Cumulative Histologic Inflammation Predicts Colorectal Neoplasia in Ulcerative Colitis: A Validation Study

Abstract

Background

Chronic inflammation in ulcerative colitis (UC) is associated with the development of colorectal neoplasia (CRN). A group at St. Mark’s Hospital reported a novel cumulative inflammatory index that predicted the development of CRN in UC patients that we validated with an independent, well-described, matched, case-controlled cohort from the University of Chicago.

Methods

Cumulative inflammatory burden (CIB) was calculated by summing each histological inflammatory activity (HIA) score and multiplying it by the length of the surveillance interval. Persistency was defined by the number of surveillance episodes (with a severity score >2) divided by the total number of surveillance procedures. T tests compared the mean and maximum HIA scores, assessing mean and maximum severity, CIB, and persistency.

Results

Sixty-two UC patients (26 patients with CRN, 36 control patients without CRN) were analyzed. Fifty-five percent were men, mean disease duration was 20.6 years, and mean age at CRN diagnosis was 43.9. Of the CRN patients, 6 (23%) had colorectal cancer, 16 (62%) had low-grade dysplasia, and 4 (15%) had indefinite dysplasia. Using mean HIA scores, we found CIB to be statistically greater in CRN patients (P = 0.04). Using maximum HIA scores, we found CIB (P = 0.02), mean severity (P = 0.03), and persistency (P = 0.01) to be significantly greater in CRN patients. Maximum severity was numerically greater for mean and maximum HIA scores but did not reach significance.

Conclusion

Cumulative histologic inflammation is significantly associated with the development of CRN in UC patients. This suggests a management strategy of controlling inflammation to reduce the risk of CRN and may influence the selection of surveillance intervals.

INTRODUCTION

Inflammatory bowel disease (IBD) involving the large intestine is a risk factor for the development of colorectal cancer (CRC).^{1, 2} Other factors associated with CRC or precancerous dysplasia include duration and extent of disease, degree of histologic inflammation, age at diagnosis, backwash ileitis, pseudopolyps, concomitant primary sclerosing cholangitis, prior dysplasia, and family history of CRC.^1-6 Currently, endoscopy with targeted or nontargeted biopsies is the recommended surveillance approach, but data demonstrating a mortality benefit are lacking. Therefore, it is clinically important to identify surveillance strategies for IBD patients to determine those who are at higher risk for dysplasia and to stratify prevention strategies.

Several studies have been published investigating the relationship between histological inflammatory activity (HIA) scores and the risk of colorectal neoplasia (CRN). A retrospective case-control study performed at St. Mark’s Hospital showed that increased HIA scores were associated with an increased risk for developing CRN.³ These findings were confirmed by both a retrospective cohort study that validated the association between severe histologic inflammation and advanced CRN and a subsequent case-control study at the Helsinki University Central Hospital that concluded that both the degree of inflammation and disease duration compounded the risk for developing CRC.^{7, 8}

Our group published a case-control study investigating the relationship between mean and maximum HIA scores and the risk of developing CRN over time.⁹ Unique to our study was the regrading of all pathology specimens for degree of inflammation. We found that average inflammation over time was more predictive of an elevated risk for CRN compared to a single severe score of inflammation. To assess the impact of cumulative inflammation on the risk of developing CRN, researchers at St. Mark’s developed an equation for cumulative inflammatory burden (CIB), calculated as the summation of each procedure’s microscopic inflammation severity score multiplied by the length of the surveillance interval.¹⁰ Using data from their single-center database, they found that both endoscopic and histologic CIB were significantly associated with the risk of developing CRN. Additional analysis revealed that mean severity, maximum severity, and persistent inflammation over time were also associated with increased CRN risk.

In this analysis, we sought to validate the previously reported CIB as an instrument to predict the development of CRN in ulcerative colitis (UC) patients.

METHODS

We used a previously described case-controlled cohort of UC patients from the University of Chicago with endoscopic biopsies obtained between May 1994 and June 2005.⁹ Biopsies using standard pinch forceps were routinely taken from the rectum and the right and left side of the colon. Cases were matched for age at diagnosis, histologic extent, and disease duration (within 5 years). In our previous publication, disease severity was defined using a modified Geboes 6-point HIA score (0 = normal; 1 = quiescent; 2 = increased granulocytes in the lamina propria only; 3 = intraepithelial granulocytes without crypt abscesses; 4 = crypt abscesses in <50% of crypts; 5 = crypt abscesses in >50% of crypts, erosion, or ulceration).⁹ Two pathologists had reviewed and scored all endoscopic biopsies. For training and agreement, the first, second, and third set of biopsies from 10 colonoscopies were reviewed independently by both pathologists and compared to ensure consistent scoring. Kappa statistics were calculated at 0.49, 0.54, and 0.60 for the first, second, and third set of biopsies respectively and showed consistent and moderate agreement between the 2 pathologists, with improvement over time. After consistency between the 2 pathologists was established, all other procedures were subsequently scored by 1 of the 2 pathologists.⁹ These same scores and data sets were used in this analysis. At our medical center, dysplasia is graded as indefinite, low-grade, or high-grade, and all readings of dysplasia or suspicion of dysplasia are routinely reviewed by multiple pathologists for confirmation. Indefinite dysplasia as read by our pathologists at the University of Chicago is considered in the spectrum of dysplastic change and is distinguished from reactive atypia as a matter of routine.

Statistical Analysis

We calculated 2 summative histologic inflammatory scores, mean and maximum, from all the biopsies taken for each colonoscopy. The CIB was calculated using the St. Mark’s equation:¹⁰

$$\text{CIB}={\displaystyle \sum {\text{H}}_{\text{i}}{\text{t}}_{\text{i}}}$$

The variable H represents the calculated HIA score of each surveillance procedure, and t is the length of time between each procedure in years. Persistency was also determined by calculating the number of surveillance episodes with mild to severe inflammation (HIA score >2) divided by the total number of surveillance procedures. We used the Student t test to compare mean and maximum HIA scores, CIB, and persistency between UC patients with and without CRN (control patients) and performed logistic regression analysis to obtain the odds ratio.

RESULTS

From the originally reported cohort of 200 patients (59 CRN patients, 141 control patients), 99 (23 CRN patients, 76 control patients) records with only a single scored colonoscopy were removed from the analysis. Four CRN patients and 9 control patients were excluded because of record errors or missing data. Of the remaining patients’ cases, 26 CRN patients had multiple colonoscopies (mean = 3; range = 1-5) and at least 1 matched control patient had more than 1 scored colonoscopy (mean = 2; range = 2-9). Sixty-two patients (26 CRN patients, 36 control patients) were analyzed.

Patient Characteristics

Twenty-six patients with UC-associated CRN and 36 control patients were included in this analysis. Six (23%) CRN patients had CRC, 16 (62%) had low-grade dysplasia, and 4 (15%) were indefinite for dysplasia. In addition, 80.8% of the CRN patients and 36.1% of the control patients were male. Four CRN patients (3 with low-grade dysplasia, 1 with indefinite dysplasia) and 2 control patients were diagnosed with primary sclerosing cholangitis. Patients with CRN and control patients were comparable in age and disease duration, disease extent, family history of CRC, and surveillance intervals (Table 1).

TABLE 1.

Demographics of Study Population (n = 62)

Characteristic	Cases (n = 26)	Controls (n = 36)
Male sex (n; %)	21; 80.8%	13; 36.1%
Age (mean; range)	42.9; 19-65	44.7; 23-69
Age at UC diagnosis (mean; range)	23.2; 4-40	23.2; 5-42
Disease duration in years (mean; range)	19.7; 5-37	21.4; 3-42
Disease extent (n; %)
Pancolitis	22; 84.6%	29; 80.6%
Left-sided	4; 15.4%	7; 19.4%
Smoking status (n; %)
Nonsmoker	19; 73.1%	29; 80.6%
Ex-smoker	4; 15.4%	2; 7.7%
Current smoker	3; 11.5%	5; 13.9%
Family history of colorectal cancer (n; %)
No	24; 92.3%	34; 94.4%
Yes	2; 7.7%	2; 5.6%
Primary sclerosing cholangitis (n; %)	4; 15.4%	2; 5.6%
Mean HIA score (median; range)	1.58; 0.29-5	1.33; 0.29-4
Maximum HIA score (median; range)	3; 1-5	3; 1-5
Number of colonoscopies per patient (median; range)	3; 2-8	2; 2-9
Number of inflammation scores per patient (median; range)	13; 1-37	11; 3-50
Surveillance interval in months (median; IQR)	25.3; 14.1-49.5	25.6; 14.9-35.3

Histologic Inflammatory Activity Scores

From 196 procedures, 2692 biopsy fragments were collected and scored. The number of procedures and histologic inflammation scores were consistent between CRN patients and control patients (Table 1).

Mean HIA scores

The CIB calculated using the mean HIA scores was statistically higher in CRN patients compared to their matched control patients (P = 0.04; Table 2). For each 1-unit increase in CIB, the risk of developing CRN increased by 1.08 (95% confidence interval [CI]; 1.01-1.17). This is consistent with the previously reported hazard ratio at St. Mark’s of 1.07 per 1-unit increase in CIB (95% CI; 1.04-1.11). Mean severity and persistency were not significantly different between groups (P = 0.15 and P = 0.58, respectively). Maximum severity trended toward significance for increased risk of CRN (odds ratio = 1.7; P = 0.06).

TABLE 2.

Case-Control Inflammation Severity Score

	Cases (n = 26)	Controls (n = 36)	Odds ratio (95% CI)	P
Mean HIA scores
Cumulative burden	12.63	7.98	1.08 (1.01-1.17)	0.0331*
Mean severity	1.82	1.58	1.76 (0.81-4.04)	0.1580
Maximum severity	2.42	1.94	1.71 (1.00-3.14)	0.0615
Persistency of inflammation	0.31	0.27	1.52 (0.34-6.85)	0.583
Maximum HIA scores
Cumulative burden	22.63	13.93	1.05 (1.01-1.11)	0.01847*
Mean severity	3.36	2.80	1.79 (1.06-3.19)	0.0358*
Maximum severity	4.15	3.64	1.52 (0.96-2.56)	0.0884
Persistency of inflammation	0.93	0.76	19.54 (1.90-469.72)	0.0305*

*Indicates P-value <0.05.

Maximum HIA scores

The CIB calculated with the maximum HIA score was also statistically significant between CRN patients and control patients (P = 0.02; Table 2). Again, CIB was significantly associated with the development of CRN. For each 1-unit increase in CIB, the risk of developing CRN increased by 1.05 (95% CI; 1.01-1.11). Mean severity and persistency of inflammation were also significantly associated with increased risk of CRC (P = 0.03 and P = 0.01, respectively). Maximum severity was numerically greater in patients who developed CRN than in the control group but did not reach statistical significance (P = 0.06).

DISCUSSION

The development of UC-associated CRN is a serious disease complication in IBD patients. Patients with UC who develop CRN have a worse prognosis than patients with sporadic CRN, and CRN accounts for 15% of mortalities in all patients with UC.¹¹ Early detection of CRN is therefore very important in the management of UC patients, and routine surveillance by endoscopy is recommended after 8 to 10 years of colitis.¹²

The metric CIB was developed by the St. Mark’s group to evaluate UC patients’ cumulative risk of developing CRN.¹⁰ One limitation to the St. Mark’s study is that all the endoscopic and HIA scores evaluated were taken from segments of the colon with the worst documented inflammation. However, neoplasia can also be located in less-inflamed portions of the rectum and colon.¹³ In our study, we used all available biopsies from the colon and rectum obtained during routine surveillance colonoscopies and scored using a 6-point scale to account for low-grade chronic inflammation.

In our cohort, we found CIB to be consistently associated with the risk of CRN using both the mean score from all biopsy fragments taken during surveillance endoscopies and the maximum score taken from the most severely inflamed segment of the colon. We also found that having persistent inflammation was associated with CRN when using only the maximum score from the most inflamed segment. Persistent inflammation was not significant when we used the average scores from the entire colon. The association between persistent inflammation and CRN when using the maximum score still suggests that reducing inflammation overall is an important endpoint for UC patients and may reduce their risk of developing CRN.

The main limitation in this study was the small sample size of patients in our case-controlled cohort who had multiple colonoscopies scored by our 2 pathologists. Other limitations in our study are similar to those we have previously discussed, including the retrospective nature of the analysis, selection bias for patients with frequent sigmoidoscopies or colonoscopies, and possible inclusion of nondiseased mucosa.⁹ In addition men constituted a greater proportion of patients with CRN than control patients.

CONCLUSION

In summary, we have provided validation of a CRC risk surveillance metric for UC patients based on microscopic inflammation severity and surveillance intervals. This metric may be useful to physicians to risk-stratify UC patients at risk for developing CRN, and it lends additional support to emerging treatment endpoints of UC patients that include mucosal healing and histologic remission.

Author contributions: OVY: Study concept and design, analysis and interpretation of data, drafting of manuscript, critical revision of manuscript. VR: Analysis and interpretation of data, drafting of manuscript. PHS: Analysis and interpretation of data. JH: Acquisition of data. JRT: Acquisition of data. CW: Acquisition of data. KEJ: Analysis and interpretation of data. DTR: Study concept and design, acquisition of data, critical revision of manuscript.

Supported by: Funded in part by the Digestive Disease Research Core Center of the University of Chicago (DK42086) and the NIH-NIDDK (R01DK068271, R01DK061931).

Presented at: Some of these data were originally presented at the European Crohn’s and Colitis Organization Congress Copenhagen, 2019, and Digestive Diseases Week, San Diego, 2019. None of these data have been published elsewhere.

Conflicts of interest: The authors have no disclosures.