Background
Periampullary adenocarcinoma, including pancreatic cancer, is a heterogenous group of tumours that originate in the area around the caput pancreatitis and the periampullary region. They have a dismal prognosis due to lack of symptomatology until late stages and lack of effective conventional therapy. Today, the 5-year survival rate for patients with pancreatic cancer is 7% and the positive survival trends seen in other major cancer types have not yet been observed in these patients [
1]. Recent research has started to explore the significance of the immune system and components of the inflammatory tumour microenvironment as potential novel treatment targets in a variety of solid cancers, including pancreatic cancer.
Professional antigen presenting cells (APC) such as dendritic cells (DC) and macrophages play a pivotal role in tumorigenesis and in the complex microenvironment of pancreatic cancer and periampullary adenocarcinomas [
2]. Tumour associated macrophages (TAM) can roughly be divided into two subtypes, M1 and M2, whereof M1 are associated with pro-inflammatory properties and M2 anti-inflammatory properties. TAMs show a complex phenotypic and behavioural variation and have the ability to promote angiogenesis, invasion, metastasis and regulate of inflammation [
3‐
5]. Although the M1–M2 polarization model may be useful, it should be pointed out that TAMs exist on a spectrum, exhibiting potent spatiotemporal plasticity in regard to phenotype [
6]. Previous research in pancreatic cancer has shown that the presence of M2 polarized TAMs is associated with poor prognosis [
7]. Another study could only establish a prognostic value to M2 TAMs, and not to the pan-macrophage tissue resident population [
8]. Further, TAMs, both CD68
+ and CD163
+, located in the stromal compartment have been shown to have prognostic value in breast cancer, illustrating that the localisation of TAMs within the histological architecture is relevant [
9].
The macrophage receptor with collagenous structure (MARCO) is a scavenger receptor involved in the recognition of pathogens through pathogen associated molecular patterns (PAMPs) [
10,
11]. MARCO is expressed by both activated DCs and a restricted population of tissue resident macrophages, and besides playing an essential role in recognising PAMPs, it is also involved in migration capacity [
10]. In addition, there is also evidence of a modulating role of MARCO on Toll-like receptors (TLRs) and thus the innate immune response to pathogens [
12,
13]. Due to its functions and interactions, MARCO has been put forward as a potential novel immunotherapy target and it was recently shown to reduce tumour growth in mouse models of cancer as well as adding to the effect of checkpoint therapy [
10,
14]. As of yet, the prognostic and potential predictive role of MARCO
+ TAMs have neither been described in periampullary adenocarcinoma nor in pancreatic cancer.
As with TAMs, tumour infiltrating DC (TIDC) show a complex phenotypic variation and has the capacity to interact with many of the plethora of cells present in the tumour microenvironment [
15]. High TIDC density has been shown to correlate with an improved prognosis in pancreatic cancer [
16], however another study could not establish any association between survival and TIDC density due to scarcity of infiltration [
17]. TIDCs have previously been reported to be either correlated with poor prognosis in cancer when expressing the immature DC marker CD1a, or to be beneficial for prognosis when expressing the mature DC markers DC-LAMP or CD83 [
18]. It should be noted that cutaneous dendritic cells, such as Langerhans cells, express CD1a even throughout maturation [
18].
There is accumulating evidence that the morphological subtype of periampullary adenocarcinoma is of larger importance than the anatomical origin, with pancreatobiliary-type (PB-type) tumours having a worse prognosis than intestinal-type (I-type) tumours [
19]. Thus, studies related to the prognostic and predictive role of components of the tumour microenvironment in periampullary adenocarcinoma need to take morphology into consideration. As of yet, we are not aware of any studies that have investigated the density and prognostic significance of TAMs, TIDCs or subtypes of these in relation to morphological subtype in periampullary adenocarcinoma. Therefore, the aim of this study was to explore the clinicopathological correlates and prognostic impact of tumour-infiltrating macrophages (CD68
+, CD163
+ and MARCO
+) and tolerogenic immature DCs (CD1a
+) in a clinically well-annotated consecutive cohort of periampullary adenocarcinoma, with particular reference to morphological subtype and adjuvant chemotherapy.
Discussion
This study is, to our best knowledge, the first to investigate the prognostic role of tumour-infiltrating CD1a+, CD68+ and CD163+ and MARCO+ immune cells in periampullary adenocarcinoma, with particular reference to morphological type. The results demonstrate that high infiltration of CD1a+ TIDC is an independent predictor of a shorter survival in patients with PB-type tumours, but does not confer any prognostic value in patients with I-type tumours. These findings add to the growing evidence of a dysfunctional inflammatory microenvironment caused by a desmoplastic stroma in these particularly aggressive tumours.
A previous study failed to establish any association between survival and DC infiltration rates in pancreatic cancer, secondary to the scarcity of local homing [
17]. The association between high CD1a
+ TIDC density and poor prognosis may be explained by their immature state or even the induction of maturation defects in TIDC in situ, initiated by the tumour microenvironment and leading to an immunosuppressive phenotype encouraging tumour tolerance and immune evasion [
24]. A study by Yamamoto et al. demonstrated a positive impact of TIDCs on survival in patients with pancreatic cancer [
16]. However, in that study, anti-fascin was used as a marker for TIDCs, therefore potentially examining another TIDC population.
In the present study we chose to not investigate mature DCs or to identify functional subtypes of TIDCs. Future studies should however focus not only on the presence of tolerogenic immature CD1a
+ DCs in tumour tissue, but also on their functional phenotype, as this may add important information on their potential immune modulatory effect in the inflammatory tumour microenvironment. There has been a steep increase of interest in the field of immunotherapy in recent years and dendritic cell vaccine has shown some promise in combination with conventional chemotherapy in pancreatic cancer [
25‐
27]. The findings of the present study indicate that it is important to take morphological type into consideration when evaluating the results from such trials.
Moreover, this study demonstrated an association between high density of CD68
+ and CD163
+ TAMs with poor prognosis in the whole cohort, but not in strata according to morphological type, and not independent of established clinical prognostic factors. Tumour-educated TAMs facilitate progression of pancreatic cancer and promote angiogenesis, remodelling of stroma, epithelial-mesenchymal transition and extravasation of tumour cells [
28,
29]. Previous studies have demonstrated that TAMs are associated with poor prognosis in pancreatic cancer [
7,
8], which is in line with the findings of the present study encompassing the full spectrum of periampullary adenocarcinoma. Even though it is mainly the anti-inflammatory subpopulation of TAMs that has been suggested to promote tumour progression, the entire CD68
+ TAM population was also found to be associated with poor prognosis in the present study. The reason for this finding is most likely that the predominant subtype of TAM in the tumour microenvironment is leaning towards pro-tumour polarisation [
30], thus making up for a large part of the CD68
+ TAM population.
Further, the novel macrophage marker MARCO, which has been shown to be a target for immunotherapy [
14], was found to be associated with poor prognosis in I-type but not PB-type tumours. The observation that the prognostic impact of MARCO
+ cells was particularly evident in patients who received adjuvant chemotherapy, as opposed to patients who did not receive any adjuvant treatment, is noteworthy, despite the lack of a significant treatment interaction. The OS of patient with low MARCO
+ TAM infiltration who received adjuvant chemotherapy with I-type morphology had a remarkably better OS than patients that did not receive adjuvant chemotherapy. This result might indicate a potential predictive role of MARCO
+ TAM infiltration for chemotherapy response, or high density of MARCO
+ TAMs could be a sign of chemotherapy resistance. Further studies are needed to validate these results, especially in intestinal cancers.
Previous research on the role of MARCO in cancer has been scarce. One previous study on hepatocellular cancer showed that decreased expression of MARCO was associated with poor prognosis [
31], however, in contrast to the present study, that study did not look at immune cell specific expression of MARCO, but rather at intra-tumoural MARCO expression. The present study found a significant association between MARCO
+ cells and CD68
+ cells which confirms the results of Sun et al, where MARCO
+ cells co-localized with CD68
+ macrophages [
31]. If the association between high MARCO
+ immune cell infiltration and poor survival rates is due to the co-localization of MARCO
+ and CD68
+ cells, interaction with chemotherapy, or because of the biological mechanisms of MARCO is yet to be determined. Further research into the role of MARCO in periampullary/pancreatic cancer as well as in other intestinal cancer and a broader spectrum of solid cancers is highly warranted.
In a translational context, the findings from the analyses of human tumours in this study are well in line with previous in vitro studies. For instance, Karnevi et al. have shown an intricate interplay between macrophages and tumour cells in vitro, where tumour derived factors drive the differentiation of macrophages into a pro-tumour phenotype [
30]. Additionally, the opposite effect has been demonstrated for dendritic cells in vitro, where tumour cell derived factors inhibit and limit the normal anti-tumour function of DCs [
32]. Further, in vivo models have shown that macrophage infiltration increases with tumour progression, and that infiltration starts very early before any invasive potential has been developed by the tumour. Collectively, these findings support the conclusions in the present study, wherein dense infiltration of TAMs and DCs are shown to be associated with poor and improved prognosis, respectively, in periampullary adenocarcinoma.
Some subgroup analyses rendered rather small numbers of cases, in particular in strata according to adjuvant or no adjuvant therapy and morphological type. Therefore, the results from the present study need to be validated in additional and preferably larger patient cohorts. However, as about half of the patients in the herein analysed patient cohort received adjuvant chemotherapy and half of the patients did not, it may give some indications to the potential predictive value of the investigated biomarkers, despite the retrospective setting.
Another potential limitation to the study is the use of the TMA technique, and in particular the fact that the tissue cores were primarily sampled from areas with tumour and not the adjacent stroma. However, a large proportion of periampullary cancers have a comparatively high stromal/tumour cell ratio, and three 1 mm cores can be considered a generous sampling. However, validating studies should ideally be specifically designed for a more comprehensive mapping of immune cell signatures. In this context, the TMA technique is likely to be superior to whole tissue section analysis, since it allows for sampling of multiple tissue types from multiple tissue blocks, and thus for a more comprehensive analysis of the inflammatory microenvironment of individual tumours.
Authors’ contributions
Conceived and designed the experiments: KJ. Performed the experiments: SL. Analysed the data: SL, EK, KL, KJ, and MK. Contributed reagents/materials/analysis tools: BN, JEB, JEL. Wrote the paper: SL, KJ. All authors read and approved the final manuscript.