Colorectal Dysplasia and Malignancy Outcomes in PSC-IBD Requiring Liver Transplantation and Colectomy: A Multicenter Study

Introduction

Concomitant primary sclerosing cholangitis (PSC) and inflammatory bowel disease (PSC-IBD) is a rare condition affecting approximately 0.96% of patients with Crohn’s disease (CD) and 2.47% of patients with ulcerative colitis (UC).¹ Stricture formation of intra-hepatic bile ducts in PSC can progress to cirrhosis requiring liver transplantation (LT), and ulcerative colitis increases the risk of developing colorectal cancer (CRC) requiring total abdominal colectomy (TAC).^2,3 Because of the relatively small number of patients undergoing LT and TAC from any single center, disease characteristics and outcomes have been challenging to report.

Patients needing both LT for PSC and TAC for IBD-associated dysplasia or malignancy require complex clinical decision making and care coordination, but literature guiding surgical management is limited, with a recent systematic review only reporting 20 cases.⁴ A multicenter study of 127 patients with PSC-IBD requiring both LT and TAC showed that 50 patients (39%) underwent TAC for dysplasia or malignancy.⁵ However, details regarding the pathology characteristics were not reported.

A more thorough understanding of the unique characteristics of PSC-IBD associated dysplasia and malignancy will add valuable insights for clinical care and research for this rare subset of IBD patients. This study aimed to describe the patterns of colorectal dysplasia, malignancy, and ileal pouch anal anastomosis (IPAA) outcomes for patients with PSC-IBD undergoing LT for PSC and TAC for dysplasia or malignancy from a multicenter collaborative sample.

Methods

Results

Of 127 patients screened, 50 (39.4%) underwent TAC for IBD with dysplasia or malignancy. Of these 50 patients, 30 (60.0%) underwent TAC for dysplasia and 20 (40.0%) underwent TAC for malignancy. There were 22 (44.9%) patients who underwent TAC first and 27 (55.1%) patients who underwent LT first (Table 1). One patient was excluded due to missing date of TAC. Most patients were male (77.3% and 70.4%) and white (81.8% and 76.9%). UC was the most frequent diagnosis in both groups (90.9% and 92.0%). Malignancy as the indication for TAC was not significantly different between groups (11 [50.0%] vs. 9 [33.3%], p = 0.26).

Dysplasia was commonly multifocal in the TAC first and LT first groups (64.3% and 70.0%, respectively; Table 2). Malignancy was commonly right sided in both groups (55.6% and 63.6%), and the most frequent histologic subtype was adenocarcinoma (88.9% and 90.9%). T3+ disease was more common in the LT first group compared to the TAC first group, but this did not reach statistical significance (5 [50.0%] vs. 2 [25.0%], p = 0.37). Nodal involvement was significantly more common in the LT first group compared to the TAC group (5 [50.0%] vs. 0 [0.0%], p=0.04). Neither group had any patients with distant metastasis at diagnosis. Adjuvant chemotherapy was noted more frequently in the LT first group, but this did not reach statistical significance (4 [57.1%] vs. 1 [14.3%], p = 0.27). The median (IQR) time from LT to diagnosis of CRC was 11.4 (2.4 – 15.4) years. In LT first patients, maintenance immunosuppressive regimens most frequently contained tacrolimus alone (9 [36.0%]) followed by tacrolimus and corticosteroids (6 [24.0%]) and tacrolimus, corticosteroids, and mycophenolate (5 [20.0%]).

There was no significant difference in CRC recurrence or 5-year overall survival between both groups (2 [18.2%] vs. 2 [22.2%] and 7 [87.5%] vs. 6 [75.0%], all p = 1.0). The Kaplan-Meier analysis showed survival probabilities decreased gradually over time for both LT and TAC first groups, but there was no significant difference between the groups (log-rank p = 0.4; Figure 1). A Cox proportional hazards model indicated belonging to one group did not increase or decrease the risk of death compared to the reference group (p = 0.814, HR 1.05; 95% CI 0.81 – 1.42). A node-stratified Kaplan-Meier analysis showed nodal status (N0 vs. N1+) was a significant driver of mortality with large effect size (p = 0.035, HR 2.45; 95% CI 1.12 – 5.40), however the surgical sequence (LT first vs. TAC first) did not yield a statistically significant survival difference (p = 0.4) (Figure 2). For patients with malignancy, median follow-up time calculated from reverse Kaplan-Meier analysis (Figure 3) showed median duration of observation of 9.92 years, median overall survival (total) 28.3 years, and median follow-up 17.5 years (IQR 8.1 – 28.3). Using Gray’s test, we found no significant difference in the incidence of recurrence between LT first and TAC first groups (p = 0.46). Similarly, the cumulative incidence of the non-cancer mortality did not differ significantly between groups (p = 0.76).

There were 22 patients who ultimately underwent restorative IPAA, with 11 in the TAC first group and 11 in the LT first group (table 3). Both groups had the same rate of one or more episodes of pouchitis (54.5%). In the TAC first group, one patient had documented pouch failure (10.0%), one required pouch excision (9.1%), and one required pouch revision (9.1%). No patients in the LT first group had documented pouch failure nor required pouch excision or pouch revision.

Discussion

PSC-IBD is rare and as such, multicenter analyses are necessary for robust investigation of disease characteristics and patient outcomes. This multicenter study from nine centers across the United States presents, to our knowledge, the largest single study of patients with PSC-IBD requiring LT and TAC for colorectal dysplasia or malignancy to include discrete pathologic characteristics and IPAA outcomes where applicable. Our findings offer valuable insights that can be utilized in clinical practice and as the foundation for further research on this rare condition.

Prior work by our collaborative has shown that long-term survival in PSC-IBD related CRC in LT patients is contingent upon progression to end-stage liver disease and liver transplant rather than the need for TAC. It is important to note that while overall survival in PSC-IBD with LT is strongly influenced by liver disease and transplant related factors, TAC in these patients with dysplasia aims to prevent progression to cancer and is an important component of their care. In TAC first or LT first patients, survival was not different among patients undergoing TAC for dysplasia, malignancy, or medically refractory disease.⁵ Though a strict conclusion cannot be drawn from our data, 5-year overall survival was comparable to survival for CRC in the general literature. The American Cancer Society estimates the 5-year relative survival rate for localized and regional CRC is 91% and 73%, respectively.⁸ Our 5-year overall survival was 87.5% in the TAC first group and 75.0% in the LT first group. This is similar to other reports in the literature. For example, a propensity score matched analysis of 46 patients with UC-related CRC compared to sporadic CRC patients demonstrated 5-year overall survival was 82.2% and 81.7%, respectively.⁹ Furthermore, a systematic review of 15 studies including 88 patients with PSC-IBD reported overall survival of 79% at 3-years.¹⁰ Additionally, we had 5 patients (all in the LT first group) with node positive disease. During the 5-year follow-up period, 2 patients died from their cancer, yielding an estimated 5-year overall survival for LT first patients with node positive disease of 3/5 (60.0%).

Despite our findings of survivability, our results suggest LT first patients may have more advanced CRC at diagnosis when compared to TAC first patients. For the LT first patients, the median time to diagnosis post LT was 11 years. 50% of these patients had node positive disease, which was significantly higher than the TAC first group (5 [50.0%] vs. 0 [0.0%], p=0.04). This differs from a report of a mean time to CRC diagnosis post LT of 3 years, with only 4.5% of patients diagnosed with stage 3 disease.¹⁰ A plausible mechanism for these findings is the imprecision of small sample size. It is emphasized, however, that neither group in our sample contained patients who presented with distant metastasis. The Cox proportional hazards model shows that while positive nodal status is a significant driver of mortality (p=0.035), the surgical sequence (LT first or TAC first) itself is not an independent risk factor (p=0.814). This explains the nodal paradox, showing, perhaps, the benefits of transplantation and intensive surveillance in the LT first group likely mitigated their higher baseline cancer stage.

It remains unclear whether LT and its associated immunosuppression are associated with a higher rate of development of CRC in PSC-IBD patients. While immunosuppression may play a role, there may be other associated factors. While determining the exact duration of IBD is difficult due to its quiescent nature, our results suggest patients who have LT first have a longer history of IBD preceding CRC or dysplasia diagnosis. Given the well-established link between duration of IBD and malignancy development, it is critically important that teams caring for patients with PSC obtain a surveillance colonoscopy early in the disease course. Care teams should consider repeating colonoscopy for patients with PSC at the same frequency recommended for surveillance of IBD, given the higher likelihood of a quiescent IBD diagnosis. While the incidence of PSC in IBD patients is low, the incidence of IBD in PSC patients may be as high as 70 – 80%.^11,12

Reconstruction with IPAA is common after TAC. IPAA can be complicated by pouchitis, pouch failure, and the need for pouch revision. Our sample of patients with PSC-IBD undergoing LT and TAC for dysplasia or malignancy had similar rates of pouchitis compared to prior reports, with 54.5% in both the TAC first and the LT first groups. A study of 160 IPAA patients reported pouchitis rates of 45.5% for non-transplanted patients and 67% for transplanted patients.¹³ In our study, only one patient in the TAC first group had documented pouch failure. The reason for pouch failure was pouchitis, not dysplasia or malignancy of the pouch. No patients in the LT first group had documented pouch failure.

Our study has limitations. This study is primarily limited by its small sample size, which reduces the statistical power to detect small differences in survival and recurrence between the TAC and LT first groups. Consequently, the non-significant p-values (p=0.4 for overall survival) should be interpreted with caution and viewed as exploratory rather than definitive. Nevertheless, the rarity of this clinical scenario combined with a robust 17.5-year median follow-up provides unique insights into the long-term safety of the LT first approach that are not typically available in larger, shorter-term multicenter registries. Given the timespan necessary to accrue patients with these rare conditions and the multicenter nature of this study, a statistically valid control arm would be challenging to identify. It is also challenging to account for individual clinical treatment decisions made over a study period of this length spanning multiple centers including the acuity with which TAC was performed and the decision whether to perform IPAA or end ileostomy, the influences of changes in medical therapy during the time span of patients accrued, and how medical advances may or may not have influenced the development of dysplasia or malignancy. Furthermore, data on surveillance colonoscopy or pouchoscopy and cancer-specific survival were not available in this cohort.

In conclusion, in this study of 50 patients with PSC-IBD who underwent TAC for dysplasia or malignancy and LT for PSC, dysplasia was commonly multifocal and arose a median of 3.7 (1.6 – 6.2) years after LT. Colorectal cancer was commonly early-stage adenocarcinoma, with a trend towards more advanced presentations in LT first patients occurring a median of 11.4 (2.4 – 15.4) years after LT. Despite more regionally advanced disease at presentation for LT first patients, 5-year overall survival was comparable to TAC first patients and the general literature. IPAA outcomes were encouraging with overall similar rates of pouchitis to prior studies and low pouch failure.