Introduction: Generating high levels of immunosuppressive adenosine (ADO) in the tumor microenvironment contributes to cancer immune evasion. CD39 and CD73 hydrolyze adenosine triphosphate into ADO; thus, efforts have been made to target this pathway for cancer immunotherapy. Our objective was optimizing a multiplex immunofluorescence (mIF) panel to explore the role of CD39 and CD73 within the tumor microenvironment. Materials and Methods: In three-time points, a small cohort (n = 8) of colorectal and pancreatic adenocarcinomas were automated staining using an mIF panel against CK, CD3, CD8, CD20, CD39, CD73, and CD68 to compare them with individual markers immunohistochemistry (IHC) for internal panel validation. Densities of immune cells and distances from different tumor-associated immune cells to tumor cells were exploratory assessment and compared with clinicopathologic variables and outcomes. Results: Comparing the three-time points and individual IHC staining results, we demonstrated high reproducibility of the mIF panel. CD39 and CD73 expression was low in malignant cells; the exploratory analysis showed higher densities of CD39 expression by various cells, predominantly stromal cells, followed by T cells, macrophages, and B cells. No expression of CD73 by B cells or macrophages was detected. Distance analysis revealed proximity of cytotoxic T cells, macrophages, and T cells expressing CD39 to malignant cells, suggesting a close regulatory signal driven by this ADO marker. Conclusions: We optimized an mIF panel for detection of markers in the ADO pathway, an emerging clinically relevant pathway. The densities and spatial distribution demonstrated that this pathway may modulate aspects of the tumor immune microenvironment.

Immune checkpoint inhibitors have improved the treatment of cancer. However, many cancers do not respond to immunotherapy or develop resistance to therapy; therefore, efforts have been made to identify new targets and drugs that can improve the antitumor immune response [1‒5].

Simultaneous angiogenesis and immunosuppression play a crucial role in tumor growth and metastasis. For this reason, some studies have demonstrated the importance of adenosine (ADO) in the proliferation, migration, and invasion of cancer cells and tumor angiogenesis and immune evasion [5‒7].

CD39 is an ectoenzyme that degrades adenosine triphosphate and adenosine diphosphate into adenosine monophosphate [7, 8]. CD73, expressed on the cell surface, degrades adenosine monophosphate into ADO [1‒3].

Extracellular accumulation of adenosine triphosphate may activate the anti-cancer immune response by stimulating the P2 receptor subtype [9]. In contrast, extracellular ADO suppresses immune cells and promotes cancer progression [4, 6, 10, 11]. High CD39 and CD73 have been found in many types of cancers and variable proportions in different immune cell subsets, stromal cells, and endothelial cells [3, 12‒15]. CD39/CD73 overexpression is a predictor of adverse clinical outcomes. Elevated CD73 in colorectal cancer (CRC) [16‒20] and pancreatic ductal adenocarcinoma (PDAC) [21] have been shown to be associated with disease progression and therapeutic resistance.

The emerging role of the ADO pathway, especially as a critical regulator of tumor immunity, has presented opportunities to develop anti-CD39 and anti-CD73 therapy for many cancers [10, 19, 22‒24]. In this regard, a phase II, randomized study in patients with locally advanced, unresectable, stage III NSCLC who had not progressed following cCRT, COAST, showed objective response rate and progression-free survival in the oleclumab and durvalumab versus durvalumab alone [25], and recent preclinical studies suggest that blockage of this pathway is a potential therapeutic approach for many types of cancer [1, 25‒27].

Multiplex immunofluorescence (mIF) is a very well-known technique used in research and clinical settings to facilitate the detection of multiple markers within the complex tumor microenvironment [28, 29]. In addition, this method allows us to simultaneously detect more than six markers on a single tissue section and recognize spatial relationships between various cells, something that conventional immunohistochemistry (IHC) cannot offer [28, 30]. In this pilot study, we designed an mIF tyramide signal amplification (TSA) panel to explore the role of CD39 and CD73 in formalin-fixed, paraffin-embedded (FFPE) CRC and PDAC tissue samples to be applied in translational studies.

Study Population and Tissue Specimens

Sequential 4-µm thickness sections from FFPE surgically resected reactive human tonsil tissue samples collected from The University of Texas MD Anderson Cancer Center (Houston, TX) was used for conventional IHC and optimization of both single IF and mIF. Moreover, we prepared sequential 4-µm thickness sections from FFPE surgically resected CRC (n = 4) and PDAC (n = 4) for mIF staining and analysis. The criteria for inclusion in the study included the following: confirmed diagnosis of CRC or PDAC among patients treated between September 8, 2003 and December 17, 2012, available Epic EHR platform clinicopathologic information such as age, gender, tumor-node-metastasis stage, tumor size, adjuvant therapy, recurrence and vital status and sufficient tumor tissue for IHC staining and designation of at least five regions of interest (ROIs) – at minimum 1.65 mm2 total tumor area.

The clinical and pathological characteristics of the patient are summarized in online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000534677). In brief, 4 CRCs were selected from 4 men with a median age of 61 years (range: 40–78 years) and 4 primary PDACs were chosen from 3 men and 1 woman with a median age of 71 years (range: 57–84 years). Tumor stages varied from I to III. Seven (87.5%) patients received adjuvant therapy; the median tumor size was 6.3 cm for CRC and 2.8 cm for PDAC.

IHC Optimization

To morphologically contextualize CD39 and CD73 expression in CRC and PDAC, we performed IHC studies using a BOND autostainer (Leica Microsystems, Vista, CA, USA) in our laboratory [28, 31] against pan-cytokeratin (CK), CD3, CD8, CD20, CD39, CD68, and CD73. For each marker, cellular expression was detected using a Novocastra Bond Polymer Refine Detection Kit (Leica Microsystems, catalog #DS9800) with a diaminobenzidine reaction to detect antibody labeling and hematoxylin counterstaining. We used different dilutions and antigen retrieval conditions for antibody optimization until obtaining a uniform staining pattern using human reactive tonsil FFPE tissue for positive and negative controls.

Single Immunofluorescence Antibody Optimization

As described in our previous papers [31‒34], after optimizing and testing the antibodies by IHC, we developed an mIF panel containing seven antibodies: CK, CD3, CD8, CD20, CD39, CD68, and CD73 plus 4′,6-diamidino-2-phenylindole (DAPI). We evaluated the antibodies using the same positive controls from the IHC optimization for single IF optimization. The cases were stained using a Leica BOND RX, Leica Biosystems autostainer, and linked with a fluorophore from the Opal 7 color IHC kit (catalog # NEL797001KT; Akoya Biosciences, Marlborough, MA, USA), including Opal Polaris 520, 540, 570, 620, 650, 690 plus TSA fluorophore Opal Polaris 480 (catalog # FP1500001KT, Akoya Biosciences) and DAPI. We proceeded with baking and dewaxing (BOND Dewax Solution, Leica Biosystems). The slides were heated at 85°C for 25 min using Bond Antigen Retrieval Tris-ethylenediaminetetraacetic acid buffer (Tris-EDTA buffer solution) or citrate buffer according to the conditions determined by IHC to open antibody epitopes. Then, the slides were incubated for 60 min at room temperature with the primary antibody at the same dilutions used for IHC staining (online supple. Table 2), followed by three times washes in 1 × 2-methyl-2H-isothiazol-3-one (catalog # AR9590, BOND Wash Solution, Leica Biosystems) and after incubated for 10 min at ambient temperature with polymer horseradish peroxidase conjugated to anti-mouse or anti-rabbit secondary antibody (Akoya Biosciences). Then, five washes were carried out with BOND Wash solution and incubated with an Opal fluorophore tyramide (Opal Polaris 480, 520, 540, 570, 620, 650, and 690) for 10 min. According to the instructions, dilutions from 1:50 to 1:150 were used to detect the different antibodies. After four successive washes in BOND Wash solution, the slides were counterstained for 5 min with DAPI. Following staining, the slides were removed from the Leica BOND RX and then mounted with ProLong Diamond Antifade Mountant (Thermo Fisher Scientific, Waltham, MA, USA). Autofluorescence (negative control) slides were included in each staining run, using primary and secondary antibodies and Opal fluorophore tyramides. Similar to IHC, tests were performed using different antibody dilutions combined with different Opal fluorophores to obtain a uniform and correct staining pattern.

mIF Optimization

Following our previously published workflow [33, 35, 36], we combined the single protocols to generate an mIF protocol once each IF protocol was optimized. First, we stained human reactive tonsil tissues with primary antibodies as positive controls (online suppl. Fig. 1). Then, as in the single IF, the different markers were applied sequentially. We also completed detection for each marker before using the next antibody. Until we obtained the same staining pattern as that brought in a single IF, we configured the sequence of antibodies and tested each sequence several times using automated protocols. As determined by the Vectra Polaris 1.0.13 multispectral imaging system (Akoya Biosciences), to obtain similar ranges of expression, we carefully adjusted dynamic ranges [34] from the different antibodies linked with their fluorophore, with 50–150 ns of exposure time for each antibody. With this adjustment, we avoid a cross-talking reaction [37] between fluorophores or an umbrella effect, which happens when the expression of one antibody blocks the expression of another antibody in the same cell compartment [34] (Fig. 1a). Additionally, autofluorescence (negative control) slides were run in each run of multiplex staining.

Fig. 1.

Overview of pilot study cohort and analysis approach. Eight tumor tissues were analyzed, including four colorectal adenocarcinoma (CRC) and four pancreatic ductal adenocarcinoma (PDAC). a Multiplex immunofluorescence (mIF) workflow. b Cell subsets, including epithelial malignant cells, immune cells (T cells, cytotoxic T cells, B cells, and macrophages) and stromal cells. c ADO pathway: role of CD39 and CD73 within tumor microenvironment.

Fig. 1.

Overview of pilot study cohort and analysis approach. Eight tumor tissues were analyzed, including four colorectal adenocarcinoma (CRC) and four pancreatic ductal adenocarcinoma (PDAC). a Multiplex immunofluorescence (mIF) workflow. b Cell subsets, including epithelial malignant cells, immune cells (T cells, cytotoxic T cells, B cells, and macrophages) and stromal cells. c ADO pathway: role of CD39 and CD73 within tumor microenvironment.

Close modal

Image Tissue Scanning and Analysis

Scanning protocols were created before scanning the seven color slides. The resulting slides were scanned in the Vectra Polaris version using the fluorescence protocol [34], and the IHC-stained slides were scanned using the brightfield protocol in the same instrument. Single IF and mIF stained slides were scanned at a magnification of 10 × (1.0 μm/pixel), from 440 to 720 nm, to extract fluorescence intensity information from the tissue samples. Using Phenochart 1.0.9 image viewer software, five individual ROIs (669 × 500 μm each) were selected by a pathologist to represent the various elements of tissue heterogeneity and then scanned at high magnification (931 × 698 μm at resolution 20 × 0.5 μm/pixel). We selected similar ROIs to cross the different time points and quantify cell phenotypes at a similar location to guarantee high analytical reproducibility. The same approach used to extract fluorescence intensity was used to build the spectral library using InForm 2.8.2 image analysis software (Akoya Biosciences), tissue segmentation, cell segmentation, cell phenotyping classification, and to export the data and generated images (Fig. 1). Therefore, an algorithm was improved and applied in all cases. We quantified the individual marker expression from the mIF and IHC. It was expressed as density per mm2 (Table 1 and online suppl. Table 3). For the samples exploratory study, the individual markers were merged with the phenoptr script from R Studio (Akoya Biosciences) to obtain the co-expression of the markers from two different compartments on the tumor: the tumor compartment inside of the tumor nets – tumor epithelial cells and the stromal compartment between tumor nets. In the final report, the cell phenotype density was normalized per mm2 from each compartment and the total compartment, and then the median value from the three-time points was obtained.

Table 1.

Cell phenotypes densities in colorectal adenocarcinoma and PDAC

Cell phenotypeColorectal adenocarcinoma (n = 4)aPDAC (n = 4)aOveralla
CK+ 3,490 (2,270–4,550) 2,056 (1,105–2,759) 2,771 (1,105–4,550) 
CK + CD39+ 3 (0–10) 5 (0–20) 4 (0–20) 
CD3+ 60 (1–122) 48 (30–81) 59 (1–122) 
CD3+CD39+ 14 (5–30) 6 (0–16) 11 (0–30) 
CD3+CD8+ 1 (0–2) 3 (0–8) 2 (0–8) 
CD3+CD8+CD39+ 1 (0–1) 1 (0–1) 1 (0–1) 
CD20+ 2 (0–4) 4 (1–9) 3 (0–9) 
CD20+CD39+ 1 (0–1) 1 (0–1) 
CD68+ 6 (4–9) 4 (1–8) 5 (1–9) 
CD68+CD39+ 1 (0–3) 1 (0–2) 1 (0–3) 
CD39 + 269 (28–630) 256 (43–549) 273 (28–629) 
CD73+ 7 (0–10) 15 (12–28) 11 (0–28) 
Cell phenotypeColorectal adenocarcinoma (n = 4)aPDAC (n = 4)aOveralla
CK+ 3,490 (2,270–4,550) 2,056 (1,105–2,759) 2,771 (1,105–4,550) 
CK + CD39+ 3 (0–10) 5 (0–20) 4 (0–20) 
CD3+ 60 (1–122) 48 (30–81) 59 (1–122) 
CD3+CD39+ 14 (5–30) 6 (0–16) 11 (0–30) 
CD3+CD8+ 1 (0–2) 3 (0–8) 2 (0–8) 
CD3+CD8+CD39+ 1 (0–1) 1 (0–1) 1 (0–1) 
CD20+ 2 (0–4) 4 (1–9) 3 (0–9) 
CD20+CD39+ 1 (0–1) 1 (0–1) 
CD68+ 6 (4–9) 4 (1–8) 5 (1–9) 
CD68+CD39+ 1 (0–3) 1 (0–2) 1 (0–3) 
CD39 + 269 (28–630) 256 (43–549) 273 (28–629) 
CD73+ 7 (0–10) 15 (12–28) 11 (0–28) 

aMedian (interquartile range, IQR).

Exploratory Cellular Distribution Analysis

Using nearest neighbor distance (NND) analysis from the spatstat package v.4.2.0 in R, we calculated the proximity of tumor cells to the different tumor-associated immune cells. NND computes the distance (µm) between each point, “i” (i.e., tumor cell) to its nearest neighbor point “j” (i.e., immune cell), precisely, whether the tumor-associated immune cells were close to (equal to or less than the median distance) or far from (more than the median distance) the CK + tumor epithelial cells. Nine cell phenotypes, including T cells, B cells, macrophages, and cells expressing CD39 and CD73, were studied using this approach (Table 2).

Table 2.

Distance between immune cells and the closest tumor cell in colorectal adenocarcinoma and PDAC

Immune cell subsetDistance to closest malignant epithelial cell (μm)Overalla
colorectal adenocarcinoma (n = 4)aPDAC (n = 4)a
CD3+ 37 (14–136) 49 (22–113) 43 (14–136) 
CD3+CD8+ 38 (27–57) 22 (13–36) 29 (13–57) 
CD20+ 37 (20–89) 57 (16–128) 45 (16–128) 
CD68+ 25 (12–45) 44 (15–137) 37 (12–137) 
CD3+CD39 33 (14–147) 56 (20–169) 43 (14–169) 
CD3+CD8+CD39+ 17 (11–23) 26 (26) 19 (11–26) 
CD20+CD39+ 18 (8–22) 26 (26) 19 (8–26) 
CD68+CD39+ 31 (7–166) 86 (29–143) 40 (7–166) 
CD3+CD73+ 70 (70) 65 (11–118) 67 (11–118) 
Immune cell subsetDistance to closest malignant epithelial cell (μm)Overalla
colorectal adenocarcinoma (n = 4)aPDAC (n = 4)a
CD3+ 37 (14–136) 49 (22–113) 43 (14–136) 
CD3+CD8+ 38 (27–57) 22 (13–36) 29 (13–57) 
CD20+ 37 (20–89) 57 (16–128) 45 (16–128) 
CD68+ 25 (12–45) 44 (15–137) 37 (12–137) 
CD3+CD39 33 (14–147) 56 (20–169) 43 (14–169) 
CD3+CD8+CD39+ 17 (11–23) 26 (26) 19 (11–26) 
CD20+CD39+ 18 (8–22) 26 (26) 19 (8–26) 
CD68+CD39+ 31 (7–166) 86 (29–143) 40 (7–166) 
CD3+CD73+ 70 (70) 65 (11–118) 67 (11–118) 

aMedian (interquartile range, IQR).

Statistical Analysis

To verify the marker expression correlation consistency across the different time points and their IHC, we used the Spearman rank correlation coefficients adjusted by Bonferroni correction. The χ2 test was used to compare categorical variables to determine their relationship. For the exploratory analysis of the cases, median densities and distances were dichotomized as high densities or long distances more than the median and low densities or close distances less or equal to the median. The Mann-Whitney and Kruskal-Wallis tests were used to analyze the associations between CD39 and CD73 co-expression with categorical factors. To assess the percentages of cell phenotypes, we divided the density of the phenotype by the density of total nucleated cells (DAPI+). All statistical analyses and data visualization were performed using R 4.2.2 (released October 2022), Phenoptr R package v.0.2.6, and/or the GraphPad Prism software program (version 9.2.0, released July 2021, GraphPad, San Diego, CA, USA) with the statistical significance set at a p value <0.05.

Marker Expression in Controls, IHC, and mIF

We optimized the various markers to obtain similar distribution patterns between IHC and mIF in the tonsil. As expected, CK is expressed by epithelial cells, and among cells surrounding the germinal centers, the T-cell marker CD3 was the most abundant, followed by CD8. B-cell markers CD20 and CD68, a macrophage marker, were localized in the germinal centers of the tonsil (online suppl. Fig. 1). In addition, we observed the ADO markers CD39 and CD73 on various immune cell populations and cells with histomorphology suggestive of endothelial cells (online suppl. Fig. 1). Our team previously reported antibody validation, staining, and pathology evaluation [32, 38, 39].

Quantification analysis from the individual markers, CK, CD3, CD8, CD20, CD39, CD68, and CD73, showed a high correlation between the cell densities across different mIF batches and the IHC-stained as shown in online supplementary Figures 2 and 3, except for CD8, probably due to the density variation observed in the different levels of histologic sample sections (online suppl. Table 3). In the case samples, we observed CD39 and CD73 membrane or membrane and cytoplasmic expression in tumor epithelial cells, immune cells, and various stromal cells. Overall, we observed a low variation of staining intensity and density expression among the three sample batches (online suppl. Fig. 2; Table 3). Likewise, the other markers, CK (epithelial cell positive), T-cell lymphocytes (CD3-positive), cytotoxic T cells (CD8-positive), B-cell lymphocytes (CD20-positive), and macrophages (CD68-positive), were observed in their immune cells’ expression, respectively, with a similar pattern of distribution on both IHC and mIF staining (Fig. 2).

Fig. 2.

Microphotographs showing markers expression in colorectal adenocarcinoma (top) and in PDAC (bottom – red arrow) and normal pancreatic tissue (bottom – white arrow).

Fig. 2.

Microphotographs showing markers expression in colorectal adenocarcinoma (top) and in PDAC (bottom – red arrow) and normal pancreatic tissue (bottom – white arrow).

Close modal

Exploratory Analysis of the ADO Pathway in the Pancreatic and Colorectal Adenocarcinoma Samples

We began our analysis by assessing cell populations: tumor epithelial cells, immune cells, and other cells (Fig. 1b, 3a). Overall, our tumor samples did not show high densities of immune cells. The predominant immune cells subset were T cells (CD3+; median, 59 cells/mm2), followed by macrophages (CD68+; median, 5 cells/mm2), B cells (CD20+; median, 3 cells/mm2), and cytotoxic T cells (CD3+CD8+; median 2 cells/mm2) (Table 1).

Fig. 3.

Immune cell densities in the CRC and PDAC cohort. a Example of multiplex immunofluorescence (mIF) image in colorectal adenocarcinoma. b Total immune cell densities in the CRC and PDAC. c Distribution of overall (combined tumor epithelial and stromal compartments) immune cell subsets densities. d Immune cell subsets densities in 4 up-front resected CRCs and 4 up-front resected PDACs. e Boxplots showing immune cells densities (T cells, cytotoxic T cells, B cells, and macrophages) in tumor and stroma compartments from our cohort. f Boxplots depicting immune cells densities (T cells, cytotoxic T cells, B cells, and macrophages) in tumor and stroma compartments from CRC and PDAC. No statistical significance was found.

Fig. 3.

Immune cell densities in the CRC and PDAC cohort. a Example of multiplex immunofluorescence (mIF) image in colorectal adenocarcinoma. b Total immune cell densities in the CRC and PDAC. c Distribution of overall (combined tumor epithelial and stromal compartments) immune cell subsets densities. d Immune cell subsets densities in 4 up-front resected CRCs and 4 up-front resected PDACs. e Boxplots showing immune cells densities (T cells, cytotoxic T cells, B cells, and macrophages) in tumor and stroma compartments from our cohort. f Boxplots depicting immune cells densities (T cells, cytotoxic T cells, B cells, and macrophages) in tumor and stroma compartments from CRC and PDAC. No statistical significance was found.

Close modal

When comparing immune cell densities between CRC and PDAC, interestingly, we observed higher densities of T cells and macrophages in CRC than in PDAC (Fig. 3b and online suppl. Table 4), but cytotoxic T cells and B cells were more common in PDAC than in CRC (Fig. 3c, d). As we expected, immune cell densities, in particular T cells, were higher in the stromal compartment than in the tumor compartment but this was not statistically significant (Fig. 3e, f and online suppl. Table 5). Although the ADO marker, CD39, we observed expressed by different immune cells including overall T cells, cytotoxic T cells, and low proportions of B cells, and macrophages (Fig. 4b, d, e) the ADO marker CD73 was not observed expressed by cytotoxic T cells, B cells, neither by macrophages in our cohort (Fig. 4e and online suppl. Table 4). Similarly, CD39 was observed in low densities by tumor epithelial cells but CD73 expression was not identified (Fig. 4a, c). Interestingly, all the samples showed expression of CD39 and CD73 by morphologically characterized as fibroblast, endothelial cells, and other immune cells without cell-typic makers in our mIF panel (Fig. 4c and online suppl. Fig. 3).

Fig. 4.

Cell subsets expressing CD39 and CD73 of up-front resected CRC and PDAC. a Bar plot percentage of tumor epithelial cells expressing CD39 (<1%) in CRC and PDAC. b Violin plot depicting the distribution of overall (combined tumor intraepithelial and stromal areas) immune cell densities expressing CD39 our cohort. c Cell subsets expressing CD39. d Representative microphotographs of marker co-expression, as determined by multiplex immunofluorescence (mIF). e Boxplot of immune cells expressing CD39 and CD73 in CRC and PDAC. No statistical significance was found.

Fig. 4.

Cell subsets expressing CD39 and CD73 of up-front resected CRC and PDAC. a Bar plot percentage of tumor epithelial cells expressing CD39 (<1%) in CRC and PDAC. b Violin plot depicting the distribution of overall (combined tumor intraepithelial and stromal areas) immune cell densities expressing CD39 our cohort. c Cell subsets expressing CD39. d Representative microphotographs of marker co-expression, as determined by multiplex immunofluorescence (mIF). e Boxplot of immune cells expressing CD39 and CD73 in CRC and PDAC. No statistical significance was found.

Close modal

Cellular Spatial Distribution

Using the NND analysis, we observed that most T cells, B cells, and macrophages were localized within 50 μm from the tumor cells (Table 2; Fig. 5a and online suppl. Table 6). Interestingly, overall macrophages were the closest cells to the tumor cells, followed by cytotoxic T cells, T cells, and B cells (online suppl. Fig. 4a). In addition, when comparing CRC and PDAC, we found that cytotoxic T cells were closest to tumor cells in PDAC (median: 22 μm), and macrophages were closest to tumor cells in CRC (median: 25 μm) (Fig. 5a and online suppl. Table 6). Importantly, overall, T cells expressing CD39 were closer to the tumor cells (median, 33 μm) than T cells expressing CD73 (median, 70 μm) (Fig. 5d and online suppl. Fig. 4b, c). On the other hand, tumor epithelial cells expressing CD39 (CK + CD39+) showed closer overall immune cells (median, within 50 μm) when compared with tumor epithelial cells that did not express CD39 (T cells, median 292 μm; cytotoxic T cells, median 126 μm; B cells, median 420 μm; macrophages, median 362 μm) (Fig. 5b, d, online suppl. Fig. 4d; online suppl. Table 7). Compared to CRC and PDAC, according to the same methodology described above, T cells were closest to tumor cells expressing CD39 in PDAC (median, 284 μm), and macrophages were closest to tumor cells expressing CD39 in CRC (median: 295 μm). We also analyzed immune cells expressing CD39 (online suppl. Fig. 4d). Interestingly, these cells are closer to tumor epithelial cells expressing CD39 (CD3+CD39+ cells, median 278 μm; CD20+CD39+ cells, median 259 μm; CD68+CD39+, median 238 μm) than the immune cells that did not express CD39, suggesting that the close interaction of immune cells and tumor epithelial cells expressing the ADO marker CD39 can interfere with the antitumor immune response. Although performed using a small cohort, we explored associations of cell densities and spatial distribution with clinicopathologic variables, finding no significant correlations (online suppl. Fig. 5).

Fig. 5.

Distances from individual immune cells to the closest tumor cell of up-front resected CRC and PDAC. a Boxplot showing distances from individual immune cells to the closest tumor cell. b Boxplot showing distances from individual cytotoxic T cells to the closest tumor cell that did not express CD39 (CK + CD39−) and to the closest tumor cell expressing CD39 (CK + CD39+) – combined CRC and PDAC. c Boxplot showing distances from individual immune cells to the closest tumor cell that did not express CD39 and to the closest tumor cell expressing CD39 in CRC and PDAC. d Median distance heat map representing the immune cells and the most common immune cells expressing CD39 and CD73 in CRC and PDAC.

Fig. 5.

Distances from individual immune cells to the closest tumor cell of up-front resected CRC and PDAC. a Boxplot showing distances from individual immune cells to the closest tumor cell. b Boxplot showing distances from individual cytotoxic T cells to the closest tumor cell that did not express CD39 (CK + CD39−) and to the closest tumor cell expressing CD39 (CK + CD39+) – combined CRC and PDAC. c Boxplot showing distances from individual immune cells to the closest tumor cell that did not express CD39 and to the closest tumor cell expressing CD39 in CRC and PDAC. d Median distance heat map representing the immune cells and the most common immune cells expressing CD39 and CD73 in CRC and PDAC.

Close modal

Markers, including CD39 and CD73, were optimized for an automated seven-color mIF panel and applied to a pilot study involving 4 PDACs and 4 CRCs for internal staining validation and exploratory analysis. Performing our staining optimization using tonsil tissue, three different time points, and their marker staining by IHC showed consistent reproducibility of the mIF panel. The ADO marker CD39 was ubiquitously expressed in various immune cell populations, including overall T cells, B cells, macrophages, and tumor epithelial cells compared to CD79, which was mostly restricted to T cells with no cytotoxic T cells in this cohort. Exploratory image analysis showed closer proximity between immune cells and tumor epithelial cells that do not express CD39 compared to tumor epithelial cells expressing CD39.

Studying the ADO pathway will be a benefit to understanding the tumor microenvironment in solid tumors better, as shown in the recent phase II study, randomized COAST, which offered improved objective response rate and progression-free survival in the combination of oleclumab and durvalumab versus durvalumab alone in patients with locally advanced, unresectable, stage III NSCLC who had not progressed following cCRT [25] or ongoing studies utilizing oleclumab, including studies in NSCLC (MAGELLAN, NCT03819465; HUDSON, NCT03334617) or pancreatic cancer (NCT03611556). Therefore, we have developed an mIF panel that can be used effectively to better understand the role of the ADO pathway in the tumor microenvironment.

In the current study, we first characterized the tumor microenvironment of the resected tumors. As part of the panel validation process, we statistically analyzed the quantified markers expression using conventional IHC and mIF cell density results at three-time points, showing the reliability of our technique.

We developed a seven-color TSA-based mIF panel using Opal fluorophores with simultaneous staining on a single paraffin tissue section [38, 40‒43], evaluating single conventional IHC and IF stains. This approach uses TSA reagents, significantly increasing sensitivity while maintaining specificity and resolution, allowing detection in low-expression targets [38, 40‒43]. However, it is essential to select the areas that can represent the whole tumor section [44]. In the optimization process, cross-talking reactions or umbrella effects should be minimized or eliminated [34, 37].

Our quality control of marker expression by IHC and mIF showed the same histologic patterns – CK as an epithelial marker; T-cell markers such as CD3 and CD8; CD20 as B-cell markers; and CD68, CD39, and CD73 as macrophages showed the same staining patterns and expression by IHC and mIF in control tissues.

In our cohort, the cell phenotypes analyzed exhibited high correlation across samples (batch 1 vs. batch 2 vs. batch 3). We demonstrate that this method of simultaneous detection of multiple markers with a nuclear counterstain in a single tissue section is scalable, reproducible, and allows for high-quality results.

Our exploratory mIF analysis showed a TME with a predominance of CD39-positive in the stromal compartment in both PDAC and CRC, as well as its interaction with the different subpopulations of immune cells. Overall, our small cohort of CRC and PDAC tumors did not show high densities of immune cells, with CRC showing higher densities than PDAC, as has already been shown in several studies. As expected, the most observed immune cells in CRC were T cells and macrophages. In most tumors, macrophages are one of the most abundant immune cells and the prognostic impact may be also associated with patient survival, such as in CRC [45‒47]. Macrophages can also promote growth, invasion, and metastatic potential of CRC [45, 48]. Interestingly, B cells and cytotoxic T cells were more common in PDAC. Recent studies have shown that PDAC is a tumor highly infiltrated by B cells, and B cells play a role in the development of PDAC [49]. In addition, when we compared the densities between the tumor and the stromal compartment, a higher density was observed in the stroma with a more pronounced difference in T cells, as expected.

We found that the frequency of CD39 T cells was higher among immune cells. In this study, we found that macrophages, cytotoxic T cells, and B cells express CD39, as previously shown in other studies [3, 14, 15]. Furthermore, it has previously been shown that the expression of CD39 in tumor-associated macrophages promotes CD8+ T-cell dysfunction in cooperation with CD73 in glioblastoma [50]. ADO also modulates B-cell responses, and recently the hypothesis that B cells expressing CD39 and CD73 have regulatory properties has been emerging [51]. In contrast, CD73 expressed by macrophages and B cells was not identified. The fact that our mIF panel did not identify tumor epithelial cells, macrophages, or B cells expressing CD73 in this pilot study is not surprising. Most studies discussing macrophages or B cells expressing CD73 focus on peripheral immune cells and not tumor-based studies [52].

Immune cell spatial configuration concerning tumor cells has been associated with antitumor immune response and patient outcomes. Characterizing the spatial organization may be potentially relevant to guide the use of available therapies that target the ADO pathway. We found that macrophages and cytotoxic T cells were closer to tumor cells in CRC and PDAC, respectively. Proximity of macrophages to tumor epithelial cells (as opposed to stroma) has been associated with improved survival [53]. Furthermore, the aggregation of macrophages within the nest of tumor epithelial cells may have a beneficial effect in terms of accumulation of cytotoxic T cells [53]. A number of studies have indicated that the accumulation of cytotoxic T cells in proximity to cancer cells correlates with increased patient survival [54, 55]. However, differential effects of T cells in PDAC are dependent on the spatial distribution, type of subpopulation involved, and accompanying macrophage infiltration [55]. Importantly, immune T cells expressing CD39 were also closer to tumor cells than T cells expressing CD73.

Interestingly, we also found that immune cells were closer to tumor epithelial cells that did not express CD39 than tumor epithelial cells expressing CD39. These results suggest CD39 may alter spatial relationships between immune and tumor cells. Further investigation will be required to determine the consequence of this pathway in immune cell spatial distribution.

Our study has several limitations. While our mIF assay offers high sensitivity and highly quantitative results for characterizing immune cells in fixed tissue, our pilot study was applied to a limited cohort of patients. Another limitation was the fact that our tumors did not show a high density of immune cells, potentially limiting our results. Finally, we did not include non-neoplastic tissue for comparative analysis.

In conclusion, mIF allows a quantitative analysis of density and spatial distribution of simultaneous markers to identify specific immune cell subsets in different tumor compartments and can provide important information for translational pathology studies. Our mIF panel was designed to analyze markers specifically selected to explore the ADO pathway and enable phenotyping of tumor cells and immune cell types expressing CD39 and/or CD73, thereby overcoming limitations of IHC studies. Furthermore, we demonstrate that the ADO pathway can modulate multiple aspects of the tumor immune microenvironment. In accordance with previous findings, our study suggests that blockage of this pathway could become an important therapeutic target for colorectal adenocarcinoma and PDAC.

The authors would like to acknowledge the Multiplex Immunoflourescence and Digital Pathology Laboratory members from the Translational Molecular Pathology Department, who contribute daily to quality multiplex IF and IHC processing. The manuscript was edited by Ann Suton of the Research Medical Library at The University of Texas MD Anderson Cancer Center.

This study was approved by the Institutional Review Board of The University of Texas MD Anderson Cancer Center (IRB number 2020-0561) and was conducted according to the principles of the Helsinki Declaration. This study reviewed data collected from patients as part of routine standard of care; no diagnostic or therapeutic interventions were performed, and no patient contact was involved. Patient consent was not required in accordance with local or national guidelines.

Jaime Rodriguez-Canales, Philip Martin, and Zachary Cooper declare that they are employees of and own stock in AstraZeneca. The authors declare that this research project was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Cibelle Lima, Auriole Tamegnon, Saxon Rodriguez, Dipen Maru, and Edwin Parra declare no conflict of interest.

This study was supported in part by Translational Molecular Pathology Department of The University of Texas MD Anderson Cancer Center CIMAC and financial support for the AstraZeneca.

C.F.L. wrote most of the manuscript and performed the image analysis. A.T. and S.R. contributed with the staining and scanning. J.R.C. and E.R.P. conceived the idea of the manuscript, and E.R.P. developed the technology in the laboratory. All the authors, C.F.L., A.T., S.R., D.M., P.L.M., Z.A.C., J.R.C., and E.R.P. edited the manuscript according to his experience.

All data generated or analyzed during this study are included in this article and its online supplementary material. Furter inquiries can be directed to the corresponding author.

1.
Chen Q, Pu N, Yin H, Zhang J, Zhao G, Lou W, et al. CD73 acts as a prognostic biomarker and promotes progression and immune escape in pancreatic cancer. J Cell Mol Med. 2020;24(15):8674–86.
2.
Roh M, Wainwright DA, Wu JD, Wan Y, Zhang B. Targeting CD73 to augment cancer immunotherapy. Curr Opin Pharmacol. 2020;53:66–76.
3.
Park HJ, Seo EH, Piao L, Park ST, Lee MK, Koh SE, et al. The preventive effect of the phenotype of tumour-associated macrophages, regulated by CD39, on colon cancer in mice. Int J Mol Sci. 2021;22(14):7478.
4.
Perrot I, Michaud HA, Giraudon-Paoli M, Augier S, Docquier A, Gros L, et al. Blocking antibodies targeting the CD39/CD73 immunosuppressive pathway unleash immune responses in combination cancer therapies. Cell Rep. 2019;27(8):2411–25.e9.
5.
Young A, Ngiow SF, Barkauskas DS, Sult E, Hay C, Blake SJ, et al. Co-inhibition of CD73 and A2AR adenosine signaling improves anti-tumor immune responses. Cancer Cell. 2016;30(3):391–403.
6.
Regateiro FS, Cobbold SP, Waldmann H. CD73 and adenosine generation in the creation of regulatory microenvironments. Clin Exp Immunol. 2013;171(1):1–7.
7.
Antonioli L, Blandizzi C, Pacher P, Haskó G. Immunity, inflammation and cancer: a leading role for adenosine. Nat Rev Cancer. 2013;13(12):842–57.
8.
Timperi E, Barnaba V. CD39 regulation and functions in T cells. Int J Mol Sci. 2021;22(15):8068.
9.
Di Virgilio F, Sarti AC, Falzoni S, De Marchi E, Adinolfi E. Extracellular ATP and P2 purinergic signalling in the tumour microenvironment. Nat Rev Cancer. 2018;18(10):601–18.
10.
Li XY, Moesta AK, Xiao C, Nakamura K, Casey M, Zhang H, et al. Targeting CD39 in cancer reveals an extracellular ATP and inflammasome driven tumor immunity. Cancer Discov. 2019;9(12):1754–73.
11.
Mayer A, Haist M, Loquai C, Grabbe S, Rapp M, Roth W, et al. Role of hypoxia and the adenosine system in immune evasion and prognosis of patients with brain metastases of melanoma: a multiplex whole slide immunofluorescence study. Cancers. 2020;12(12):3753.
12.
Beavis PA, Stagg J, Darcy PK, Smyth MJ. CD73: a potent suppressor of antitumor immune responses. Trends Immunol. 2012;33(5):231–7.
13.
Tahkola K, Ahtiainen M, Kellokumpu I, Mecklin JP, Laukkarinen J, Laakkonen J, et al. Prognostic impact of CD73 expression and its relationship to PD-L1 in patients with radically treated pancreatic cancer. Virchows Arch. 2021;478(2):209–17.
14.
Timperi E, Barnaba V. CD39 regulation and functions in T cells. Int J Mol Sci. 2021;22(15):8068.
15.
Bono MR, Fernández D, Flores-Santibáñez F, Rosemblatt M, Sauma D. CD73 and CD39 ectonucleotidases in T cell differentiation: beyond immunosuppression. FEBS Lett. 2015;589(22):3454–60.
16.
Yang H, Yao F, Davis PF, Tan ST, Hall SR. CD73, tumor plasticity and immune evasion in solid cancers. Cancers. 2021;13(2):177.
17.
Messaoudi N, Cousineau I, Arslanian E, Henault D, Stephen D, Vandenbroucke-Menu F, et al. Prognostic value of CD73 expression in resected colorectal cancer liver metastasis. Oncoimmunology. 2020;9(1):1746138.
18.
Wu XR, He XS, Chen YF, Yuan RX, Zeng Y, Lian L, et al. High expression of CD73 as a poor prognostic biomarker in human colorectal cancer. J Surg Oncol. 2012;106(2):130–7.
19.
Harvey JB, Phan LH, Villarreal OE, Bowser JL. CD73’s potential as an immunotherapy target in gastrointestinal cancers. Front Immunol. 2020;11:508.
20.
Hu G, Wu P, Cheng P, Zhang Z, Wang Z, Yu X, et al. Tumor-infiltrating CD39+ γδTregs are novel immunosuppressive T cells in human colorectal cancer. Oncoimmunology. 2017;6(2):e1277305.
21.
Zhao J, Soto LMS, Wang H, Katz MH, Prakash LR, Kim M, et al. Overexpression of CD73 in pancreatic ductal adenocarcinoma is associated with immunosuppressive tumor microenvironment and poor survival. Pancreatology. 2021;21(5):942–9.
22.
Chen S, Wainwright DA, Wu JD, Wan Y, Matei DE, Zhang Y, et al. CD73: an emerging checkpoint for cancer immunotherapy. Immunotherapy. 2019;11:983–97.
23.
Allard B, Longhi MS, Robson SC, Stagg J. The ectonucleotidases CD39 and CD73: novel checkpoint inhibitor targets. Immunol Rev. 2017;276(1):121–44.
24.
Baghbani E, Noorolyai S, Shanehbandi D, Mokhtarzadeh A, Aghebati-Maleki L, Shahgoli VK, et al. Regulation of immune responses through CD39 and CD73 in cancer: novel checkpoints. Life Sci. 2021;282:119826.
25.
Herbst RS, Majem M, Barlesi F, Carcereny E, Chu Q, Monnet I, et al. COAST: an open-label, phase II, multidrug platform study of durvalumab alone or in combination with oleclumab or monalizumab in patients with unresectable, stage III non-small-cell lung cancer. J Clin Oncol. 2022;40(29):3383–93.
26.
Beavis PA, Divisekera U, Paget C, Chow MT, John LB, Devaud C, et al. Blockade of A 2A receptors potently suppresses the metastasis of CD73+ tumors. Proc Natl Acad Sci USA. 2013;110(36):14711–6.
27.
Hajizadeh F, Masjedi A, Heydarzedeh Asl S, Karoon Kiani F, Peydaveisi M, Ghalamfarsa G, et al. Adenosine and adenosine receptors in colorectal cancer. Int Immunopharmacol. 2020;87:106853.
28.
Parra ER. Immune cell profiling in cancer using multiplex immunofluorescence and digital analysis approaches. In: Streckfus CF, editor. Immunohistochemistry. IntechOpen; 2018. 1–13.
29.
Taube JM, Akturk G, Angelo M, Engle EL, Gnjatic S, Greenbaum S, et al. The Society for Immunotherapy of Cancer statement on best practices for multiplex immunohistochemistry (IHC) and immunofluorescence (IF) staining and validation. J Immunother Cancer. 2020;8(1):e000155.
30.
Parra ER, Francisco-Cruz A, Wistuba II. State-of-the-art of profiling immune contexture in the era of multiplexed staining and digital analysis to study paraffin tumor tissues. Cancers. 2019;11(2):247.
31.
Parra ER. Novel platforms of multiplexed immunofluorescence for study of paraffin tumor tissues. J Cancer Treat Diagn. 2017;2(1):43–53.
32.
Parra ER, Ferrufino-Schmidt MC, Wistuba II, Tamegnon A, Zhang J, Solis L, et al. Immuno-profiling and cellular spatial analysis using five immune oncology multiplex immunofluorescence panels for paraffin tumor tissue. Sci Rep. 2021:11–8511.
33.
Parra ER, Uraoka N, Jiang M, Cook P, Gibbons D, Forget MA, et al. Validation of multiplex immunofluorescence panels using multispectral microscopy for immune-profiling of formalin-fixed and paraffin-embedded human tumor tissues. Sci Rep. 2017;7(1):13380.
34.
Parra ER, Jiang M, Solis L, Mino B, Laberiano C, Hernandez S, et al. Procedural requirements and recommendations for multiplex immunofluorescence tyramide signal amplification assays to support translational oncology studies. Cancers (Basel). 2020;12(2):255.
35.
Francisco-Cruz A, Parra ER, Tetzlaff MT, Wistuba II. Multiplex immunofluorescence assays. Methods Mol Biol. 2020; 2055:467–95.
36.
Gorris MAJ, Halilovic A, Rabold K, van Duffelen A, Wickramasinghe IN, Verweij D, et al. Eight-color multiplex immunohistochemistry for simultaneous detection of multiple immune checkpoint molecules within the tumor microenvironment. J Immunol. 2018;200(1):347–54.
37.
Bacia K, Petrášek Z, Schwille P. Correcting for spectral cross-talk in dual-color fluorescence cross-correlation spectroscopy. Chemphyschem. 2012;13(5):1221–31.
38.
Rocha P, Salazar R, Zhang J, Ledesma D, Solorzano JL, Mino B, et al. CD73 expression defines immune, molecular, and clinicopathological subgroups of lung adenocarcinoma. Cancer Immunol Immunother. 2021;70(7):1965–76.
39.
Lazcano R, Barreto CM, Salazar R, Carapeto F, Traweek RS, Leung CH, et al. The immune landscape of undifferentiated pleomorphic sarcoma. Front Oncol. 2022;12:1008484.
40.
Yu M, Guo G, Huang L, Deng L, Chang CS, Achyut BR, et al. CD73 on cancer-associated fibroblasts enhanced by the A2B-mediated feedforward circuit enforces an immune checkpoint. Nat Commun. 2020;11(1):515.
41.
Zhou L, Jia S, Chen Y, Wang W, Wu Z, Yu W, et al. The distinct role of CD73 in the progression of pancreatic cancer. J Mol Med Berl. 2019;97(6):803–15.
42.
Surace M, DaCosta K, Huntley A, Zhao W, Bagnall C, Brown C, et al. Automated multiplex immunofluorescence panel for immuno-oncology studies on formalin-fixed carcinoma tissue specimens. J Vis Exp. 2019;143.
43.
Viratham Pulsawatdi A, Craig SG, Bingham V, McCombe K, Humphries MP, Senevirathne S, et al. A robust multiplex immunofluorescence and digital pathology workflow for the characterisation of the tumour immune microenvironment. Mol Oncol. 2020;14(10):2384–402.
44.
Sun B, Laberiano-Fernández C, Salazar-Alejo R, Zhang J, Solorzano Rendon JL, Lee J, et al. Impact of region-of-interest size on immune profiling using multiplex immunofluorescence tyramide signal amplification for paraffin-embedded tumor tissues. Pathobiology. 2023;90(1):1–12.
45.
Wang H, Tian T, Zhang J. Tumor-associated macrophages (TAMs) in colorectal cancer (CRC): from mechanism to therapy and prognosis. Int J Mol Sci. 2021;22(16):8470.
46.
Väyrynen JP, Haruki K, Lau MC, Väyrynen SA, Zhong R, Dias Costa A, et al. The prognostic role of macrophage polarization in the colorectal cancer microenvironment. Cancer Immunol Res. 2021;9(1):8–19.
47.
Christofides A, Strauss L, Yeo A, Cao C, Charest A, Boussiotis VA. The complex role of tumor-infiltrating macrophages. Nat Immunol. 2022;23(8):1148–56.
48.
Zhang Y, Zhao Y, Li Q, Wang Y. Macrophages, as a promising strategy to targeted treatment for colorectal cancer metastasis in tumor immune microenvironment. Front Immunol. 2021;12:685978.
49.
Minici C, Testoni S, Della-Torre E. B-lymphocytes in the pathophysiology of pancreatic adenocarcinoma. Front Immunol. 2022;13:867902.
50.
Takenaka MC, Gabriely G, Rothhammer V, Mascanfroni ID, Wheeler MA, Chao CC, et al. Control of tumor-associated macrophages and T cells in glioblastoma via AHR and CD39. Nat Neurosci. 2019;22(5):729–40.
51.
Figueiró F, Muller L, Funk S, Jackson EK, Battastini AMO, Whiteside TL. Phenotypic and functional characteristics of CD39high human regulatory B cells (Breg). Oncoimmunology. 2016;5(2):e1082703.
52.
Feng LL, Cai YQ, Zhu MC, Xing LJ, Wang X. The yin and yang functions of extracellular ATP and adenosine in tumor immunity. Cancer Cel Int. 2020;20:110.
53.
Reis-Sobreiro M, Teixeira da Mota A, Jardim C, Serre K. Bringing macrophages to the frontline against cancer: current immunotherapies targeting macrophages. Cells. 2021;10(9):2364.
54.
Raskov H, Orhan A, Christensen JP, Gögenur I. Cytotoxic CD8+ T cells in cancer and cancer immunotherapy. Br J Cancer. 2021;124(2):359–67.
55.
Carstens JL, Correa de Sampaio P, Yang D, Barua S, Wang H, Rao A, et al. Spatial computation of intratumoral T cells correlates with survival of patients with pancreatic cancer. Nat Commun. 2017;8:15095.