Introduction

Mucinous adenocarcinomas are defined by pools of extracellular mucin that exceed 50% of tumor volume. Tumors with lesser amounts of mucin are often referred to as adenocarcinomas with mucinous differentiation. The studies that address the influence of mucinous differentiation on survival report contradictory results, although an impact that is independent of other key prognostic factors such as AJCC stage, extramural venous invasion, and perineural invasion has not been demonstrated1,2,3,4,5. In the metastatic setting the prognosis of patients with mucinous adenocarcinoma is worse than conventional adenocarcinomas6,7,8. The current guideline that require >50% extracellular mucin for a diagnosis of mucinous adenocarcinoma is arbitrary, and there is limited data on the prognostic impact of lesser amounts of mucin9.

Histologic grading of colonic mucinous adenocarcinomas has evolved over the last two iterations of the WHO guidelines. The 2010 guidelines state that proficient mismatch repair (pMMR) mucinous adenocarcinomas behave as high-grade tumors10. However, the most recent edition of WHO guidelines11 recommend grading based on the extent of glandular differentiation, akin to non-mucinous adenocarcinomas: cases showing ≤50% gland formation are considered high grade. There is only limited data to support either of these guidelines12,13,14.

Deficient (d)MMR colon carcinomas are preferentially associated with mucinous differentiation15. However, there is only limited data regarding the immune phenotype of mucinous adenocarcinomas. In one analysis, tumor-infiltrating lymphocytes, as determined on a hematoxylin and eosin-stained slide, was a strong independent predictor of disease-free survival in mucinous adenocarcinoma16. In this study, compared with non-mucinous tumors, mucinous adenocarcinomas were associated with increased tumor infiltrating lymphocytes. However, there is only limited data regarding key lymphocyte subsets or other therapeutically relevant immune markers in mucinous adenocarcinoma17.

A significant proportion of dMMR colon carcinomas respond favorably to immune checkpoint inhibitor therapy targeting programmed cell death 1 (PD1) and cytotoxic T lymphocyte antigen 4 (CTLA4), however, pMMR carcinomas are generally unresponsive to current checkpoint therapy18,19. The increased number of CD8 positive cells and high mutational burden is believed to account for the favorable response of dMMR tumors to immunotherapy20. However, not all dMMR colonic tumors show favorable response to checkpoint inhibitors and the mechanisms underlying this resistance are largely unknown.

Herein, using a large cohort of resected colonic carcinomas we address 3 unresolved questions regarding mucinous adenocarcinoma: (1) the differences in key clinicopathologic-molecular features and oncological outcomes between mucinous adenocarcinomas, adenocarcinomas with mucinous differentiation and conventional adenocarcinomas, (2) grading of mucinous adenocarcinomas, and (3) immunological differences between (a) mucinous adenocarcinomas and conventional adenocarcinomas and, (b) tumors with ≤10% mucin and >10% mucin.

Materials and methods

We evaluated 903 consecutive patients with colon carcinomas resected at the Massachusetts General Hospital. We identified 62 (6.9%) mucinous adenocarcinomas. Signet-ring carcinoma and patients who had received neoadjuvant chemoradiation were excluded. We also excluded medullary carcinomas and neuroendocrine carcinomas. Two tumor locations were recognized: right (cecum to splenic flexure) and left (splenic flexure to rectum). The study was approved by the hospital IRB.

Morphological and immunohistochemical analysis

The slides were reviewed by two pathologist (VD and AC) who were not aware of clinical, outcome, and molecular data. Mucinous differentiation was scored in increments of 5%. Based on this analysis we identified three classes: (1) mucinous adenocarcinoma with >50% mucin, (2) adenocarcinoma with mucinous differentiation with 10–50% mucin, and (3) conventional adenocarcinoma with less than 10% mucin. Tumor grading was performed as per WHO guidelines (≤50% vs. >50% glandular forming area) (Fig. 1)11. We also recorded the AJCC stage, perineural and extramural venous invasion. We also recorded MMR status based on immunohistochemistry for MSH2, MSH6, MLH1, and PMS2.

Fig. 1: Colon carcinoma with mucin.
figure 1

Mucinous adenocarcinoma (A, B). Extracellular mucin occupied >90% of this tumor. Adenocarcinoma with mucinous differentiation (C, D). Extracellular mucin occupied 10% of this neoplasm.

Full size image

Additional immunohistochemistry was performed on tissue microarrays. The central portion of each tumor was represented by a 0.2 cm core of tissue. We specifically avoided areas with mucin. Detailed information about the clone, dilution, type of antibody, and company are listed in Table 1. Tumor samples were analyzed for BRAF (V600E) mutations using an immunohistochemical platform. We have previously demonstrated high concordance of BRAF V600E as assessed by sequencing21.

Table 1 Immunohistochemical assays used in this study.
Full size table

The monoclonal antibody HC-10 recognizes b2m-free HLA-A3, -A10, -A28, -A29, -A30, -A31, -A32, -A33, and all b2m-free -HLA-B (excluding -B5702, -B5804, and -B73) and -HLA-C heavy chains and is referred to as antibody targeting Class I HLA in this study. The monoclonal antibody LGII-612.14 recognizes a monomorphic epitope expressed on the b chain of HLA DR, -DQ, and -DP antigens, and is referred to as antibody targeting Class II antigens. Heat-induced epitope-retrieval (HIER) was used prior to immunostaining. The sections were incubated in 10 mM sodium-citrate (pH6.0) or 10 mM Tris (pH9.0) buffered solutions containing 0.05% Tween at 125 °C for 5 min using a decloaking chamber (Biocare Medica). Staining was performed on the automated LabVision Autostainer 360 (Thermo Scientific) platform using a secondary ImmPRESS polymer detection system (Vector Laboratories) and horseradish-peroxidase-conjugated rabbit anti-sheep secondary antibodies (Thermo Scientific), according to the manufacturers protocols. The Vulcan Fast Red Chromogen Kit 2 (red staining; Biocare Medical) or the DAB Quanto System (brown staining; Thermo Scientific) were applied as substrates. With every batch, a known tissue-positive control was used.

Automated quantification

The stained TMA slides were scanned using Aperio ScanScope digital slide scanner (Aperio ScanScope CSO, Leica Biosystems Imaging, CA) and automated quantification of CD8, CD163, and FoxP3 positive cells was performed using HALO image analysis platform (HALO 2.3; Indica Laboratories, NM). Tumor area within each tissue core was manually annotated. The number of positive CD8, CD163 and FoxP3 cells were calculated and expressed per mm2.

Manual quantification (Semiquantitative)

The percentage of tumor cells staining for HLA class 1, HLA class II, beta-2-microglobulin and PD-L1 was recorded in increments of 5% on the TMA slides. We also manually quantitated the number of intratumoral PD-L1 positive inflammatory cells and the results were expressed per mm2.

Statistical analysis

Follow-up duration was calculated from the time of operation to the time of death or last follow-up. Survival curves were plotted using the Kaplan–Meier method. Differences in overall survival between groups were analyzed by the log-rank test. Multivariate survival analyses were performed using the backward conditional Cox regression method. Comparison of categorical variables among groups was performed using chi-square test and Fisher exact test. Pair-sample t-test was used to compare differences between immune subsets and mucinous differentiation.

All analyses were performed using SPSS version 21. A p value of < 0.05 was considered statistically significant.

Results

Patient characteristics

All patients underwent surgical resection between 2004 and 2014. Of the 903 patients, 50.5% were female, and the median age was 68.3 years (range 26 to >90). 52.7% (455/864) involved the right colon. 16.6% (147–887) were classified as high-grade adenocarcinomas. No patient received neoadjuvant therapy. The median follow-up duration was 53 months (SD 42.8, range 1–170).

MMR immunohistochemistry was performed on 793 patients. 20.3% (161/793) of adenocarcinomas were classified as dMMR. On immunohistochemistry. Tumors with dMMR status were more often positive for BRAF reactivity (p = 0.0001).

42.6% (26/61) of mucinous adenocarcinoma and 38.9% (327 of 841) of conventional adenocarcinoma patients received adjuvant therapy (p = 0.88).

Consistent with prior literature, tumor grade (p = 0.0001), AJCC stage (p = 0.0001), extramural venous invasion (p = 0.0001), small vessel invasion (p = 0.0001), and perineural invasion (p = 0.0001) correlated with disease specific survival on univariate analysis. However, MMR status did not correlate with disease specific survival (P = 0.12).

On multivariate analysis, AJCC stage (p = 0.001), extramural venous invasion (p = 0.001) and perineural invasion (p = 0.006) correlated with disease specific survival.

Clinical outcome of patients classified as mucinous adenocarcinomas and adenocarcinoma with mucinous component

To assess the clinical significance of varying degrees of mucinous differentiation, we evaluated the relationship between the quantity of mucin and disease-specific survival. Disease-specific survival of tumors classified as mucinous adenocarcinoma, although lower, did not differ from conventional adenocarcinomas (estimated mean survival 119 months vs 133.2 months, respectively, p = 0.08) (Fig. 2). No difference in disease-specific survival was noted at a lower cut-off at 10% (estimated mean survival for tumors with ≥10% mucin 122.9 months vs tumors with <10% mucin 134.5 months, p = 0.08) (Fig. 2).

Fig. 2: Kaplan–Meier curves.
figure 2

The figure highlighs the lack of difference in survival of patients with mucinous adenocarcinoma (A) and tumors with 10% mucin or more (B).

Full size image

In this study we further categorized colonic adenocarcinomas into 3 groups: (1) Conventional adenocarcinoma <10% mucin (n = 675), (2) adenocarcinomas with mucinous differentiation with 10–50% mucin (n = 166), and (3) mucinous adenocarcinomas with >50% mucin (n = 62). There was no statistically significant difference in disease-specific survival between adenocarcinomas with mucinous differentiation and conventional adenocarcinomas (p = 0.4)

Clinical, pathologic, and molecular characteristics of mucinous adenocarcinomas

Mucinous adenocarcinomas were more often classified as high-grade and localized to the right side (Table 2). Compared to conventional adenocarcinomas, mucinous adenocarcinomas more often affected women (58.1% vs 48.9%), although this difference was not statistically significant. Similarly, mucinous adenocarcinomas were more often positive for BRAF (27.1% vs 15.5%). Of note, there was no difference in the TNM stage between dMMR mucinous and dMMR non-mucinous adenocarcinomas (p = 0.37).

Table 2 Comparison of the clinical, molecular and immunohistochemical features of mucinous adenocarcinoma, adenocarcinoma with mucinous component and conventional colonic adenocarcinomas.
Full size table

Clinical, pathologic and molecular characteristics of adenocarcinoma with mucinous differentiation

There were no statistically significant differences in key clinical, histological, and molecular features between mucinous adenocarcinoma and adenocarcinoma with mucinous differentiation (Table 2). Of note, there was no difference in the TNM stage between dMMR tumors with ≥10% mucin and dMMR tumor with <10% mucin (p = 0.98).

Grading and MMR status

High-grade mucinous adenocarcinomas showed shorter disease-specific survival (estimated survival 84.9 months) than low grade mucinous adenocarcinomas (estimated survival 121 months) (p = 0.01). On the other hand, there was no difference between disease-specific survival between dMMR and pMMR mucinous adenocarcinomas (p = 0.86).

A similar pattern was seen with adenocarcinomas with mucinous differentiation (i.e. 10% cut off). When compared to low-grade adenocarcinoma with mucinous differentiation, high-grade adenocarcinomas with mucinous differentiation showed decreased survival (p = 0.0001). dMMR adenocarcinoma with mucinous differentiation showed longer disease-specific survival than pMMR patients with mucinous differentiation (estimated mean survival 143.1 months versus 130.1 months, respectively), however, this difference was not statistically significant (p = 0.12).

We also evaluated survival statistics in non-mucinous carcinomas. Although pMMR non-mucinous adenocarcinomas showed a shorter disease-specific survival (estimated mean survival 130.8 months) than dMMR non-mucinous adenocarcinomas (estimated mean survival 145.4 months), this difference was not statistically significant (p = 0.08).

Correlation with CD8

Given the ability of immunoscore to independently predict outcome in colonic adenocarcinomas22,23, we quantitated the number of CD8 positive cells across the entire cohort of 903 patients (Fig. 3A–D). As predicted, dMMR tumors showed higher numbers of CD8 positive cells (p = 0.0001) than pMMR tumors. We used the 25th percentile of CD8 count to define two subsets of tumors: 1) those with fewer than 94 CD8 cells per mm2, and 2) tumors with ≥ 94 CD8 cells per mm2. When we evaluated the entire cohort, colonic adenocarcinomas with fewer than 94 CD8 positive cells showed significantly worse disease-specific survival (p = 0.0001). Although mucinous adenocarcinomas showed fewer CD8+ cells compared to conventional adenocarcinomas (338.6 versus 408.7 per mm2) this difference was not statistically significant (p = 0.17) (Table 3); a similar pattern was noted when tumors were classified based on the 10% mucin cut point (Table 3).

Fig. 3: Immune phenotype of colon carcinoma.
figure 3

Mucinous (A) and non-mucinous adenocarcinoma (B). The mucinous adenocarcinoma showed large pools of extracellular mucin (not shown). Representative immunohistochemical images for mucinous adenocarcinomas are shown on the left (C, E, G) and those for non-mucinous adenocarcinoma are shown on the right (D, F, H, I). Immunohistochemistry for CD8 (C, D), PD-L1 (E, F), and CD163 (G, H).

Full size image
Table 3 Comparison of intratumoral immune cells and HLA complex protein expression with mucin.
Full size table

Correlation of other immune markers with mucinous adenocarcinoma

Mucinous adenocarcinomas also showed fewer PD-L1 positive immune cells (Fig. 3E, F), although there was no difference in the number of CD163 (Fig. 3G, H) or FoxP3 positive cells. Notably, there was no difference between mucinous and conventional adenocarcinomas regarding expression of HLA complex proteins (HLA class I, class II and Beta-2-microglobulin); similarly, there was no correlation with PD-L1 expressed on tumor cells (Table 3). Similar results were seen when tumors with <10% mucin were compared to tumors with ≥10% mucin (Table 3).

Correlation between immune milieu of mucinous adenocarcinoma when stratified by MMR status

To further evaluate the immunological milieu of mucinous adenocarcinomas, we stratified tumors based on MMR status. dMMR mucinous adenocarcinomas showed fewer CD8, CD163, and PD-L1 positive immune cells than pMMR conventional adenocarcinoma (Table 4). Notably, no such differences were identified between pMMR mucinous and pMMR conventional adenocarcinomas (Table 4). These differences in dMMR tumors, and the lack of such differences, with pMMR tumors (with the exception of PD-L1 immune cells), were also noted when tumors with <10% mucin were compared to tumors with ≥10% mucin (Table 4).

Table 4 Analysis of dMMR tumors and pMMR tumors. The table compares intratumoral immune cells and tumor cell HLA complex protein expression with mucin.
Full size table

Remarkably, all dMMR mucinous adenocarcinomas were negative for PD-L1 on tumor cells. dMMR tumors also showed higher expression of HLA class I and class II expression.

Discussion

Based on this study evaluating the immune milieu in 62 mucinous and 841 consecutive conventional colonic adenocarcinomas we draw 4 broad conclusions:

  1. 1.

    Concordant with several prior studies the diagnosis of mucinous adenocarcinoma does not predict disease-specific survival. Based on survival we could not validate a cut point for a diagnosis of mucinous adenocarcinoma, although based on the impact of mucin on the immune milieu at a cut point as low as 10% our data could support the redefinition of this variant of colonic adenocarcinoma.

  2. 2.

    The study supports grading of mucinous adenocarcinomas based on gland formation and not MMR status.

  3. 3.

    There was no difference in the number of CD8 positive cells between mucinous and non-mucinous adenocarcinomas

  4. 4.

    dMMR mucinous adenocarcinomas showed fewer CD8, CD163, and PD-L1 positive immune cells than dMMR conventional adenocarcinomas. These differences were not observed in the pMMR cohort. Similar differences were observed when we compared tumors with <10% mucin with tumors that showed ≥10% mucin. These findings raise the possibility that the immune milieu of dMMR mucinous adenocarcinomas differs from dMMR conventional adenocarcinomas and that lesser quantities of mucin may alter the intratumoral immune environment

In this study, mucinous adenocarcinomas did not correlate with statistically significant inferior survival and a lower cut point (10% extracellular mucin) showed similar results. The impact of the mucinous adenocarcinomas subtype on outcome remains controversial. Regardless, there appears to be consensus that the mucinous subtype is not an independent predictor of outcome. A meta-analysis published in 2012 showed that mucinous adenocarcinomas showed a slightly worse survival rate as compared to non-mucinous carcinomas (HR 1.06, 95% CI 1.3–1.1)3, although, no difference was observed on the multivariate portion of the analysis. Interestingly, in the metastatic setting, patients with mucinous adenocarcinoma generally show a worse outcome compared to conventional adenocarcinoma; the median overall survival rates varied from 8–14 months for patients with metastatic mucinous adenocarcinoma compared to 17.9–23.4 months for patients with conventional adenocarcinoma6,7,23.

In this study, we defined 2 subgroups of mucin-producing adenocarcinomas: (1) adenocarcinomas with mucinous differentiation—mucin between 10 and 50%, and (2) mucinous adenocarcinoma with >50% mucin. Interestingly, there were no differences in key demographic, histological and molecular features (Table 2) between these two categories, and the disease-specific survival of adenocarcinomas with mucinous differentiation was identical to mucinous adenocarcinomas.

The grading of mucinous adenocarcinoma has also been debated over the last 2 decades. While some studies have reported improved outcome of dMMR mucinous adenocarcinomas, others have reported contradictory results12,14,16,24,25,26,27. A recent study of 1643 stage II and stage III colonic adenocarcinomas found that dMMR mucinous adenocarcinomas were associated with superior disease-free survival when compared to pMMR mucinous adenocarcinomtas, while tumor grade did not stratify these patients16. Our findings support the use of the traditional ‘gland formation’ based grading for mucinous adenocarcinoma, although we would acknowledge the smaller size of the current cohort compared to two significantly larger series that both support MMR based grading10,16.

The limited prognostic impact of the mucinous adenocarcinoma category on disease-specific survival diminishes the clinical relevance of this subtype. Furthermore, based on current NCCN guidelines, a diagnosis of mucinous adenocarcinoma has limited impact on the therapeutic approach. However, given the success of checkpoint inhibitors in dMMR colon carcinomas, a subset overrepresented in the mucinous adenocarcinoma category28,29, the immunological differences between mucinous adenocarcinoma and conventional adenocarcinoma identified in this study may have therapeutic implications. Of note, there have been no prior attempt to comprehensively assess the immune signature of mucinous adenocarcinomas of the colon.

We identified fewer CD8 positive cells in mucinous adenocarcinomas, although this difference was not statistically significant. We considered potential confounding factors that could account for the decrease in CD8 cells in mucinous adenocarcinomas, specifically displacement of intratumoral stroma and immune cells by mucin. There are however 3 key findings that counter this hypothesis: (1) we choose a non-mucinous core for core for the microarray, (2) these differences were also noted at the 10% cut point, and tissue cores from these patients generally did not include mucin, and (3) In a prior analysis, mucinous adenocarcinoma showed higher numbers of lymphocytes than conventional adenocarcinomas16.

Given the success of PD1 inhibitor in dMMR colon carcinomas, we focused our attention on these tumors that feature high mutational burden and increased numbers of CD8 positive cells. It is notable, that although dMMR tumors tend to show high PD-L1 expression, therapeutic decisions are dictated by MSI and/or MMR status, and checkpoint inhibitors are now approved for tumors with this molecular alteration. Notably, relative to dMMR conventional adenocarcinomas, dMMR mucinous adenocarcinomas showed fewer CD8, CD163 positive, and PD-L1 positive immune cells, and these results were also observed in tumors with as little as 10% mucin. Remarkably, these differences were not observed between pMMR mucinous and pMMR conventional colon adenocarcinomas. We also evaluated the HLA complex, beta-2- microglobulin light chain, and PD-L1 expression on tumor cells to assess other key regulators of the adaptive anti-tumor response. Successful anti-PD-1/PD-L1 therapy, at a minimum, requires the presence of cytotoxic T cells, and expression of HLA class 1 and beta-2-microglobulin on the surface of tumor cells. Of note, mucinous adenocarcinoma showed decreased expression of PD-L1 on tumor cells but upregulation of HLA class I and II proteins.

These findings are supported by the knowledge that mucin forms an important part of the epithelial barrier and is closely tied with the regulation of inflammation, proliferation, and survival of epithelial cells.

Mucin protects epithelial cells by curbing the activation of inflammation under non-pathological conditions. Overexpression of mucin can lead to an overall immunosuppression via mechanisms involving receptor masking, reduced recruitment, inhibition of cytolytic activity thereby hampering the T cell effector response which in turn aids in promoting growth and survival of cancer cells30,31. The loss of expression of mucin 2 (MUC2) in mouse models has shown to increase intestinal inflammation32. This indirect evidence suggests that mucin may have a role in impairing anti-tumor immune response, although alternative mechanisms cannot be excluded.

While we did not specifically identify the cell expressing PD-L1, morphologically, an overwhelming majority of PD-L1 positive cells resembled macrophages. Of note, dMMR mucinous adenocarcinomas showed fewer PD-L1 positive immune cells than dMMR conventional adenocarcinomas. It is now believed that both tumor and immune PD-L1 positive cells are involved in checkpoint-mediated blockade33,34. Lymph node-based PD-L1 expressing antigen-presenting cells can inhibit T cell activation and prevent recruitment of T cells to the tumor microenvironment35. In a mouse model of colon carcinoma, loss of PD-L1 on bone marrow cells but not tumor cells resulted in loss of efficacy to PD-L1 blockade35. Our data emphasizes the need to separately quantify PD-L1 on tumor and tumor immune infiltrate.

A hypermutator phenotype generated by dMMR tumors is expected to recruit a more robust immune response. We hypothesize that although this hypermutator phenotype generates a robust immune response, the mucin dampens the intratumoral immune infiltrate. This hypothesis could explain the combined decrease in CD8, CD163, and PD-L1 positive cells as well as the lack of PD-L1 on dMMR mucinous adenocarcinomas.

There are several limitations of our study that deserve mention. Although the overall size of the cohort is large, since we did not enrich for either dMMR or mucinous adenocarcinomas, these tumor types are relatively underrepresented in this study. However, studies that enrich for dMMR tumors do not account for the bias associated with this analysis. Secondly, although we evaluated two critical molecular parameters, MMR and BRAF, we do not present data of other common genetic alterations in colonic adenocarcinomas. Finally, we used tissue microarrays and not whole sections and thus the study would not account for intratumoral heterogeneity associated with these cells. However, we would emphasize the strong correlation between CD8 and disease-specific survival, a widely recognized marker of outcome in colon carcinoma. Conversely, the strength of this study lies in the large cohort, and extensive immunological phenotyping of these tumors. Another confounding factor that deserves mention is the possibility the mucin may have decreased the number of infiltrating immune cells. However, when creating the tissue microarray, we ensured that mucin is kept to a minimum. We would also highlight vitually identical differences at the 10% cut point. Finally, there were no differences between pMMR mucinous and non-mucinous tumors.

Although the diagnosis of colonic mucinous adenocarcinoma in the non-metastatic setting has limited clinical significance, the immunological differences between mucinous and conventional adenocarcinomas, particularly within the dMMR category, are intriguing and these alterations may influence response to checkpoint inhibitor therapy. We recommend that future clinical trials of checkpoint inhibitor therapy specifically address the impact of mucinous differentiation.