Abstract
Aim: This study used a multicenter questionnaire survey to evaluate the morphology and progression of the initial lesion in cases of colitis-associated colorectal neoplasia (CRN). Patients and Methods: Endoscopic images of lesions that had been definitively diagnosed as CRN by pathological examination were retrospectively reviewed. Results: This resulted in the identification of 54 initial lesions in 49 patients. The 54 initial lesions fell into the following categories: 22 endoscopically visible localized lesions consisting of 18 elevated lesions and 4 depressed lesions, as well as 32 lesions that were not endoscopically visible as localized and consisted of 20 active-phase mucosal lesions and 12 remission-phase mucosal lesions. Nineteen of the lesions eventually became advanced cancers, while 35 lesions eventually became early-stage cancers. The final lesions were 40 elevated lesions, 5 flat or depressed lesions and 9 stenotic lesions. The form of growth of the advanced cancers was progressive stenosis or increased elevation. For approximately 69% of the early-stage cancers, the growth form was increasing elevation or development of elevation. For 73.6% of the advanced cancers, the initial lesion underwent rapid growth and became advanced cancer within 3 years; they accounted for 25.9% of the total cancers. Approximately 40% of the initial lesions of CRN were endoscopically visible as localized lesions, while approximately 60% were judged to be inflammatory mucosal lesions. Conclusion: It will be necessary to proactively take biopsy inflammatory mucosal lesions in order to discover tumors early and periodic surveillance should be performed with the knowledge that tumors may grow very quickly.
Much remains unclear regarding the process of colon cancer development in ulcerative colitis (UC). Even endoscopic image findings have been insufficient for accurate diagnosis of early-stage lesions. Blackstone et al. (1) proposed that endoscopic images of a sessile, elevated lesion be called a dysplasia-associated lesion or mass. More recently, Matsumoto et al. (2) proposed 6 endoscopic morphological classifications for colorectal neoplasia (CRN); however, there has been no international consensus.
CRN is known to be an important complication of UC for which appropriate preventive endoscopy is considered necessary. The objectives of this study were to analyze and identify findings that would contribute to early diagnosis and shed light on the growth rate or, so-called, “lead time” (3, 4) of CRN. The ultimate goal was to generate data that would aid in the creation of a guideline for early diagnosis of CRN. Many institutions in Japan carefully archive endoscopic records and images. We reasoned that retrospective analysis of such images might yield findings useful for identifying the initial lesions in the developmental process leading to CRN. Such findings could then be incorporated into future surveillance colonoscopy (SC).
Questionnaire contents.
The approach was to conduct a nationwide, multicenter, questionnaire survey to collect information on CRN cases. The data were then analyzed for cases for which early-stage lesions of CRN could be identified by retrospective review of the endoscopic images.
It has been said overseas that most cases of CRN are interval cancers that showed no abnormality on prior examinations (3, 4). In addition, there is no consensus regarding the appropriate frequency for examinations. For that reason, if prior examinations included images of sites where lesions developed, such cases were included in the analysis even if there was no judgment of a localized lesion.
Retrospective view of progression of initial lesions to final lesions. There were 22 retrospectively endoscopically visible localized lesions. Thirty-two lesions were not endoscopically visible as localized lesions on retrospective review of the endoscopic photographs. Nineteen final lesions were advanced cancer, 8 of which had progressed from visible lesions and 11 that had progressed from invisible lesions. The remaining 35 final lesions were early-stage cancers or dysplasia, comprising of 14 that had progressed from visible lesions and 21 that had progressed from invisible lesions.
This study was reviewed and approved by the Institutional Review Board of Fukuoka University Chikushi Hospital.
Patients and Methods
A multicenter questionnaire survey targeting institutions was conducted. Instructions were given to describe pertinent examples in the protocol and return the form to our department.
The subjects were cases: (i) with confirmed diagnosis of UC; (ii) having lesions that were pathologically diagnosed as CRN based on endoscopic or surgical excision findings as of October 31, 2013; or (iii) having endoscopic images taken before CRN was detected that show the site of the eventual CRN lesion. Endoscopic images were reviewed by the original institution and central reading members (K.Y, T.M, T.H). In principle, cases were excluded if more than 4 years had passed since the previous endoscopic examination. However, cases were included if multiple examinations had been performed during that interval. This provision is due to using 36 months as the basis for calculating having missed/overlooked cancers on colonoscopy (3). Table I shows the contents of the questionnaire used.
Results
Courses of all cases. Responses to the questionnaire were received from 15 institutions, yielding 54 lesions in 49 patients. The clinical background data showed more males than females, while the mean patient age at the time of diagnosis of CRN was 50.5 years and the mean disease duration until diagnosis of CRN was 15.9 years. Most were of the total colitis type and the clinical course showed nearly equal numbers of the chronic persistent type and the relapse-remission type. The location of about 80% of the lesions was the sigmoid colon and rectum. Endoscopic examination was performed because of the presence of symptoms in 16 patients; however, 33 patients had been asymptomatic and many of them had been diagnosed as a result of SC (Table II).
Clinical characteristics of the 49 patients (54 lesions).
Figure 1 shows the relationships between the retrospectively determined initial lesions and the final lesions. The 54 initial lesions were classified into endoscopically visible localized lesions and lesions that were not endoscopically visible as localized. The latter were further classified into active-phase UC mucosal lesions and remission-phase UC mucosal lesions. There were 22 retrospectively endoscopically visible localized lesions, consisting of 12 IIa, 2 IIa+IIc, 3 Is, one Is+IIb, 3 IIc, and one IIc+IIb.
Morphology of retrospectively examined initial lesions for the 19 final lesions that were advanced cancers. The initial lesions for the 19 final lesions that had become advanced cancers and were retrospectively reviewed consisted of 8 lesions that were visible and 11 invisible lesions. The final images of the 8 visible lesions showed stenosis for 3 lesions and increased elevation for 5 lesions. The final images of the 11 invisible lesions showed 4 lesions with stenosis and 7 lesions with increased elevation. For all 19 final lesions, the tumor had increased in diameter.
Macroscopic type of advanced cancer final lesions (n=19).
Thirty-two lesions were not endoscopically visible as localized lesions on retrospective review of the endoscopic photographs. They consisted of 20 active-phase UC mucosal lesions and 12 remission-phase UC mucosal lesions (Figure 1).
Nineteen final lesions were advanced cancer, 8 of which had progressed from visible lesions and 11 that had progressed from invisible lesions. The remaining 35 final lesions were early-stage cancers or dysplasia, comprising of 14 that had progressed from visible lesions and 21 that had progressed from invisible lesions (Figure 1).
Courses of the 19 final lesions that were advanced cancers (Table III). The macroscopic types of 19 advanced cancer final lesions were: one Is-like advanced; one IIa+IIb-like advanced; 2 IIc+IIb-like advanced; 2 IIa-like advanced; one IIa+IIc-like advanced; one Is+IIb-like advanced; one IIa+Is+IIc-like advanced; one type 2; 3 type 2; one type 3; one type 4; and 3 type 5. The mean observation period for the 19 lesions was 32.8±26.1 months. Eight of the final lesions showed stenosis; the tumor diameter increased for all lesions. The mean observation period was 33.1±27.7 months for 8 lesions for which the initial lesion was visible and 32.6±26.1 months for 11 lesions for which the initial lesion was invisible (Table III). Some of the patients with a long observation period were examined on multiple occasions. Moreover, review of the cancer growth rate showed that the initial lesion underwent rapid growth and became advanced cancer within 3 years for 73.6% of the advanced cancers; they accounted for 25.9% of the total cancers.
A visible lesion that progressed to advanced cancer after 19 months. Left picture: IIa (visible). Right picture: advanced cancer (IIa-like advanced, increased elevation). A rectal IIa lesion was diagnosed as group 3 on biopsy. After 19 months, it had grown into a IIa-like advanced cancer. This is a case of rapid growth into an advanced cancer within 2 years.
An invisible lesion that developed into advanced cancer after 14 months. Left picture: Active-phase UC mucosa (invisible). Right picture: advanced cancer (IIa+IIc-like advanced). Initially, the mucosa showed low-active phase UC but it was not elevated. Later, the same site showed localized redness and was slightly elevated. This is a case of rapid growth.
Morphology of retrospectively examined initial lesions for the 19 final lesions that were advanced cancers (Figure 2). The initial lesions for the 19 final lesions that had become advanced cancers and were retrospectively reviewed consisted of 8 lesions that were visible (4 IIa (Figure 3); 3 IIc; one IIa+IIc) and 11 invisible lesions (5 active-stage UC (Figures 4, 5 and 6) and 6 remission-phase UC). The final images of the 8 visible lesions showed stenosis for 3 lesions (2 IIc+IIb-like advanced; one type 2) and increased elevation for 5 lesions (2 IIa-like advanced (Figure 3); one Is+IIa-like advanced; 2 type 1). The final images of the 11 invisible lesions showed 4 lesions with stenosis (one IIa+IIc-like advanced (Figure 4); one type 3 (Figure 5); one type 4; one type 5) and 7 lesions with increased elevation (one Is-like advanced; one IIa+Is+IIc-like advanced (Figure 6); one IIa+IIb-like advanced; 2 type 2; 2 type 5). For all 19 final lesions, the tumor had increased in diameter (Figure 2).
Morphology of the 35 final lesions that were early-stage cancers or dysplasia (Table IV). The macroscopic types of 35 final early-stage-cancer or dysplasia lesions were: 7 IIb; 12 IIa; 6 Is+IIa; 4 Is; 2 IIa+IIc; 2 IIa+IIb; and 2 Is+IIb. The mean observation period for the 35 lesions was 38.6±21.2 months. The change in morphology consisted of formation of elevation or increased elevation in 24 lesions and stenosis in 2 lesions. The mean observation period was 32.2±17.5 months for 14 lesions for which the initial lesion was visible and 41.8±23.2 months for 21 lesions for which the initial lesion was invisible. Two lesions underwent stenosis (Table IV). Some of the patients with a long observation period were examined on multiple occasions.
An invisible lesion that developed into advanced cancer after 74 months. Left picture: Active-phase UC mucosa (invisible). Right picture: advanced cancer (stenosis, type 2). The initial site had been of the chronic persistent type. The initial lesion was also uneven and rugged but it was not localized; therefore, biopsy was not performed. Multiple subsequent examinations showed progressive stenosis and the lesion developed into a type 2 advanced cancer.
An invisible lesion that developed into advanced cancer after 13 months. Left picture: Active-phase UC mucosa (invisible). Right picture: advanced cancer (IIa+Is+IIc-like advanced). The initial lesion was diagnosed as active-phase mucosal UC. It later became elevated and underwent rapid growth.
Retrospectively observed morphology of initial lesions of the 35 final lesions that had become early-stage cancers or dysplasia (Figure 7). The initial lesions for the 35 final lesions that had become early-stage cancers or dysplasia and were retrospectively reviewed consisted of 14 lesions that were visible and 21 invisible lesions. The final images of the 14 visible lesions showed increased elevation for 10 lesions (2 IIa+IIc; 4 IIa; 2 Is; 2 Is+IIa) but no change in the elevation for 4 lesions (2 IIa (Figure 8); one Is+IIb; one Is).
The final images of the 21 invisible lesions showed 3 invisible lesions (all IIb (Figure 9)), 2 lesions with stenosis (both IIb), 14 lesions showing formation of elevation (6 IIa; one Is+IIb; 4 Is+IIa; one Is; 2 IIa+IIb) and 2 lesions with localized redness (IIb) (Figure 7).
Retrospectively observed morphology of the initial lesions of the 35 final lesions that had become early-stage cancers or dyspl asia. The initial lesions for the 35 final lesions that had become early-stage cancers or dysplasia and were retrospectively reviewed consisted of 14 lesions that were visible and 21 invisible lesions. The final images of the 14 visible lesions showed increased elevation for 10 lesions but no change in the elevation for 4 lesions. The final images of the 21 invisible lesions showed 3 invisible lesions and 14 lesions showing formation of elevation.
Macroscopic type of early-stage cancer or dysplasia final lesions (n=35).
Discussion
Several reports based on experience with many patients with CRN have been published in the United States and Europe (5-9). Recent reports document a downward trend in the incidence of CRN over time (10-14). Few institutions in Japan have had much experience with CRN (15-17). Hardly any comparisons have been conducted of resected specimens and endoscopic images and there have been almost no retrospective studies of endoscopic images for CRN lesions, such as the current study. Thus, little is known regarding the natural history of CRN. Accordingly, we reasoned that a study, such as the present one, was needed to enable early diagnosis of CRN, which is expected to increase markedly in the coming years.
However, there was a problem encountered during this study that involved the method for recording the macroscopic type of the lesions. We have almost no criteria for classification of the macroscopic type of CRN (1) and, although Matsumoto et al. (2) attempted to draw up endoscopic classifications, both overseas and in Japan there seems to be no appropriate method for classification of the macroscopic type of CRN lesions. In the present study, we relied primarily on the descriptions of the macroscopic type that were reported by the institutions. However, some aspects of changes in endoscopic images cannot be adequately expressed using the standard descriptions devised for advanced cancers in Japan by the Japanese society for cancer of the colon and rectum. Therefore, for this study, the macroscopic types for early cancers were used. We must hope that the future will see establishment of standardized descriptions and classifications for the macroscopic type of CRN lesions in Japan.
Based on the present findings, we estimated the morphology of the initial lesions leading to development of CRN and discussed their progression and morphological changes.
The 54 initial lesions that were reviewed fell into the following categories: 22 endoscopically visible localized lesions (41%) and 32 lesions (59%) that were not endoscopically visible as localized. At 30-40 months prior to diagnosis of CRN, nearly 60% of the patients could not even be suspected of being at risk of CRN. We, thus, can understand that random biopsies are needed to detect these cancers at an early stage. In fact, few of the patients included in this study underwent biopsy at the time of the initial examination. Thus, it was not possible to investigate the efficacy of such biopsy as further research into this is necessary.
A visible lesion that remained an early-stage cancer after 36 months. Left picture: IIa (visible). Right picture: early-stage cancer (IIa, elevation unchanged). The central portion was discolored and elevated. After 36 months, it was detected as a broad and low IIa lesion.
An invisible lesion that remained an early-stage cancer after 56 months. Left picture: Active-phase UC mucosa (invisible). Right picture: early-stage cancer (IIb, invisible). An extremely well-differentiated adenocarcinoma and dysplasia were also present. The final image also shows redness but its scope could not be accurately discerned.
The review of the morphological changes that occurred in initial lesions found that a majority of the sites that eventually developed into early-stage cancers showed formation of elevation or increased elevation (69%). Many of the sites that eventually became advanced cancers showed progression of stenosis (37%) or increased elevation (100%). The analysis of the rate of growth of CRN showed that the initial lesion underwent rapid growth and became advanced cancer within 3 years in 73.6% of the advanced cancers. Such rapid-growth lesions accounted for 25.9% of the total lesions. Thus, cancer growth had been rapid in quite a few patients.
From a retrospective point of view development of CRN has been studied, as in the present study, as cancers that were missed in a previous endoscopic examination. For example, Wang et al. (18) calculated the rate of early/missed colorectal cancers detected by SC of older inflammatory bowel disease (IBD) patients. When SC was performed within 36 months, they estimated rates of 15.8% for 544 patients with UC. These rates were nearly 3 times higher than the 5.8% rate in non-IBD patients. Wang et al. concluded that it is easy to miss early-stage lesions in older IBD patients, thus requiring great caution and performance of SC.
It is, thus, clear that colonoscopy has a high rate of missing tumors associated with IBD. There have been a number of reports regarding the accuracy of SC for UC patients. In Japan, Matsuoka et al. (19) retrospectively analyzed the clinicopathological features of CRN discovered by SC of patients who eventually underwent surgery. CRN was detected in 83 of 1,274 patients who underwent surgery for UC. Forty of those patients had undergone SC and 30 of them were stage 0 or 1. On the other hand, 22 CRN patients who did not undergo SC had early-stage cancer. The difference between these groups was significant (p=0.04). They, thus, concluded that lesions can be detected at an earlier stage if SC is performed.
The cohort in this study included many patients who had undergone SC and, for that reason, a majority of the detected CRN lesions were still in an early stage, i.e. 35 lesions.
Rubin et al. (20) investigated the sensitivity of SC. That is, SC was performed 1,339 times for 622 patients with UC and, during that time, 75 lesions of cancer and dysplasia were detected in 46 patients. The sensitivity of 128 examinations was investigated in a follow-up study. The sensitivity of the initial endoscopy and the sensitivity of the final endoscopy were examined. The results showed that 58.5% of dysplasia lesions were visible, while 80% of cancers were visible. As a result, the sensitivity for the lesions as a whole was 61.3%. The sensitivity for the patients as a whole was calculated as 76.1%. The follow-up study also examined the usefulness of SC. That is, in the first endoscopy, 36 lesions were recognized and diagnosed in 15 patients. However, in the follow-up study, 39 lesions were recognized in 31 patients. In this study, the rate of missed detection per lesion was 38.7%. That incidence of lesions missed in the immediately previous examination is similar to what was found in the present study.
Rutter et al. (3, 4) analyzed the precision of SC in around 600 patients. The cumulative colorectal cancer (CRC) incidence was 2.5% for 20 years, 7.6% for 30 years and 10.8% for 40 years. Overall, 5% of 600 patients developed cancer (3). That group showed that long-term SC detects cancer in 5% of patients and that their prognosis is not poor; thus, they concluded that, although not fail-proof, their SC regime is effective. The mean number of biopsies per site was 8. It is noteworthy that interval cancers that were missed by the previous examination accounted for 50% of the detected cancers. Moreover, 13 patients had advanced cancer and it was noted that quite a few showed rapid growth after having been missed. Thus, although SC is effective for UC colorectal cancers, the cancer detection rate is low and cancers develop over time. Based on the above, the sensitivity of SC was 77.3% for lesions and 89.3% for the patients as a whole. Stated from the reverse perspective, the miss-rate for lesions was 23%. Thus, few studies have evaluated SC over time. It is said that, since evaluation of SC involves a lead-time bias, careful investigations must be carried out in retrospective studies (3, 4). As noted above, looking at the diagnostic performance of SC overseas, malignant lesions are missed at a rate of about 20-35% per SC.
In the present study, it was concluded that 22 initial lesions that later developed into CRN had been endoscopically visible as tumors. They consisted of 18 elevated lesions and 4 depressed lesions. This finding suggests that the efficiency of diagnosis of early-stage lesions can be improved by focusing on elevated lesions.
Conclusion
Approximately 40% of early-stage images of UC-associated CRN were visible on endoscopic examinations performed about 3 years earlier and approximately 60% of them were judged to be inflammatory mucosal lesions. It will be necessary to proactively biopsy inflammatory mucosal lesions in order to discover tumors early, whereas periodic surveillance should be performed with the knowledge that tumors may grow very quickly.
Most of the institutions that participated in the present study had not performed dye spraying during initial endoscopy. Numerous studies have pointed out that chromocolonoscopy improves the detection rate of CRN (17, 22-24). In the future, dye spraying should definitely be included in SC protocols.
This study had a number of limitations, including the following: Biopsies had not been performed at the time of diagnosis of the initial lesions. This meant that diagnosis of lesions had been uncertain. At the time of diagnosis, dye spraying had not been performed during endoscopy. The macroscopic type and scope of lesions could not be clearly discerned. In addition, the pathological diagnoses of the final lesion did not use the institutions' diagnostic descriptions as a reference and follow-up studies were not performed. For these reasons, it was not possible to fully analyze the infiltration patterns, tissue structure of the lesions, etc. Despite these limitations, we are confident that the present findings and conclusions are both valid and valuable.
Acknowledgements
The Authors are very thankful to both editors of “Anticancer Research” and “Stomach and Intestine”, which is a Japanese Journal, and admitted to publish the present paper as secondary publication from a Japanese article (Matsui T, Yamasaki K, et al: Stomach and Intestine 49(10): 1517-1534, 2015).
- Received April 5, 2016.
- Revision received May 20, 2016.
- Accepted May 24, 2016.
- Copyright© 2016 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
