Tumor budding in pre-neoadjuvant biopsy and post-neoadjuvant resection specimens is associated with poor prognosis in intrahepatic cholangiocarcinoma—a cohort study of 147 cases by modified ITBCC criteria

Intrahepatic cholangiocarcinoma (iCCA), the second most lethal primary liver cancer after hepatocellular carcinoma, is expected to experience a ten-fold increase in global incidence over the next two to three decades [1]. iCCA patients have a 5-year survival rate of less than 10% due to late diagnosis, aggressive biological behavior, and limited options for effective treatment [2, 3]. Surgical resection is currently the only curative-intent treatment for early- to intermediate-stage resectable iCCA patients. Nevertheless, the 3-year recurrence rate is reported to be as high as 80%, which negatively impacts the outcome of surgery [4, 5].

In the past, when conventional chemotherapy was the mainstream treatment modality in late-stage iCCA, limited evidence showed the benefit of neoadjuvant therapy (NAT) [6, 7]. In the era of immunotherapy, the combination of immune checkpoint inhibitors with chemotherapy and/or targeted shows promising results in treating advanced iCCA by downstaging the tumor, increasing the rate of complete resection, and reducing the rate of local recurrence [8‐11]. The phase III randomized clinical trial TOPAZ-1 (NCT03875235) has shown significant improvement in treating advanced iCCA patients. The regimen used in this trial (durvalumab plus gemcitabine and cisplatin) has since been approved by the FDA for locally advanced or metastatic biliary tract cancer [12, 13]. Another phase III clinical trial (NCT04003636) with a larger patient cohort, involving pembrolizumab plus gemcitabine and cisplatin, also demonstrated that advanced iCCA patients can benefit from systematic immunotherapy [10]. Both results indicate that more iCCA patients can benefit from systematic immunotherapy, which also enlightens the future for borderline resectable iCCA patients [12, 14‐16]. Evidence from clinical practice also reveals that initially unresectable or borderline resectable iCCA patients can benefit from combined immunotherapy by downstaging the tumor and eventually undergoing surgical resection [17‐19]. Nevertheless, the effectiveness of combined immunotherapy varies significantly among iCCA patients [20, 21]. Clinicopathological parameters and biomarkers that could stratify patients into different prognostic and potentially therapeutic subgroups would be highly beneficial. Conventional pathological parameters, such as tumor size and grade, are known to be profoundly altered after treatment [22, 23]. To date, reliable clinicopathological parameters that can predict the survival of post-NAT iCCA patients are not well reported in the literature.

Tumor budding (TB) is a histomorphological parameter defined as the presence of single cell or tumor cell clusters comprising fewer than five cells within or at the invasive front of the tumor [24, 25]. High TB counts have been reported to indicate poor prognosis in various types of cancers, including recently published results in iCCA [26‐30]. Notably, the International Tumor Budding Consensus Conference (ITBCC) criteria have been recommended and adopted into the routine pathology report protocol for colorectal cancer resection specimens [31], after its prognostic value was validated in large prospective and population-based cohorts [32‐34]. On the other hand, high TB counts have also been reported as an adverse prognostic factor in biopsy specimens from different types of cancers [35, 36]. Increasing evidence from post-NAT studies investigating TB in rectal [37], esophageal [38], gastric [39] cancers, and perihilar cholangiocarcinoma [40] suggests that even after NAT, where conventional histopathological grading is usually omitted, the prognostic power of TB is retained.

In this study, we evaluated tumor budding in pre-NAT biopsy specimens and subsequent surgically resected iCCA specimens from patients who received NAT, to determine its prognostic significance.

A total of 147 cases were included in our study, comprising 95 cases of biopsy material obtained from patients before NAT and 139 cases of post-NAT resection with viable tumor cells for TB assessment. All surgical margins were negative, with a range of 0.1 to 5.0 cm. Eight cases were not included in post-treatment analysis, because 5 cases showed complete tumor necrosis and 3 cases of resection specimen slides were not available for us. A total of 87 cases had paired pre-NAT biopsy and post-NAT resection specimens for TB assessment. The median follow-up time was 29 months, during this follow-up 79 patients (53.7%) developed disease progression and 32 patients (21.8%) died. TB counts could be readily identified in both biopsy and resection specimens (Fig. 1). cTNM stage was confirmed before NAT. TRS did not show significant association with TB grade (P = 0.395) (Table 1). TRS grade 3 did not show significant association with either OS (P = 0.870) or RFS (P = 0.205) in the univariate analysis (Tables 2,3), as well as on Kaplan–Meier curves for either OS (P = 0.382) or RFS (P = 0.052) (Supplemental Fig. 1).

A biopsy is not routinely performed for surgically resectable iCCA. However, it is a prerequisite to establish the diagnosis and facilitate subsequent NAT [42]. TB counts have been reported to be applicable in biopsy specimens, correlated well with TB counts in the following surgical resection specimens and prognostic value for survival in several other types of cancers [35, 36, 43]. A recent study [26] on treatment-naive iCCA resection specimens modified ITBCC criteria by assessing TB in whole tumor areas rather than at the invasive front because the authors believed determination of the invasive front was frequently difficult. This study also yielded a positive correlation between high TB counts and poor prognosis. These findings provide scientific and histological evidences to evaluate TB in iCCA biopsy specimens even if the tumor invasive front may not be representative on the limited biopsied tumor tissue. In our study, a relatively weak correlation between TB counts in paired pre-NAT biopsy and post-NAT resection supported the reliability of TB counts reporting in biopsy specimens (Fig. 3). The optimal prognostic cut-off value of TB counts in biopsy varies from 1 to 10 buds in the literature [43]. In our study, we have observed TB counts of 3 to be the best to predict OS, which is similar to the cut-off value in rectal adenocarcinomas [35, 36, 44]. However, such prognostic value is not well achieved in predicting RFS, possibly because not all resection cases’ biopsy slides are available to be included in the study. Further study, preferably a prospective randomized clinical trial with a larger case number, is needed to determine the best cut-off value and its prognostic power in the treatment selection of iCCA.

Recent evidence showed that a proportion of iCCA patients could benefit from NAT to downstage the iCCA in order to achieve surgical resection [20]. Pathological examination of the resection specimens delivers crucial information in the post-NAT setting in many types of cancers, whereas this topic is not well addressed in iCCA [37‐40]. In our study, the AJCC tumor stages only showed suboptimal prognostic value, as the cTNM stages and ypTNM stages did not achieve statistical significance (Tables 2 and 3). The TRS also fails to demonstrate prognostic value for post-NAT iCCA patients (Supplemental Fig. 1). Therefore, additional biomarkers that are able to stratify these patients into different prognostic and potentially therapeutic groups would be highly beneficial. TB has been identified as a highly valuable histopathological parameter in a variety of cancer types [29, 30, 40, 45‐47] and recent studies also demonstrated its prognostic value in treatment-naive BTCs [26‐28, 40, 48]. Our study investigated TB according to the ITBCC criteria, a reproducible TB grading scheme validated by multi-center prospective studies in colon cancer, in post-NAT iCCA resection specimens and identified TB as a strong and independent prognostic factor for OS and RFS. The TB-positive group outperforms all other clinicopathological parameters, including the TNM stage, in multivariate analysis for survival analysis (Tables 2 and 3). Our finding, together with other similar TB studies in colorectal [37], esophageal [38, 49], gastric [39] cancer, and perihilar cholangiocarcinoma [40] post-NAT resection specimens, strongly supported that TB counts retained its significant prognostic power in cancer resection specimens after NAT and shall be considered to be included into the routine pathological report protocol.

Additionally, we have observed that a two-tier grading scheme, merging ITBCC intermediate and high groups into one “positive” group, showed better prognostic power than the original three-tier grading scheme in ITBCC (Fig. 4, Supplemental Fig. 3). Practically, a two-tier scheme is simpler, more cost-effective, and reproducible than a three-tier scheme in routine practices.

The statistically borderline significant, but weak positive, correlation of TB counts in biopsy and in subsequent post-NAT resection specimens (Fig. 3) suggests that TB may not only be a morphological feature of tumor growth but also related to the intrinsic aggressive biological characteristics of the tumor. As TB retained an association with poor prognosis before and after NAT, a consistent biomarker in different spatiotemporal conditions of the tumor, it may be related to certain key biological factors in driving cancer invasion and metastasis. Emerging evidences suggest that TB may play a role in epithelial-mesenchymal transition (EMT) and other molecular pathways of tumor microenvironment in promoting cancer progression [25]. Our results, in a post-NAT setting, also showed high TB counts are related to high perineural invasion. The specific pathologenetic and molecular aspects of these phenomena are worth further investigation, particularly by the latest sub-histology level analysis technology such as single cell sequencing and spatial multi-omics [50].

This study has several limitations. Patients who had undergone a biopsy but were not eligible for surgical resections were not included. And our study is a single-center retrospective cohort study in nature. A prospective multicenter study is preferred to further validate our results in the future.

In summary, our study demonstrated that tumor budding, evaluated by modified ITBCC criteria, provides independent prognostic information in pre-NAT biopsy and post-NAT resection iCCA specimens. Our findings, in line with other evidences, support tumor budding to be included as a routine parameter in the pathology report of iCCA.