Introduction
Ulcerative colitis (UC) is a chronic inflammatory disease of the colon that causes clinical symptoms such as bloody stools, diarrhea, and abdominal pain [1 ]
[2 ]. Currently, one of the most important therapeutic targets for UC is endoscopic remission [3 ]
[4 ], but the clinical symptoms do not always reflect the endoscopic findings [5 ]. Some patients in clinical remission still have colonic mucosal inflammation [6 ]. Several studies demonstrated that endoscopic remission is associated with better long-term prognosis, namely a reduction in disease relapse, hospitalization, and surgery [7 ]
[8 ]
[9 ]
[10 ]. Histological healing has drawn more attention for achieving better long-term outcomes beyond endoscopic remission [11 ]
[12 ]
[13 ]
[14 ]. Previous studies have demonstrated that histological indices can assess active colonic inflammation more precisely than endoscopic indices even in patients with clinically mild UC [15 ]
[16 ]. These findings suggest that endoscopic evaluation is not enough to assess microscopic inflammation, which is detected as endoscopically invisible lesions, and histological healing can lead to better clinical outcomes. The recent Selecting Therapeutic Targets in Inflammatory Bowel Disease (STRIDE II) program [17 ] stated that a therapeutic target should be set to improve long-term outcomes in inflammatory bowel disease treatment. In STRIDE II, histological healing was proposed as a considerable target. However, there are some concerns about assessing histological healing. Biopsy can result in complications such as bleeding. In addition, histopathological evaluation is a burden for endoscopists and pathologists.
The LASEREO endoscopic system (Fujifilm Medical Co., Tokyo, Japan) uses a semi-conductor laser as the light source and has linked color imaging (LCI) [18 ]. LCI is an image-enhanced technology that can enhance slight color differences in the red region of the mucosa. Because it can emphasize redness, LCI is expected to effectively diagnose disease conditions that cause mucosal redness, such as inflammation.
Some reports have evaluated endoscopic inflammation of UC using LCI. Uchiyama et al. showed that LCI can subdivide the Mayo endoscopic subscore (MES), a widely used endoscopic score, with WLI [19 ]. They evaluated 193 areas in 52 patients with UC by LCI. LCI patterns were classified from A to C (A, no redness; B, redness with visible vessels; and C, redness without visible vessels). In the results, even MES 0 was sub-classified by LCI, and LCI classification was well correlated with the histopathological Matts score. Takagi et al. examined 26 UC patients with MES 0 and demonstrated that the non-relapse rate was significantly higher in patients with no redness by LCI observation [20 ].
However, a large multicenter study is still needed to confirm the usefulness of LCI in clinical practice for UC. In addition, the correlation of LCI findings with histopathology has not yet been fully investigated. Here, we conducted a multicenter study to assess LCI and histological findings to establish a new LCI endoscopic evaluation index for UC.
Patients and methods
Patient enrollment
This study (score evaluation for observation of patients with UC using LCI: The SOUL study) was conducted at three tertiary institutions (Kyorin University Hospital, Kyoto Prefectural University of Medicine, and Fukuoka University Chikushi Hospital) between November 2016 and February 2019 (UMIN000030412). Patients who were in clinical and endoscopic remission were enrolled in this study. UC diagnosis was determined based on clinical, endoscopic, radiological, and histological criteria according to the clinical practice guidelines for inflammatory bowel disease by the Japanese Society of Gastroenterology [21 ]. Patient characteristics, treatments, and blood examination results were obtained from the medical records at each institution.
Clinical disease activity
Clinical disease activity was evaluated using the partial Mayo score, which was previously used in several clinical trials [22 ]
[23 ]
[24 ]. Clinical remission was defined as a rectal bleeding subscore of 0 (no rectal bleeding) and a stool frequency subscore of either 0 (normal stool frequency for the patient) or 1 (1 or 2 more daily stools than normal).
Endoscopic evaluation
Endoscopic activity with WLI was assessed using MES [25 ]. Endoscopic remission was defined as MES ≤ 1. All examinations were performed with a high-definition EG-L590WR endoscope as part of the LASEREO endoscopic system (Fujifilm Medical Co., Tokyo, Japan). Colonoscopy was performed on the enrolled patients, and the rectum and sigmoid colon were observed.
First, observation was performed with WLI, and MES was evaluated. Patients with MES ≥ 2 were excluded. Next, observation was performed with LCI. Referring to the previous LCI classification based on the report by Uchiyama et al. [19 ], we evaluated the degree of redness. We also evaluated the area of inflammation and lymphoid follicles, which have not been analyzed as an item by conventional endoscopic observation with WLI. Briefly, endoscopic findings with LCI were classified as LCI index-R (redness), LCI index-A (area of inflammation), and LCI index-L (lymphoid follicular) ([Fig. 1 ]). In addition, each LCI index was classified as Grade 0–2 or 0–3. LCI index-R was defined as: Grade 0, no redness; Grade 1, redness with visible vessels; and Grade 2, redness without visible vessels. LCI index-A was defined as: Grade 0, up to 1 of 4 screen; Grade 1, up to 2 of 4 screen; Grade 2, up to 3 of 4 screen; and Grade 3, over Grade 2. LCI index-L was defined as: Grade 0, no lymphoid follicular; Grade 1, slightly visible lymphoid follicles; Grade 2, clearly visible lymphoid follicles; and Grade 3, lymphatic follicles with angiogenesis. These evaluations were made by three central endoscopic judgment committee members who are experts in colonoscopy for UC. Discussions were held among the three members, and when two or more of them had the same judgment this was adopted as the final evaluation.
Fig. 1 Definition and representative images of each component of the LCI index. The LCI index consist of three components: LCI index-R (Redness), LCI index-A (Area of inflammation), and LCI index L (Lymphoid follicular). LCI index-R was classified into Grade 0 to 2 according to the degree of redness with/without visible vessels. LCI index-A was classified into Grade 0 to4 according to the extent of redness on the screen. LCI index-L was categorized into Grade 0 to 3 according to the visibility of lymphoid follicles. LCI; linked color imaging.
Histopathological evaluation
Biopsy specimens were obtained from areas with inflammation and lymphoid follicles observed by LCI. Biopsy of areas with inflammation was performed at the sites at which the endoscopist judged that the inflammation was most severe and mild. One or more samples were collected from each site. When inflammation was not endoscopically observed, a biopsy of the normal mucosa was performed. The histopathological findings were evaluated by Geboes histopathological score (GHS) [26 ]
[27 ]. Histological healing was defined as GHS < 2B.1 (neutrophil infiltration in < 5 % of crypts, no crypt destruction, and no erosions, ulcerations, or granulation tissue) with reference to definition of histologic improvement in the phase 3 clinical trial of ustekinumab for moderate to severe UC [28 ].
Blinded histopathological evaluations were performed two central expert pathologists, and the final results were decided through a discussion and consensus by the two pathologists.
Ethics statement
This study was conducted in accordance with the guidelines of the Declaration of Helsinki and with the approval of each institution’s ethics committee (as core facilities, Kyorin University School of Medicine, approval number 707). All patients provided written informed consent.
Statistical analysis
The data for LCI index, MES, and GHS were used for the analyses in the present study. The proportion and score of each component of the LCI index were compared between MES 0 vs. 1 and GHS < 2B.1 vs. ≥ 2B.1 using Fisher’s exact test and Mann-Whitney U test, respectively. We set the threshold of the LCI index as the 75th percentile, and evaluated the sensitivity, specificity, positive predictive value, and negative predictive value according to histological healing. The concordance rate of the pathologist's judgment was evaluated using the Cohen’s κ coefficient [29 ]. κ-values were classified as follows: < 0, poor agreement; 0–0.2, slight agreement; 0.2–0.4, fair agreement; 0.4–0.6, moderate agreement; 0.6–0.8, substantial agreement; and 0.8–1.0 almost perfect agreement.
Statistical analysis was performed using SPSS software, version 25 (IBM Corp., Armonk, New York, United States) and R version 4.0.3 (The R Foundation for Statistical Computing, Vienna, Austria). P < 0.05 was considered statistically significant.
Results
Patient characteristics
We enrolled 98 UC patients with clinical remission and performed colonoscopies. Four patients with MES ≥ 2 and one patient with insufficient endoscopic images were excluded. In addition, one patient was excluded because of insufficient mucosal biopsy sample ([Fig. 2 ]). The clinical characteristics of 92 patients are shown in [Table 1 ]. The most common extent of colitis was pancolitis (60 patients, 65.2 %). As concomitant medications, 5-aminosalicylic acid, azathioprine, prednisolone, anti-tumor necrosis factor-α treatment, and tacrolimus were used in 80 (86.9 %), 23 (25.0 %), 1 (1.1 %), 10 (10.9 %), and two patients (2.2 %), respectively.
Fig. 2 Flow chart of this study. Of the enrolled 98 patients with UC, 169 samples in 92 patients were analyzed in this study.
Table 1
Patient characteristics at baseline (n = 92).
Characteristics
Total (n = 92)
Median age, y (IQR)
46 (36–60)
Male sex, n (%)
48 (53.3)
Median body mass index, kg/m2 (IQR)
20.5 (17.8–21.9)
Median disease duration, y (IQR)
8.0 (4.3–13.0)
Extent of colitis, n (%)
Pancolitis
60 (65.2)
Left-sided colitis
29 (31.5)
Proctitis
3 (3.3)
Concomitant drug, n (%)
5-aminosalicylic acid
80 (86.9)
Azathioprine
23 (25.0)
Prednisolone
1 (1.1)
TNF inhibitors
10 (10.9)
Oral tacrolimus
2 (2.2)
Total Mayo score, mean ± SD
0.76 ± 0.82
Median C-reactive protein level, mg/dL (IQR)
0.04 (0.02–0.10)
Median albumin, g/dL (IQR)
4.3 (4.2–4.5)
Median hemoglobin, g/dL (IQR)
14.0 (13.1–15.1)
IQR, interquartile range; SD, standard deviation; TNF, tumor necrosis factor.
Proportion of MES and LCI index
In 92 patients, the endoscopic findings of 85 samples of sigmoid colon and 84 samples of rectum were evaluated ([Table 2 ]). Among the 169 samples, MES was confirmed and finalized as MES 0 for 117 of 169 cases (69.2 %) and MES 1 for 52 of 169 cases (30.8 %). In LCI index, the proportion of cases in which the diagnosis of three central endoscopic judgment committee members was consistent in two or more members were 164 of 169 cases (97.0 %) in LCI-index R, 154 of 169 cases (91.1 %) in LCI-index A, and 151 of 169 cases (89.3 %) in LCI-index L. In LCI index-R, there were 22 of 169 cases (12.4 %) of Grade 0, 117 of 169 cases (69.8 %) of Grade 1, and 30 of 169 cases (17.8 %) of Grade 2. In LCI index-A, there were 113 of 169 cases (66.9 %) of Grade 0, 34 of 169 cases (20.2 %) of Grade 1, 17 of 169 cases (10.0 %) of Grade 2, and five of 169 cases (2.9 %) of Grade 3. In LCI index-L, there were 124 of 169 cases (73.8 %) of Grade 0, 27 of 169 cases (16.0 %) of Grade 1, 14 of 169 cases (8.4 %) of Grade 2, and four of 169 cases (2.4 %) of Grade 3.
Table 2
Proportion of MES and LCI index.
Endoscopic findings
Total (n = 169)
MES (0/1)
117/52
LCI Index
Redness (0/1/2)
22/117/30
Area of inflammation (0/1/2/3)
113/34/17/5
Lymphoid follicular (0/1/2/3)
124/27/14/4
LCI, linked color imaging; MES, Mayo endoscopic subscore; WLI, white light imaging.
Proportion of each LCI index component in patients with MES 0 and 1
The proportion of each LCI index in MES 0 and 1 is shown in [Fig. 3 ]. Regarding LCI index-R in MES 0, there were 20 of 117 cases (17.1 %) of Grade 0, 92 of 117 cases (78.6 %) of Grade 1, and five of 117 cases (4.3 %) of Grade 2. For MES 1, there were one of 52 cases (1.9 %) of Grade 0, 26 of 52 cases (50.0 %) of Grade 1, and 25 of 52 cases (48.0 %) of Grade 2. The proportion of LCI index-R (0 of 1 of 2) was associated with MES (0 of 1) (P < 0.0001). Regarding LCI index-A in MES 0, there were 98 of 117 cases (83.8 %) of Grade 0, 13 of 117 cases (11.1 %) of Grade 1, 4 of 117 cases (3.4 %) of Grade 2, and two of 117 cases (1.7 %) of Grade 3. In MES 1, there were 15 of 52 cases (28.8 %) of Grade 0, 21 of 52 cases (40.4 %) of Grade 1, 13 of 52 cases (25.0 %) of Grade 2, and three of 52 cases (5.8 %) of Grade 3. The proportion of LCI index-A (0 of 1 of 2 of 3) was associated with MES (0 of 1) (P < 0.0001). Regarding LCI index-L in MES 0, there were 92 of 117 cases (78.6 %) of Grade 0, 16 of 117 cases (13.7 %) of Grade 1, five of 117 cases (4.3 %) of Grade 2, and four of 117 cases (3.4 %) of Grade 3. In MES 1, there were 32 of 52 cases (61.5 %) of Grade 0, 11 of 52 cases (21.2 %) of Grade 1, and nine of 52 cases (17.3 %) of Grade 2. The proportion of LCI index-L (0 of 1 of 2 of 3) was associated with MES (0 of 1) (P = 0.009).
Fig. 3 Proportion of each LCI index in MES 0 and 1. a Proportion of LCI index-R. The proportion of LCI index-R (0/1/2) was associated with MES (0/1) (P < 0.0001). b Proportion of LCI index-A. The proportion of LCI index-A (0/1/2/3) was associated with MES (0/1) (P < 0.0001). c Proportion of LCI index-L. The proportion of LCI index-L (0/1/2/3) was associated with MES (0/1) (P = 0.009). The proportion of each component of the LCI index and MES was compared using Fisher’s exact test. LCI, Linked color imaging; MES, Mayo endoscopic subscore.
Proportion of each LCI index in patients with and without histological healing
Regarding the diagnoses made by the two pathologists, the concordance rate using the κ-score is shown in [Table 3 ]. "Chronic inflammatory infiltrate,” "crypt destruction,” and "crypt destruction" had a moderate agreement, and "structural changes,” "eosinophils in lamina propria,” "neutrophils in lamina propria,” and "neutrophils in epithelium" had a stationary agreement. The overall histological healing rate was 84.0 % (142 of 169). The proportions of each LCI index in histological healing and non-healing are shown in [Fig. 4 ]. Regarding the LCI index-R in histological healing, there were 21 of 142 cases (14.8 %) of Grade 0, 101 of 142 cases (71.1 %) of Grade 1, and 20 of 142 cases (14.1 %) of Grade 2. In histological non-healing, there were one of 27 (3.7 %) of Grade 0, 16 of 27 cases (59.3 %) of Grade 1, and 10 of 27 cases (37.0 %) of Grade 2. The proportion of LCI index-R (0 of 1 of 2) was associated with histological healing or non-healing (P = 0.013). Regarding LCI index-A in histological healing, there were 103 of 142 cases (72.5 %) of Grade 0, 27 of 142 cases (19.0 %) of Grade 1, nine of 142 cases (6.3 %) of Grade 2, and three of 142 cases (2.1 %) of Grade 3. In histological non-healing, there were 11 of 27 cases (40.7 %) of Grade 0, seven of 27 cases (25.9 %) of Grade 1, eight of 27 cases (29.6 %) of Grade 2, and one of 27 cases (3.7 %) of Grade 3. The proportion of LCI index-A (0 of 1 of 2 of 3) was associated with histological healing or non-healing (P = 0.0014). Regarding LCI index-L in histological healing, there were 106 of 142 cases (74.6 %) of Grade 0, 23 of 142 cases (16.2 %) of Grade 1, 11 of 142 cases (7.7 %) of Grade 2, and two of 142 cases (1.4 %) of Grade 3. In histological non-healing, there were 18 of 27 cases (66.7 %) of Grade 0, four of 27 cases (14.8 %) of Grade 1, four of 27 cases (14.8 %) of Grade 2, and one of 27 cases (3.7 %) of Grade 3. The proportion of LCI index-L (0 of 1 of 2 of 3) was not associated with histological healing or non-healing (P = 0.382).
Table 3
Concordance rate of pathological judgment using κ score.
Geboes histopathological score
κ score
Structural changes
0.702 (0.650–0.754)
Chronic inflammatory infiltrate
0.453 (0.395–0.511)
Eosinophils in lamina propria
0.670 (0.616–0.724)
Neutrophils in lamina propria
0.784 (0.719–0.849)
Neutrophils in epithelium
0,729 (0.651–0.807)
Crypt destruction
0.491 (0.275–0.707)
Erosion/ulceration
0.551 (0.431–0.671)
Fig. 4 Proportion of each LCI index in histological healing/non-healing. a Proportion of LCI index-R. The proportion of LCI index-R (0/1/2) was associated with histological healing/non-healing (P = 0.013). b Proportion of LCI index-A. The proportion of LCI index-A (0/1/2/3) was associated with histological healing/non-healing (P = 0.0014). c Proportion of LCI index-L. The proportion of LCI index-L (0/1/2/3) was not associated with histological healing/non-healing (P = 0.382). The proportion of each component of the LCI index was compared using Fisher’s exact test. Histological healing was defined as GHS < 2B.1. GHS, Geboes histopathological score; LCI, linked color imaging.
Next, we performed an additional analysis to examine if there was a difference in each component of the LCI index between the subjects with and without histological healing ([Table 4 ]). LCI index-R (P = 0.0036) and A (P = 0.0004) were significantly lower in the subjects with histological healing compared with those without histological healing. No significant difference in LCI index-L score was found between the subjects with or without histological healing (P = 0.322).
Table 4
Relationship between each LCI index and histological healing (GHS < 2B.1).
LCI index
Histological healing
P value[1 ]
(+)
(−)
Redness (0/1/2)
21/101/20
1/16/10
0.003
Area of inflammation (0/1/2/3)
103/27/9/3
11/7/8/1
0.0004
Lymphoid follicular (0/1/2/3)
106/23/11/2
18/4/4/1
0.322
GHS, Geboes histological score; LCI, linked color imaging.
1 Mann-Whitney U test.
Relationship between LCI index and histological healing in patients with endoscopic remission
In LCI index-R and A, which had significant differences compared with histological healing, we aimed to set a threshold of these components considering their convenience in daily practice. Because the 75th percentile of LCI index-R score in the subjects with histological healing was Grade 1, we set the threshold of LCI index-R to predict histological healing at Grade 1. Grade ≤ 1 vs. 2 of LCI index-R was significantly associated with histological healing vs. non-healing (P = 0.007). The 75th percentile of LCI index-A score in the subjects with histological healing was also Grade 1. Grade ≤ 1 vs. ≥ 2 of LCI index-A was significantly associated with histological healing vs. non-healing (P = 0.001) ([Table 5 ]).
Table 5
Diagnostic capability of LCI index for histological healing (GHS < 2B.1).
LCI index
Histological healing
Sensitivity
Specificity
PPV
NPV
P value
(+)
(−)
Redness
Grade ≤ 1[1 ]
122
17
0.859
0.370
0.877
0.333
0.007
Grade 2
20
10
Area of inflammation
Grade ≤ 1[2 ]
129
18
0.908
0.333
0.878
0.409
0.001
Grade ≥ 2
13
9
LCI, linked color imaging; NPV, negative predictive value; PPV, positive predictive value.
1 The 75th percentile of LCI index-R score in the subjects with histological healing.
2 The 75th percentile of LCI index-A score in the subjects with histological healing.
Discussion
Previously, Uchiyama et al. reported that MES subdivision was obtained from redness evaluated by LCI. They demonstrated that mucosal redness without visible vessels was diagnosed in 4.6 % and 34.6 % of patients with MES 0 and MES1, respectively [19 ]. In our study, LCI index-R Grade 2, representing redness without visible vessels, was observed in 4.3 % and 48.0 % of patients with MES 0 and MES1, respectively, consistent with the previous report by Uchiyama et al. LCI index may be able to subdivide MES 0 and 1, defined as endoscopic remission in WLI.
In the present study, LCI index-A was significantly correlated with histological healing in patients with UC in clinical remission. This finding provides the insight that LCI index-A – an index evaluating the area of inflammation from the viewpoint of redness as well as decreased visibility of vessels – could be a novel marker in endoscopic monitoring for UC.
The weakness of the previous endoscopic score is that it is evaluated by point (strongest point of inflammation). However, in clinical practice, inflammation has a certain regionality. We believe that the advantage of this study is that it incorporates regionality (LCI-index A) into the endoscopic score and emphasizes redness by LCI (LCI-index R). In colonoscopy with LCI, the boundary of areas with redness and loss of vascular pattern is easily determined. It is an advantage of LCI.
Histological healing is generally considered to be in a higher hierarchy than endoscopic remission [14 ]. Various studies showed that histologically active inflammation can exist even in mucosa with endoscopic remission [30 ]. However, there is insufficient evidence that histological healing is a superior predictor to endoscopic remission in predicting a preferable long-term prognosis. Furthermore, attention should be paid to the cutoff value defined for histological healing. In a Phase 3 clinical trial of ustekinumab for moderate to severe UC, the proportion of patients who achieved histologic improvement (defined as neutrophil infiltration in < 5 % of crypts, no crypt destruction, and no erosions, ulcerations, or granulation tissue) was numerically greater than that of patients who achieved endoscopic improvements [28 ]. Thus, the hierarchy between endoscopic and histological findings may change depending on the cutoff value for histological healing.
Histological findings are limited as a point evaluation, while endoscopic findings can be an area evaluation. However, for UC characterized by continuous lesions, no endoscopic scoring system is currently available that considers areas of inflammation. Yoon et al. reported that UC patients who achieved MES 0 had a lower risk of relapse within 12 months than who achieved MES 1. Furthermore, they reported that patients who achieved MES 0 with histological healing had a lower risk of relapse [31 ]. However, these were analyses in MES that did not include evaluation of area. Although the long-term prognosis was not evaluated in our study, the LCI index, which includes the evaluation of area, may be involved in predicting relapse. In addition, in LCI index, lymphoid follicles were included in an evaluation item as LCI index-L. In this study, the relationship between LCI index and histological healing was evaluated for each index, and only LCI index-L was not associated with histological healing. In UC, lymphoid follicles may precede or accompany typical mucosal lesions, and these are considered to be an early lesion in UC [32 ]. In our study, all patients had a certain disease duration, and lymphoid follicles observed were unlikely to be early lesions in UC. Lymphoid follicles have not been analyzed by conventional endoscopic observation with WLI. Although LCI index-L was not correlated with histopathological findings, there is still a possibility that lymphoid follicles may affect the subsequent clinical course.
In addition, in our study, we evaluated only the association between each LCI index and histological healing, not the total score. The reason for this is that it was difficult to compare the importance of each index and grade. It also included that only the LCI index L was not associated with histologic cure. Adding the LCI index-L to the total score is considered difficult.
Our study had several limitations. First, there might be a selection bias because the subjects were selected by the attending physicians. Second, inflammation was evaluated only in the sigmoid colon and the rectum, and not in the entire colon. Although this study included many patients with total colitis, all patients in clinical remission were selected; thus, the evaluation was limited to the sigmoid colon. Third, because the target patients were limited to MES ≤ 1, much of the actual biopsy tissue showed mild inflammation in GHS. This might affect LCI-index evaluation, especially specificity. However, we believe that there is no point in assessing histopathological inflammatory activity by biopsy for assessment of disease activity, which is invasive, in cases with MES ≥ 2 with apparent endoscopic inflammation. Of course, a biopsy is necessary for evaluation such as infection or cancer. Finally, this study examined the correlation between LCI findings and histopathology, and a further prospective analysis is needed to determine if LCI index can predict long-term prognosis in UC patients with clinical remission.
Conclusions
We developed an LCI index that is useful for predicting histological healing in UC patients with clinical remission and MES ≤ 1. Additional studies are required to investigate the usefulness of the LCI index in predicting long-term prognosis.
