Small duct and large duct type intrahepatic cholangiocarcinoma reveal distinct patterns of immune signatures

Cholangiocarcinoma (CCA) is a highly heterogeneous malignancy originating from the intrahepatic biliary epithelium (iCCA) or from extrahepatic bile ducts (eCCA). Patients with CCA have a poor prognosis and the incidence of iCCA is rising globally, accounting for about 10–15% of primary liver cancers (Bertuccio et al. 2013; Bridgewater et al. 2014). With surgical resection being the sole curative treatment option, the prognosis for iCCA patients remains unfavorable (Groot Koerkamp and Fong 2014). The new standard of care in the palliative setting is the combination of chemotherapy with gemcitabine and cisplatin and the immune checkpoint inhibitors durvalumab or pembrolizumab, leading to a median overall survival of 12.8 and 12.7 months, respectively (Oh et al. 2022; Kelley et al. 2023).

In recent years, histopathological characterization of iCCA revealed two distinct subtypes according to the size of the affected bile duct, which led to implementation in the WHO classification (WHO Classification of Tumours 2019, 5th ed. Vol. 1. Digestive System Tumours, 2019. [Online]. Available: https://publications.iarc.fr/ Book-And-Report-Series/ Who-Classification-Of- Tumours/Digestive-System-Tumours-2019.'; Kendall et al. 2019; Aishima and Oda 2015). On the one hand, small duct type iCCA (SD-iCCA) was found to be more peripheral in the liver and resembling a ductular and cholangiolar type (Liau et al. 2014; Chung and Park 2022). On the other hand, the large duct type (LD-iCCA) arises from large intrahepatic ducts closer to the liver hilum and contains mainly mucin-producing columnar tumor cells (Hayashi et al. 2016; Sigel et al. 2018; Chung and Park 2022). Remarkably, both subtypes differ in underlying diseases, survival, response to chemotherapy, and molecular alterations, emphasizing clinically relevant subtype heterogeneity (Kinzler et al. 2022; Aishima and Oda 2015; Chung et al. 2020; Kendall et al. 2019; Gerber et al. 2022).

Immunotherapy emerged in the last decade and revolutionized treatments and outcomes across multiple cancer entities, including CCA (Pan et al. 2020; Greten et al. 2023; Fiste et al. 2021). Although the results of the TOPAZ-1 and KEYNOTE-966 trial have opened new perspectives for palliative CCA patients (Oh et al. 2022; Kelley et al. 2023), immunotherapeutic approaches in the management of iCCA patients remain challenging as the immunosuppressive tumor microenvironment (TME) plays a pivotal role in iCCA progression and, thereby, potential response to immunotherapeutic agents (Greten et al. 2023; Banales et al. 2020). The highly reactive TME comprises a variety of immune cells, including cancer-associated fibroblasts, tumor-associated macrophages, endothelial cells, and lymphocytes (Xia et al. 2022b; Job et al. 2020; Banales et al. 2020), but detailed characteristics are lacking. Here, improved characterization of the immune landscape in iCCA holds substantial clinical potential, both for predicting response to immunotherapy and for identifying novel treatment strategies. A few preliminary studies investigated immune signatures as predictive biomarkers in iCCA (Xia et al. 2022b; Yoon et al. 2021; Konishi et al. 2022; Yugawa et al. 2021; Jing et al. 2019). However, differences in immune signatures between SD- and LD-iCCA remain unknown, and exploratory studies are lacking so far.

We hypothesized that the heterogeneity of iCCA subtypes is reflected in their immune patterns and that these differences could hold a significant potential for diagnostic and therapeutic personalized medicine.

By applying NanoString® technology, we exploratorily identified substantial differences in the local immune patterns of patient with SD- and LD-iCCA, including signatures of inflammation and immune response. By analyzing treatment-naïve tumor samples, our results suggest that the immune signatures are an intrinsic trait of the tumor types. Further, we corroborated our results by performing a multitude of complementary NanoString® technology analysis like pathway scoring, gene set enrichment analysis, differential expression analysis and cell type profiling to create a holistic model of the immune signatures in both iCCA subtypes. Considering the rapid emergence of immune-oncology diagnostics and treatment, our results provide insights and give evidence which can be used for biomarker discovery and to inform future studies with therapeutic intend before clinical translation.

Our data indicate that immune-related pathways are broadly downregulated in SD-iCCA, with the exception of complement signatures. Among 27 strongly differently expressed genes, 20 were downregulated in SD-iCCA with CEACAM6, DMBT1 and CD79A showing the strongest effect. From the 7 upregulated genes, CRP showed the highest differential expression and the complement factors C5, C4BPA, C8A and C8B comprised 57% of the upregulated genes. Remarkably, total TIL signatures were reduced in SD-iCCA. Cell type signatures differed in both iCCA subtypes and chemokine as well as cytokine-cytokine receptor interaction pathways were broadly downregulated in SD-iCCA.

In recent years, the nihilistic approach to the treatment of iCCA has been replaced by new therapies in the field of molecular and immunotherapeutic regimens. However, the heterogeneity of both iCCA subtypes and the lack of predictive biomarkers remain a major challenge in determining which patient subgroup will benefit from immunotherapy in order to provide stratified medicine. In addition to the known heterogeneity of SD- and LD-iCCA in terms of survival, response to chemotherapy, and molecular alterations, this study is the first to address differences in immune patterns. Our data demonstrate a downregulation of DMBT1 in SD-iCCA compared to LD-iCCA. DMBT1 is a mucin-like molecule that exerts functions in the regulation of epithelial differentiation and inflammation participating in mucosal immune defense (Mollenhauer et al. 2000; Bathum Nexoe et al. 2020). So far, only two studies investigated the role of DMBT1 in iCCA. Goeppert and colleagues did not observe differences in DMBT1 expression in iCCA compared to normal biliary tissue while they identified a significant decrease of DMBT1 expression in iCCA compared to biliary intraepithelial neoplasia (BilIN) 3 suggesting a tumor suppressing role of DMBT1 in early cholangiocarcinogenesis (Goeppert et al. 2017). In line, Sasaki et al. demonstrate a decreased expression of DMBT1 in tissue of invasive iCCA compared to intraductal papillary neoplasms (Sasaki et al. 2003). Against this background, one would rather suspect DMBT1 downregulation in LD-iCCA as this subtype is commonly associated with impaired survival and a more aggressive tumor biology (Kinzler et al. 2022). However, lack of DMBT1 expression in non-neoplastic biliary tissue of CCA patients was associated with poor survival while no significant impact on outcome was observed when DMBT1 expression was reduced in cancer cells (Goeppert et al. 2017). Interestingly, overexpression of DMBT1 was shown in primary sclerosing cholangitis as well as in hepatolithiasis (Bisgaard et al. 2002; Sasaki et al. 2003), two common risk factors for the occurrence of LD-iCCA (Aishima and Oda 2015). This may explain the upregulation of DMBT1 in LD-iCCA as hepatolithiasis was significantly more present in in this cohort in our study. It should be noted that both studies did not differentiate between iCCA subtypes, which may hinder comparability to our data. Further, we found a downregulation of CEACAM6 in SD-iCCA compared to LD-iCCA. CEACAM6 is a member of the immunoglobulin cell adhesion molecule superfamily, and its overexpression is associated with poor prognosis and invasiveness in vivo and in vitro in iCCA (Ieta et al. 2006; Liu et al. 2022; Kurlinkus et al. 2021). In addition, high levels of CEACAM6 are suggested as a screening parameter especially for eCCA (Rose et al. 2016), which is consistent with our data showing overexpression of CEACAM6 in LD-iCCA, as LD-iCCA are generally mucin-secreting tubular adenocarcinomas resembling perihilar and distal CCA (Kendall et al. 2019). Intriguingly, Ieta et al. could demonstrate that CEACAM6 overexpression is associated with chemoresistance to gemcitabine in vitro (Ieta et al. 2006) while two recently published studies revealed significant shorter progression-free survival for LD-iCCA receiving gemcitabine-based chemotherapy (Kinzler et al. 2022; Yoon et al. 2021). Thus, our data suggest that CEACAM6 could serve as a potential chemoresistant marker to gemcitabine especially in patients suffering from LD-iCCA.

Complement proteins, as a part of tumor microenvironment, can play a pivotal role in local immune response in various cancer entities (Roumenina et al. 2019). A recent proteomic analysis demonstrated that complement factors were significantly increased in CCA patients (Son et al. 2020) while it was shown that the reduced expression of complement factor H-related 3 is associated with poor prognosis and immune regulation in CCA patients (Wang et al. 2022). As such, presence of complement factor H-related 3 negatively correlated with tumor infiltrating lymphocytes like CD8 + T cells (Wang et al. 2022). In line with these findings, our data show an upregulation of various complement factors and a concomitant decrease of TIL in SD- compared to LD-iCCA. As part of the adaptive immune system, TIL can either target tumor cells to prevent carcinogenesis and cancer progression, or cancer cells can adopt strategies to evade the immune responses against the cancer, thus promoting tumor progression (Gooden et al. 2011). TIL comprise of highly heterogeneous immune cells, including CD8 + T cells. In iCCA, two studies have shown that an increase in TIL is associated with favorable outcome (Xia et al. 2022a; Yoon et al. 2021) while Goeppert et al. confirmed the prognostic value of TIL only for eCCA, but not for iCCA (Goeppert et al. 2013). For the first time, we could show increased CD8 + /TIL ratios in SD-iCCA compared to LD-iCCA in our study, which is in line with Xia et al. and Yoon et al. as this subtype is generally associated with better overall survival (Kinzler et al. 2022). However, Yoon et al. investigated TIL solely by determining the CD8 + status by immunohistochemistry in a sub cohort of PD-L1-inhibitor treated patients in recurrent or unresectable CCA that underwent upfront chemotherapy, which likely affected the immune landscape (Yoon et al. 2021), while we used NanoString® technology in treatment naïve tumor tissue. Thereby, our results were not potentially cofounded by prior treatments and thereby indicate a tumor-intrinsic trait.

In general, the presence of TIL- and chemokine-infiltrated TME is associated with better response to immune checkpoint blockade in CCA (Binnewies et al. 2018). So far, both iCCA subtypes are traditionally merged in the clinical context and are treated similarly with regard to chemo- and immunotherapeutic approaches. This long-held hypothesis is challenged by our finding that the immune signatures of LD-iCCA comprise higher levels of total TIL and chemokine signaling. Therefore, our results might suggest that this subtype is potentially more responsive to immunotherapy. However, studies investigating the potentially different response to immunotherapy in SD- and LD-iCCA with a possible link to TIL expression are needed.

Our study has several limitations that warrant discussion. We performed a retrospective analysis and selection bias cannot be ruled out. Our study population was small, and generalizability may not be presumed as this study was exploratory and hypothesis-generating in nature. However, we used treatment naïve tumor tissue of surgically resected specimen which ensured that the immune landscape of our samples was not altered due to prior anti-cancer treatments. Given these considerations, the substantial novelty of our data and the fact that data on this topic are lacking so far, the results of the present study are of high clinical relevance. Our aim was to exploratively analyze different immune patterns using NanoString® technology. We anticipate our results to accelerate and inform future work using confirmatory analysis such as immunofluorescence staining, flow cytometry, qPCR, in vitro, and in vivo experiments in prospective studies to augment and validate our results prior to clinical translation.

In conclusion, our study is the first to demonstrate that SD- and LD-iCCA are associated with dedicated local patterns of immune profiles. The substantial differences hold clinically relevant promise for biomarker discovery and treatment planning, especially using immune checkpoint therapy aimed at subtype-specific, personalized medicine. We anticipate our findings to inform future work which is needed to build upon and corroborate our findings, and thereby improve our understanding of iCCA biology for improved diagnostics, and treatment approaches.