conventional colonoscopy in 10% of patients and has a higher chance of detecting intraepithelial neoplasia in flat lesions compared with conventional colonoscopy.43 These studies further highlight the possibility of missing LGD because of sampling bias. Thirdly, only 145 of 482 (30%) patients in our analysis (excluding 26 of 508 patients who had colectomy at initial diagnosis) had colectomy either for progression to an advanced lesion or for nonprogressive LGD. Considering the possibility of sampling bias and the drawbacks of nontargeted surveillance the number of concurrent cancers in noncolectomised patients remain uncertain. Finally, the presence of ongoing inflammation and the degree of inflammation significantly correlates with the development of dysplasia and cancer.44 The risk of CRC in UC is reduced by 80% in patients on long-term 5-aminosalycilates.2 The impact of 5-aminosalycilates on the incidence rate of any advanced lesion could not be determined in this analysis because medication status in the subgroup of patients with LGD was not clearly documented in the selected studies.

We have also shown that once LGD is diagnosed during surveillance there is a ninefold increase in the risk of developing cancer and a 12-fold increase in risk from any advanced lesion compared to patients with no dysplasia. Lynch et al. has pointed out that 75% of patients will have some form of dysplasia if surveyed long enough.27. With these figures in mind colectomy for patients with flat LGD seems a reasonable practice as in some Scandinavian centres.45 In contrast to the above argument Tine et al. in a recent retrospective study found that six of eight patients (75%) with flat LGD did not progress to an advanced lesion after a median surveillance period of 17.8 years.46 One can argue that 6 monthly intensive surveillance is the least one should offer these patients and colectomy considered for persistent nonprogressive LGD although there is no clear consensus regarding this strategy. Patients with chronic colitis found to have DALMs on surveillance should be offered a colectomy. However, it is also important to determine whether polypoidal or raised dysplastic lesions found on surveillance are sporadic adenomas or associated with colitis (DALM or ALM) since sporadic adenomas found outside an area of inflammation could safely be removed by polypectomy. A similar approach has been considered for ALM or even for genuine DALMs.25, 47 We have also shown a significant increased trend in incidence of cancer with time. Our meta-regression showed the number of surveillance biopsies had a significant influence on the incidence of any advanced lesions (P = 0.002) and highlights the importance of adhering to an intensive surveillance protocol.

As the inter-observer variation between pathologists for LGD is high one would expect this to influence the overall incidence rates of cancer in LGD as in the St Marks study.33 The problem with accurate histological interpretation of LGD is to differentiate it from regenerative epithelium in the presence of underlying inflammation.48 Repeat sampling after high doses 5-ASAs can potentially circumvent this problem to certain extent in the same way as high dose proton pump inhibitors in patients with LGD in Barrett’s epithelium. Although our meta-regression analysis showed no significant influence of this variable on the overall incidence rates of either cancer or an advanced lesion this should be interpreted with extreme caution as only 28, 29 of the seven studies included in the analysis had more than one pathologist reviewing histology. The true influence of histological interpretation can only be assessed if the cancer incidence rate in studies where two or more pathologist’s interpreted histology is compared to studies where only one pathologist was involved in histological interpretation of LGD. The numbers in this meta-analysis are small for any useful comparison.

This is the first study to estimate the number needed to colonoscope (NNC) to detect an advanced lesion or cancer (any stage) once LGD (flat or DALM) is detected on surveillance. This was estimated to be 6 and 8, respectively over a mean duration of 5.2 years. The NNC for LGD with DALM’s could not be calculated separately because of the paucity of data. We suspect this to be lower for DALM than the above estimate. These estimates are valuable in calculating the cost effectiveness and the cost benefit of cancer surveillance when LGD is detected. To date similar data are not available for useful comparison of invasive cancer surveillance programmes in Barrett’s oesophagus or nondysplastic chronic ulcerative colitis. This will contribute to a difficult but informed management strategy when LGD is found on surveillance.

We attempted to determine the incidence of LGD in chronic ulcerative colitis. The incidence data is possibly contaminated by the prevalence data and therefore should be interpreted with caution. Where possible prevalent LGD has been excluded when incident figures are calculated. We have shown that there is a geographical variation in the incidence of LGD with an incidence of 12 of 1000 pyd in Europe, of 28 of 1000 pyd in the US and 15 of 1000 pyd in UK. This geographical variation may represent a true population difference with genetic, environmental and dietary factors playing a role. This is difficult to prove without comparative trials between continents and to date there is no evidence to suggest this. The relatively low incidence of LGD in Europe may be a reflection of more aggressive medical therapy or a policy of early colectomies in some Scandinavian countries.45 An alternative explanation for the high incidence in USA may be the more intensive surveillance protocols in some American centres. American surveillance studies on an average do more biopsies per colonoscopy than either European or British studies. However, our meta-regression analysis showed that influence of biopsies per colonoscopy on LGD incidence rates were not significant.

We acknowledge that there is selection bias in our inclusion criteria in that non-English language publications have been excluded. It is also possible that some relevant studies may have eluded our search. The other obvious drawback is that unpublished data has not been included in our analysis. It is possible that many of the unpublished data are small studies showing low incidence of cancer rates.

The overall positive predictive value of LGD for concurrent cancer was 25.5% and for progression to CRC was 7.6%. This compares to Bernstein et al. results of 19% and 8%, respectively.39 However, the predictive value of DALM for any concurrent advanced lesion and progression to any advanced lesion was 41% and 64%, respectively. These figures are in keeping with other studies.29, 39 Low-grade dysplasia in the presence of DALM constitutes a high risk of CRC and is an indication for colectomy. Patients should be made aware of the risk and routinely advised to undergo colectomy if these lesions are detected. On the other hand, there is a one in four chance of any concurrent advanced lesions when only flat LGD is found on surveillance and a lesser risk of progressing to an advanced lesion. These figures allow both the gastroenterologists and patients to make informed choices regarding colectomy in the presence of such lesions.

To conclude we have shown that there is a high incidence of CRC or any advanced lesion in the presence of LGD and an increased trend of cancer incidence over time. We have also given useful figures for the number needed to colonoscope to detect CRC or any advanced lesion once LGD is diagnosed on surveillance. Low-grade dysplasia in the presence of DALM has a high positive predictive value for concurrent CRC and progression to any advanced lesion. Flat LGDs has a positive predictive value of 22% for harbouring concurrent CRC and 14.6% PPV for progression to any advanced lesion. The decision regarding continuing surveillance, colectomy, endoscopic polypectomy or endomucosal resection (for sporadic adenomas and ALMs) should be an informed one involving a multidisiplinary team. Cancer risk estimates in our meta-analysis will contribute significantly to this decision-making process.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices
  • 1
    Ekbom A, Helmick C, Zack M, et al. Ulcerative colitis and colorectal cancer: a population based study. N Eng J Med 1990; 323: 122833.
  • 2
    Eaden JA, Abrams KA, Mayberry JF. Cancer risk in ulcerative colitis: A meta-analysis and systematic review. Gut 2001; 48(4): 52635.
  • 3
    Leidenius M, Kellokumpun I, Husa A, et al. Dysplasia and carcinoma in longstanding ulcerative colitis: an endoscopic and histological surveillance programme. Gut 1991; 32: 152125.
  • 4
    Ullman TA, Loftus VE Jr, Kakar S, et al. The fate of low-grade dysplasia in ulcerative colitis. Am J Gastrol 2002; 97(4): 9227.
    Direct Link:
  • 5
    Eaden AJ, Ward AB, Mayberry JF. How gastroenterologists screen for colonic cancer in ulcerative colitis: an analysis of performance. Gastrointestinal Endosc 2000; 51(2): 12328.
  • 6
    Bernstein CN, Weinstein MW, Levine SD, et al. Physicians’ perceptions of dysplasia and approaches to surveillance colonoscopy in ulcerative colitis. Am J Gastroenterol 1995; 90(12): 210614.
  • 7
    Melville DM, Jass JR, Morson BC, et al. Observer study of the grading of dysplasia in ulcerative colitis: comparison with clinical outcome. Hum Pathol 1989; 20: 100814.
  • 8
    Lim CH, Dixon MF, Vail A, Forman D., et al. Ten year follow up of ulcerative colitis patients with and without low-grade dysplasia. Gut 2003; 52: 112732.
  • 9
    Rozen P, Baratz M, Fefer F, et al. Low incidence of significant dysplasia in a successful endoscopic surveillance program of patients with ulcerative colitis. Gastroenterology 1995; 108: 136170.
  • 10
    Ullman TA, Croog V, Harpaz N, et al. Progression of flat low-grade dysplasia to advanced neoplasia in patients with ulcerative colitis. Gastroenterology 2003; 125: 131119.
  • 11
    Eaden AJ, Mayberry JF. Guidelines for screening and surveillance of asymptomatic colorectal cancer in patients with inflammatory bowel disease. Gut 2002; 51(Suppl. 5): V102.
  • 12
    Winawer S, Fletcher R, Bond J, et al. Colorectal cancer screening and surveillance: clinical guidelines and rational update based on new evidence. Gastroenterology 2003; 124: 54460.
  • 13
    Lennard- Jones JE. Screening for colorectal cancer in ulcerative colitis. Gut 1990; 32: 72233.
  • 14
    The University of York NHS centre for reviews and Dissemination. Undertaking systematic reviews of research on effectiveness: CRD guidelines for those carrying out our commissioning reviews. New York: York Publishing Services Ltd, 1996.
  • 15
    Riddell RH, Goldman H, Ransohoff DF, et al. Dysplasia in inflammatory bowel disease: standardised classification with provisional clinical applications. Hum Pathol 1983; 14: 93168.
  • 16
    Stata Statistical Software (computer programme). Stata Corp. College Station. Texas: Stata Corporation 1997.
  • 17
    Sharp SJ, Sterne JAC. Meta analysis. Stata Tech Bull 1997; 38: 914.
  • 18
    Thompson S, Sharp S. Explaining homogeneity in Meta-analysis. Stat Med 1999; 18: 2693708.
  • 19
    DerSimion, Laird L. Random effects regression model for meta-analysis. Stat Med 1995; 14: 395411.
  • 20
    Sharp SJ. Meta analysis regression. Stata Tech Bull 1998; 42: 1622.
  • 21
    Dear K, Begg C. An approach to assessing publication bias prior to performing a meta-analysis. Stat Sci 1992; 7: 23745.
  • 22
    Egger M, Smith JD, Schneider M, et al. Bias in meta-analysis detected by a simple graphical test. BMJ 1997; 315: 62934.
  • 23
    Rosenstock E, Farmer GR, Petras R, et al. Surveillance for colonic carcinoma in ulcerative colitis. Gastroenterology 1989; 89: 134246.
  • 24
    Lindberg B, Persson B, Veress B, et al. Twenty years’ colonoscopic surveillance in patients with ulcerative colitis. Scand J Gastroenterol 1996; 31: 1195204.
  • 25
    Befrits R, Ljung T, Jaramillo E, et al. Low grade dysplasia in extensive longstanding inflammatory bowel disease. Dis Colon Rectum 2002; 45: 61520.
  • 26
    Woolrich JA, DaSilva DM, Korelitz IB. surveillance in the routine management of ulcerative colitis: the predictive value of low-grade dysplasia. Gastroenterology 1992; 103: 4318.
  • 27
    Lynch DA, Lobo AJ, Sobala MG, et al. Failure of colonoscopic surveillance in ulcerative colitis. Gut 1993; 34: 107580.
  • 28
    Blackstone MO, Riddell HR, Rogers Gerald BH, et al. Dysplasia–associated lesion or mass (DALM) detected by colonoscopy in longstanding ulcerative colitis: An indication for colectomy. Gastroenterology 1981; 80: 36674.
  • 29
    Biasco G, Brandi G, Paganelli MG, et al. Colorectal cancer in patients with ulcerative colitis. A prospective cohort study in Italy. Cancer 1995; 75(8): 204550.
  • 30
    Lennard-Jones JE, Melville DM, Morson BC, et al. Precancer and cancer in extensive ulcerative colitis: findings among 401 patients over 22 years. Gut 1990; 31: 80006.
  • 31
    Hata K, Watanabe T, Kazama S, et al. Early surveillance colonoscopy programme improves survival in patients with ulcerative colitis associated colorectal cancer: results of a 23 year surveillance programme in the Japanese population. Br J Cancer 2003; 89: 123236.
  • 32
    Brostrom O, Lofberg R, Ost A, et al. Cnacer surveillance pf patients with longstanding ulcerative colitis: a clinical, endoscopical, and histological study. Gut 1986; 27: 140813.
  • 33
    Connell WR, Lennard-Jones JE, Williams BC, et al. Factors affecting the outcome of endoscopic surveillance for cancer in ulcerative colitis. Gastroenterology 1994; 107: 93444.
  • 34
    Jonsson B, Ashgren L, Andersson LO, et al. Colorectal cancer surveillance in patients with ulcerative colitis. Br J Sur 1994; 81: 68991.
  • 35
    Lofberg R, Brostrom O, Karlen P, et al. Colonoscopic surveillance in longstanding ulcerative colitis. Gastroenterology 1990; 99: 102131.
  • 36
    Nugent WF, Haggitt RC, Gilpin AP. Cancer surveillance in ulcerative colitis. Gastroenterology 1991; 100: 124148.
  • 37
    Lashner BA, Kane SV, Hanauer SB, et al. Hazard rates for dysplasia and cancer in chronic ulcerative colitis. Dig Dis Sci 1989; 34: 153641.
  • 38
    Bernstein NC, Shanahan F, Weinstein MW. Are we telling the truth about surveillance colonoscopy in ulcerative colitis Lancet 1994; 343: 7174.
  • 39
    Rubin PH, Friedman S, Harpaz N, et al. Colonoscopic polypectomy in chronic colitis: conservative management after endoscopic resection of dysplastic polyps. Gastroenterology 1999; 117: 1295300.
  • 40
    Eaden JA, Ward BA, Mayberry JF. How gastroenterologists screen for colonic cancer in ulcerative colitis: an analysis of performance. Gastrointest Endosc. 2000; 51(2): 12328.
  • 41
    Rutter MD, Saunders DP, Schofield G, et al. Pancolonic indigo carmine dye spraying for the detection of dysplasia in ulcerative colitis. Gut 2004; 53: 25660.
  • 42
    Hurlstone DP, Sanders DS, Lobo AJ, McAlindon ME, Cross SS. Indigo carmine-assisted high-magnification chromoscopic colonoscopy for the detection and characterisation of intraepithelial neoplasia in ulcerative colitis: a prospective evaluation. Endoscopy 2005; 37(12): 118692.
  • 43
    Trecca A, Gaj F, Pascuito A, Serafinin S, Bella A. Advanced diagnostic tools: virtual colonoscopy and magnifying endoscopy. Tech Coloprocol 2004; Suppl. 2: s27982.
  • 44
    Rutter M, Saunders B, Wilkinson K, et al. Severity of inflammation is a risk factor for colorectal neoplasia in ulcerative colitis. Gastroenterology 2004; 126: 45159.
  • 45
    Langholz E, Munkholm P, Davidsen M, et al. Colorectal cancer risk and mortality in patients with ulcerative colitis. Gastroenterology 1992; 103: 144451.
  • 46
    Jess T, Edward VL, Velayos FS, et al. Incidence and prognosis of colorectal dysplasia in inflammatory bowel disease: A population-based study from Olmstead County, Minnesota. Inflammatory Bowel Diseases 2006; 12(8): 66976.
  • 47
    Engelsgjerd M, Farraye FA, Odze RD. Polypectomy may be adequate treatment for adenoma-like dysplastic lesions in chronic ulcerative colitis. Gastroenterology 1999; 117: 128894.
  • 48
    Barbara MJ, Appelman HD. Dysplasia and the gut: the diagnosis is harder than it seems. J Clin Gastroenterol 2002; 43(2): 11116.

Appendices

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims and Methods
  5. Exclusion criteria
  6. Definitions
  7. Results
  8. Incidence of colorectal cancer and advanced lesions in LGD
  9. Risk of developing crc/high-grade dysplasia in LGD
  10. Incidence and prevalence of LGD in chronic UC
  11. LGD as a predictor of advanced lesions
  12. Discussion
  13. Acknowledgements
  14. References
  15. Appendices

Appendix A

Type of Study

Country where study done

Total number of patients with UC undergoing surveillance

Duration of follow-up in patients with ulcerative colitis in patient years duration

Total number of patients with low-grade dysplasia

Duration of follow-up in patients with LGD in pyd

Mean number of surveillance colonoscopies

Mean number of biopsies per colonoscopy

Duration of colitis when LGD detected

Mean age of patients when LGD detected

Number of pathologists reviewing histology

Number of patients with cancer/HGD on immediate colectomy (within 6 months of diagnosis of LGD)

Number of patients with cancer/HGD among patients had colectomy for persistent LGD

Number of patients progressing to CRC pre-operatively

Number of patients progressing to HGD/DALM pre-operatively

Operative histology

no dysplasia/LGD/DALM/HGD/Cancer

Stage of CRC

Dukes A/Dukes B/Dukes C/Dukes D

Appendix B

Flowchart showing the operative histology of patients with LGD colectomised for progression to any advanced lesion in chronic ulcerative colitis

(ND, no dysplasia; LGD, low-grade dysplasia; HGD, high-grade dysplasia; CRC, colorectal cancer)

inline image