Analysis of the immune landscape of small bowel neuroendocrine tumors

in Endocrine-Related Cancer
Correspondence should be addressed to J Strosberg: jonathan.strosberg@moffitt.org

Immune checkpoint inhibitors have shown promising results in different cancers, and correlation between immune infiltration, expression of programmed death-ligand 1 (PD-L1) by tumor cells and response to immunotherapy has been reported. There is limited knowledge regarding the immune microenvironment of small bowel (SB) neuroendocrine tumors (NETs). This work was aimed at characterizing the immune landscape of SB NETs. Expression of PD-L1 and programmed death-1 (PD-1) was evaluated by immunohistochemistry in 102 surgically resected, primary NETs of the duodenum, jejunum and ileum. Extent and characteristics of the tumor-associated immune infiltrate were also assessed and investigated in their prognostic potential. We detected the expression of PD-L1 in ≥1 and ≥50% of tumor cells in 40/102 (39%; 95% CI, 30–49%) and 14/102 (14%; 95% CI, 8–22%) cases respectively. Intratumor host immune response was apparently absent in 35/102 cases (34%; 95% CI, 25–44%), mild to moderate in 46/102 samples (45%, 95% CI, 35–55%), intense in 21/102 tumors (21%, 95% CI, 13–30%). Expression of PD-L1 and extent of immune infiltration were significantly higher in duodenal NETs as compared with jejunal/ileal NETs. A marked peritumoral host response was organized as ectopic lymph node-like structures in 18/102 cases (18%; 95% CI, 11–26%). Neither PD-L1 expression nor the degree of immune infiltration showed any prognostic significance. Overall, the immune landscape of SB NETs is heterogeneous, with adaptive immune resistance mechanisms prevailing in duodenal NETs. Clinical trials of immune checkpoint inhibitors should take into account the immune heterogeneity of SB NETs.

Abstract

Immune checkpoint inhibitors have shown promising results in different cancers, and correlation between immune infiltration, expression of programmed death-ligand 1 (PD-L1) by tumor cells and response to immunotherapy has been reported. There is limited knowledge regarding the immune microenvironment of small bowel (SB) neuroendocrine tumors (NETs). This work was aimed at characterizing the immune landscape of SB NETs. Expression of PD-L1 and programmed death-1 (PD-1) was evaluated by immunohistochemistry in 102 surgically resected, primary NETs of the duodenum, jejunum and ileum. Extent and characteristics of the tumor-associated immune infiltrate were also assessed and investigated in their prognostic potential. We detected the expression of PD-L1 in ≥1 and ≥50% of tumor cells in 40/102 (39%; 95% CI, 30–49%) and 14/102 (14%; 95% CI, 8–22%) cases respectively. Intratumor host immune response was apparently absent in 35/102 cases (34%; 95% CI, 25–44%), mild to moderate in 46/102 samples (45%, 95% CI, 35–55%), intense in 21/102 tumors (21%, 95% CI, 13–30%). Expression of PD-L1 and extent of immune infiltration were significantly higher in duodenal NETs as compared with jejunal/ileal NETs. A marked peritumoral host response was organized as ectopic lymph node-like structures in 18/102 cases (18%; 95% CI, 11–26%). Neither PD-L1 expression nor the degree of immune infiltration showed any prognostic significance. Overall, the immune landscape of SB NETs is heterogeneous, with adaptive immune resistance mechanisms prevailing in duodenal NETs. Clinical trials of immune checkpoint inhibitors should take into account the immune heterogeneity of SB NETs.

Introduction

Neuroendocrine tumors (NETs) of the small bowel (SB) represent the most common type of gastrointestinal NET (Frilling et al. 2012). The incidence of SB NETs has increased substantially in recent years, and prolonged survival durations have been reported over time as therapeutic options improved (Dasari et al. 2017). Nevertheless, the long-term outcome of patients with stage IV disease remains poor and new treatments including immune checkpoint inhibitors are currently being investigated (Pavel & Sers 2016).

During tumorigenesis, cancer cells exploit a number of immune checkpoint pathways including the programmed death-1 (PD-1)/programmed death-ligand 1 (PD-L1) axis to avoid detection by the adaptive immune system. Checkpoint blockade has demonstrated dramatic antitumor activity in various solid cancers, and tumor PD-L1 expression as well as tumor lymphocyte infiltration have been described as predictors of response (Teng et al. 2015).

Very little is known regarding the immune microenvironment of SB NETs, as well as their expression of immune checkpoint molecules. Circulating antigen-specific CD8+ T cells have been detected in patients with midgut carcinoids (Vikman et al. 2008). However, a decreased proliferative capability has been shown for T cells from NET patients as compared with those from normal controls, probably as result of suppressed levels of systemic Th1-promoting cytokines and increased numbers of immunosuppressive Treg lymphocytes (Vikman et al. 2009). Recently, tumor expression of PD-L1 has been reported in 22% of 32 patients with primarily high-grade, foregut- or hindgut-derived neuroendocrine neoplasms (Kim et al. 2016). In a very heterogeneous cohort of 57 patients with gastroenteropancreatic (GEP) neuroendocrine neoplasms, the expression of PD-L1 was observed in 0, 78 and 100% of G1, G2 and G3 tumors respectively (Cavalcanti et al. 2017). In a series of 37 patients with poorly differentiated digestive neuroendocrine carcinomas (NECs), PD-L1 positivity was demonstrated in 14% of tumors (Roberts et al. 2017). In another study of 62 patients with SB G1/G2 NETs, expression of PD-L1 within tumor cells was detected in 13% of samples, while infiltration of PD-1-positive lymphocytes was observed in 23% of cases (Lamarca et al. 2018)

In order to characterize the immune landscape of primary SB NETs, we investigated the expression of PD-L1 and PD-1 on 102 surgically resected tumors of the duodenum, jejunum and ileum. We also assessed the extent and characteristics of the tumor-associated immune infiltrate, providing the first description of the immune microenvironment of SB NETs.

Patients and methods

Patient selection

Approval for data collection and analysis was obtained from the Institutional Review Board (IRB) of the University of South Florida (Tampa, FL, USA). Patient consent was not required by IRB given the retrospective nature of the study and the fact that the study posed no risk to patients. We retrospectively examined 102 patients with SB NETs whose primary tumors were resected at Moffitt Cancer Center between 2000 and 2015. The following information was collected by review of patient medical records: demographics, date of initial diagnosis, location of primary tumor, initial tumor stage according to the American Joint Committee on Cancer (AJCC) 2010 classification (Rindi et al. 2007), as well as follow-up and survival data. Pathological information including tumor grade by World Health Organization (WHO) criteria (Rindi & Arnold 2010), tumor size and presence of perineural or lymphovascular invasion were obtained by review of surgical pathology reports.

Evaluation of tumor lymphocytic infiltration

Selected hematoxylin and eosin (H&E)-stained slides were inspected by two experienced neuroendocrine pathologists (D C and S A D) to confirm the original diagnosis. Formalin-fixed paraffin-embedded (FFPE) blocks representative of the tumor/non-tumor interface were chosen for both assessment of immune infiltration and immunohistochemistry (IHC) analysis. Lymphocytes were counted in five representative areas and their average number was scored into four categories: no infiltration (0 cells/5 HPF), mild infiltration (1-10/5 HPF), moderate infiltration (11-50/5 HPF) and intense infiltration (>50/5 HPF). Sections were examined independently by D C and S A D; in case of disagreement, a consensus was reached after joint review at a multihead microscope. The readers were blinded to the patient clinical outcome.

Immunohistochemistry

Immunohistochemistry (IHC) was used to evaluate the expression of PD-L1 and PD-1 on all available cases (n = 102). Intra- and peri-tumoral immune infiltration was characterized by using mAbs against CD3, CD4, CD8, CD20, CD21, CD27, CD79a, CD138, FOXP3 and Ki-67 in a subset of samples (n = 20) randomly selected within the original cohort. FFPE tissue sections of 4 µm in thickness were stained using the avidin-biotin complex method, as previously described (Coppola et al. 2011). Normal human tonsil tissue was used as a positive control for all staining. Parallel tests without the primary antibody served as negative control. Supplementary Table 1 (see section on supplementary data given at the end of this article) details dilutions, incubation times, clone and source for each mAb. For PD-L1 expression, both the intensity of the staining and the percentage of positive tumor cells were estimated. Staining intensity was scored considering 0 as negative or trace, 1 as weak, 2 as moderate and 3 as high. Various cut-offs were used for determining PD-L1 positivity, as summarized in Supplementary Table 2. Immunoreactivity for T cell markers including PD-1, CD3, CD4, CD8, CD27 and FOXP3 was detected in the tumor-infiltrating lymphocytes, while CD21 was used as marker of dendritic cells. B lymphocytes and B lymphocyte precursors were localized by CD20 and CD79a respectively, while CD138+ cells were interpreted as plasma cells. Immune cell proliferation was assessed by Ki-67 labeling. Immunoreactive cells were counted using the same criteria described above for the determination of lymphocytic infiltration on H&E-stained sections.

Statistical analysis

Descriptive statistics were used for patient demographics. The distribution of covariates was compared across groups using χ2, Fisher’s test or ANOVA for categorical variables and Mann–Whitney test for continuous data. Pearson’s test was used to assess the correlation between immune infiltration and the count of CD3+ cell. Overall survival (OS) was measured from date of initial diagnosis until death from any cause or last known follow-up. Cancer-specific survival (CSS) was calculated from the date of diagnosis until cancer-related death. Survival curves were estimated using the Kaplan–Meier method and compared by the log-rank test. Exact 95% CIs were calculated for each proportion of interest. All tests were two-sided and statistical significance was declared when P ≤ 0.05. Statistical analysis was conducted using MedCalc statistical software 12.7 (MedCalc Software bvba, Ostend, Belgium).

Results

Demographics and tumor characteristics

Demographic variables and clinicopathological characteristics are listed in Table 1. The median age at diagnosis was 60 years. Presence of the carcinoid syndrome was documented in 23% of patients at baseline. The majority of patients (87%) had ileal primaries, and the median tumor size was 1.9 cm. Sixty-nine percent had lymph node metastases, and stage IV disease was diagnosed in 54% of cases. Liver (47%) and peritoneum (10%) were the main sites of distant metastases. The majority of tumors (92%) were low grade. Most patients were either treatment naïve (48/102) or had received only somatostatin analogs (41/102) for a median of 2 months (range 1–13 months) before resection of the primary tumor.

Table 1

Patient demographics and tumor characteristics.

Characteristicsn of patients (n = 102)%
Age at diagnosis (years)
 Median60
 Range27–95
Sex
 Male5251
 Female5049
Race
 White8684
 Black1010
 Hispanic55
 Unknown11
Carcinoid syndrome
 Yes2323
 No7977
Tumor location
 Duodenum1010
 Jejunum22
 Ileum8785
 Unknown33
Tumor size (cm)
 Median1.9
 Range0.3–8
T stage (AJCC 2010 classification)
 T144
 T21414
 T34039
 T43635
 Tx88
N stage (AJCC 2010 classification)
 N02120
 N17169
 Nx1011
Macroscopic mesenteric lymph nodes
 Yes3130
 No7170
M stage (AJCC 2010 classification)
 M04746
 M15554
TNM stage (AJCC 2010 classification)
 I22
 IIA55
 IIB44
 IIIA00
 IIIB3332
 IV5554
 Unknown33
Sites of distant metastases
 Liver4847
 Peritoneum1010
 Ovary55
 Bone55
 Others44
Tumor grade (WHO 2010)
 G19492
 G288
Lymphovascular invasion
 Yes7170
 No1717
 Unknown1413
Perineural invasion
 Yes6362
 No1919
 Unknown2019
Systemic treatment before surgery
 None4140
 Somatostatin analogs Unknown48134713
Follow-up (months)
 Median61
 Range1–182

PD-1 and PD-L1 expression in SB NETs

All samples were assessable for PD-1 and PD-L1 expression. No immunoreactivity for PD-1 was observed in tumor cells across all examined cases. When PD-L1 positivity was defined by membranous staining in either ≥1 or ≥5% of tumor cells, we interpreted 40 of 102 cases as positive (39%; 95% CI, 30–49%). In PD-L1-positive tumors, the staining intensity was weak, moderate and high in 21 (52%; 95% CI, 37–67%), 15 (38%; 95% CI, 24–53%) and 4 (10%; 95% CI, 4–23%) samples respectively. When PD-L1 positivity was defined by membranous staining in ≥50% of tumor cells, a positive staining was detected in 14/102 (14%; 95% CI, 8–22%) samples. Staining intensity was weak, moderate and high in eight (57%; 95% CI, 33–79%), three (21%; 95% CI, 8–48%) and three cases respectively (Fig. 1A). Overall, interobserver agreement rate for PD-L1 expression was 85%. To mitigate potential biases deriving from the time-dependent fading of PD-L1 expression in FFPE samples, cases older than 3 years (n = 72; 71%) were analyzed separately. In this group, PD-L1 positivity was observed in 28/72 (39%; 95% CI, 28–50%) and 8/72 (11%; 95% CI, 6–20%) samples when the cut-off was set at ≥1 or ≥50% of tumor cells respectively.

Figure 1

Download Figure

Figure 1

PD-L1 expression in FFPE samples of small intestine NETs. (A) Small bowel NETs showed different intensities of PD-L1 expression (1+, weak; 2+, moderate; 3+ intense). Scale bar: 20 µm. By using either the ≥1% (B) or ≥50% (C) cut-offs for positivity interpretation, expression of PD-L1 was more common in duodenal NETs as compared with jejunal/ileal NETs.

Citation: Endocrine-Related Cancer 26, 1; 10.1530/ERC-18-0189

As depicted in Fig. 1B and C, duodenal NETs showed significantly higher PD-L1 positivity compared with jejunal/ileal NETs. Of the 14 samples showing PD-L1 expression in ≥50% of tumor cells, seven were duodenal. There were no differences in PD-L1 expression by other clinicopathological characteristics (Table 2). Rates of PD-L1 expression resulted similar when local tumors (N0M0) were compared to N+ tumors and M+ tumors.

Table 2

PD-L1 expression by clinicopathological characteristics.

CharacteristicsPD-L1 positive, ≥1% (n = 40)PD-L1 negative <1% (n = 62)PPD-L1 positive, ≥50% (n = 14)PD-L1 negative <50% (n = 88)P
Age in years, median (range)58 (27–85)61 (37–95)0.4365 (49–75)59 (27–95)0.4
Sex
 Male21 (52%)31 (50%)0.845 (36%)47 (53%)0.26
 Female19 (48%)31 (50%)9 (64%)41 (47%)
Race
 White34 (85%)52 (84%)110 (71%)76 (86%)0.23
 Others6 (15%)10 (16%)4 (29%)12 (14)
Carcinoid syndrome
 Yes9 (22%)14 (23%)10 (0%)19 (22%)0.07
 No31 (78%)48 (77%)14 (100%)69 (78%)
Tumor location
 Duodenum8 (20%)2 (3%)0.00137 (50%)3 (3%)<0.0001
 Jejunum/ileum29 (72%)60 (97%)7 (50%)82 (94%)
 Others3 (8%)0 (0%)0 (0%)3 (3%)
Tumor size in cm, median (range)2 (0.3–8)1.9 (0.3–8)0.801.2 (0.6–8)2 (0.3–8)0.23
T stage (AJCC classification)a
 T12 (6%)2 (3%)0.780 (0%)4 (5%)0.18
 T26 (18%)8 (13%)4 (36%)10 (12%)
 T314 (43%)26 (43%)4 (36%)36 (43%)
 T411 (33%)25 (41%)3 (28%)33 (40%)
N stage (AJCC classification)a
 N08 (24%)13 (22%)0.114 (33%)17 (21%)0.55
 N125 (76%)46 (78%)8 (67%)63 (79%)
Macroscopic mesenteric lymph nodes
 Yes13 (32%)18 (29%)0.833 (21%)28 (32%)0.54
 No27 (68%)44 (71%)11 (79%)60 (68%)
M stage (AJCC classification)
 M019 (47%)28 (45%)0.849 (64%)38 (43%)0.16
 M121 (53%)34 (55%)5 (36%)50 (57%)
TNM stage (AJCC classification)a
 I0 (0%)2 (3%)0.720 (0%)2 (2%)0.59
 II3 (8%)6 (10%)1 (8%)8 (9%)
 III13 (35%)20 (32%)6 (50%)27 (31%)
 IV21 (57%)34 (55%)5 (42%)50 (58%)
Tumor grade (WHO 2010)
 G137 (93%)57 (92%)113 (93%)81 (92%)1
 G23 (7%)5 (8%)1 (7%)7 (8%)
Lymphovascular invasiona
 Yes25 (81%)46 (81%)18 (80%)63 (81%)1
 No6 (19%)11 (19%)2 (20%)15 (19%)
Perineural invasiona
 Yes19 (68%)44 (81%)0.186 (67%)57 (78%)0.43
 No9 (32%)10 (19%)3 (33%)16 (22%)
Sample age, year of collection
 2000–20059 (22%)7 (11%)0.263 (21%)13 (15%)0.71
 2006–201018 (45%)28 (45%)5 (36%)41 (46%)
 2011–201513 (33%)27 (44%)6 (43%)34 (39%)
Prior treatment with somatostatin analogsa
 Yes17 (47%)28 (53%)0.674 (29%)39 (52%)0.15
 No19 (53%)25 (47%)10 (71%)36 (48%)
Follow-up in months, median (range)66 (23–162)59 (1–182)0.2261 (23–120)60 (1–182)0.99

aPatients with unknown status of these parameters were excluded from the analysis.

Immune infiltration in SB NETs

Intratumor infiltration of immune cells was observed in 67/102 samples (66%; 95% CI, 56–74%), while no apparent host immune response was noted in the remaining 35 cases (34%; 95% CI, 26–44%). The density of immune infiltration was mild to moderate in 46/67 cases (69%; 95% CI, 57–78%) and intense in 21/67 samples (31%, 95% CI, 21–43%) (Fig. 2). In 18/102 tumors (18%; 95% CI, 11–26%), there was a marked peritumoral host response that was organized as ectopic lymph nodes (ELNs) (Fig. 2D). This response was particularly evident at the invasive edge of the tumors and was observed in 4, 5 and 9 samples with absent, moderate and intense intratumor immune infiltration respectively (P = 0.005). The presence of ELNs appeared to be associated with PD-L1 expression in ≥50% of tumor cells (P = 0.0003), but there was no statistically significant correlation with PD-L1 staining in ≥1% of tumor cells (P = 0.06). ELNs were detected more frequently in duodenal tumors as compared with jejunoileal NETs (P = 0.0009). As summarized in Table 3, the density of the immune infiltration was significantly higher in duodenal tumors as compared with jejunoileal NETs (P = 0.0007). Moreover, there was a significant association between tumor PD-L1 expression and the extent of host immune response. No differences were seen between local tumors, N+ tumors and M+ tumors in terms density of immune infiltration. Figure 3 summarizes the main features of the immune microenvironment of SB NETs.

Figure 2

Download Figure

Figure 2

Histopathologic examples of immune infiltration in SB NETs. (A) Absence of intratumor lymphocyte infiltrates. Examples of mild and intense immune infiltration are depicted in (B) and (C) respectively. Scale bar: 20 µm. (D) ELNs (black arrows) at the invasive tumor edge. Magnification: 200×.

Citation: Endocrine-Related Cancer 26, 1; 10.1530/ERC-18-0189

Figure 3

Download Figure

Figure 3

Heterogeneity of the immune microenvironment of SB NETs. (A) In the overall cohort, one-third of cases showed features suggestive of adaptive immune resistance. (B) and (C) depict the immune landscape of duodenal and jejunal/ileal NETs respectively. A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0189.

Citation: Endocrine-Related Cancer 26, 1; 10.1530/ERC-18-0189

Table 3

Intratumor immune infiltration by clinicopathological characteristics.

CharacteristicsAbsence of immune infiltration (n = 35)Low-density immune infiltration (n = 46)High-density immune infiltration (n = 21)P
Age in years, median (range)57 (34–77)60 (35–85)65 (27–95)0.55
Sex
 Male17 (49%)24 (52%)11 (52%)0.94
 Female18 (51%)22 (48%)10 (48%)
Race
 White27 (77%)41 (89%)18 (86%)0.33
 Others8 (23%)5 (11%)3 (14%)
Carcinoid syndrome
 Yes5 (14%)15 (33%)3 (14%)0.09
 No30 (86%)31 (67%)18 (86%)
Tumor location
 Duodenum3 (9%)2 (4%)5 (24%)0.0007
 Jejunum/ileum32 (91%)44 (96%)13 (62%)
 Others0 (%)0 (0%)3 (14%)
Tumor size in cm, median (range)1.5 (0.3–4)2 (0.3–8)2.5 (0.6–8)0.30
T stage (AJCC classification)a
 T12 (6%)2 (4%)0 (0%)0.98
 T26 (17%)6 (14%)2 (13%)
 T314 (40%)19 (43%)7 (47%)
 T413 (37%)17 (39%)6 (40%)
N stage (AJCC classification)a
 N07 (21%)10 (23%)4 (27%)0.92
 N126 (79%)34 (77%)11 (73%)
Macroscopic mesenteric lymph nodes
 Yes13 (37%)12 (26%)6 (29%)0.55
 No22 (63%)34 (74%)15 (71%)
M stage (AJCC classification)
 M018 (51%)18 (39%)10 (48%)0.53
 M117 (49%)28 (61%)11 (52%)
TNM stage (AJCC classification)a
 I1 (3%)1 (2%)0 (0%)0.72
 II4 (11%)2 (5%)3 (16%)
 III13 (37%)14 (31%)6 (32%)
 IV17 (49%)28 (62%)10 (53%)
Tumor grade (WHO 2010)
 G131 (89%)44 (96%)19 (91%)0.48
 G24 (11%)2 (4%)2 (9%)
Lymphovascular invasiona
 Yes25 (78%)35 (88%)11 (69%)0.25
 No7 (22%)5 (12%)5 (31%)
Perineural invasiona
 Yes24 (80%)30 (77%)9 (69%)0.74
 No6 (20%)9 (23%)4 (31%)
Sample age, year of collection
 2000–20056 (17%)5 (11%)5 (24%)0.35
 2006–201015 (43%)25 (54%)6 (29%)
 2011–201514 (40%)16 (35%)10 (47%)
Follow-up in months, median (range)57 (0–182)70 (5–169)52 (23–162)0.22
PD-L1 expression (≥1% cut-off)
 Yes7 (20%)19 (41%)14 (67%)0.0023
 No28 (80%)27 (59%)7 (33%)
PD-L1 expression (≥50% cut-off)0.0137
 Yes3 (9%)4 (9%)7 (33%)
 No32 (91%)42 (91%)14 (67%)

aPatients with unknown status of these parameters were excluded from the analysis.

IHC characterization of the intratumor immune response

Intratumor infiltration of PD-1-positive lymphocytes was observed in 9/102 samples (9%; 95% CI, 5–16%). Of these, seven were of ileal origin and two of duodenal origin. The density of the infiltration was considered mild and intense in 3 and 6 tumors respectively. There was no correlation between PD-L1 expression and intratumor infiltration by PD-1-positive lymphocytes (P = 0.07), while PD-1-positive infiltration significantly correlated with PD-L1 staining intensity (P < 0.0001). No association was found between infiltration by PD-1-positive cells and tumor grade (P = 0.54).

To further investigate the cellular makeup of the host immune response, we carried out serial IHC studies in 20 tumors randomly selected from our cohort (n = 9 cases with mild/moderate immune infiltration; n = 11 cases with intense immune infiltration on H/E). Infiltrating CD3+ and CD20+ lymphocytes were found in 17/20 (85%; 95% CI, 64–95%) and 10/20 (50%; 95% CI, 30–70%) samples. There was a significant correlation between the density of the immune infiltration by H/E and CD3+ cell count (r=0.77; 95% CI, 0.39–0.93; P = 0.0013). The degree of CD3+ infiltration positively correlated with the expression of PD-L1 by tumor cells (P = 0.03). Infiltrating CD4+ and CD8+ lymphocytes were detected in 15/20 (75%; 95% CI, 53–89%) and 16/20 (80%; 95% CI, 58–92%) tumors respectively. Tumor-infiltrating CD27+ memory T cells were observed in 17/20 (85%; 95% CI, 64–95%) cases, while FOXP3+ T regulatory cells were found in 5/20 (25%; 95% CI, 11–47%) samples. Scattered CD138+ plasma cells were detected within the tumor parenchyma in only 2/20 (10%; 95% CI, 3–30%) samples. No CD21+ dendritic cells could be found within tumors.

IHC characterization of ELNs

We assessed the composition of ELNs in all ELN-positive samples (n = 18) by IHC. As depicted in Fig. 4, the majority of CD3+ T cells were located in the parafollicular cortex-like or marginal-like zones and expressed PD-1. CD20+ B cells and their CD79a+ precursors were present almost exclusively within the follicular structures, in close contact with CD21+ dendritic cells and CD27+ memory T cells. No or very low-density FOXP3+ cells were detected within ELNs. ELNs were uniformly activated, as suggested by the expression of Ki-67 within the pseudo-germinal centers in all cases analyzed, thus supporting the hypothesis that the described lymphoid aggregates represent secondary and/or tertiary lymphoid structures.

Figure 4

Download Figure

Figure 4

Cellular composition of ELNs. By IHC, the cellular makeup of ELNs resembles secondary or tertiary lymphoid structures. (A) CD3+ T cells are mainly located in the parafollicular cortex, with a similar distribution for CD4+ (B) and CD8+ cells (C). CD20+ B cells (D) and CD21+ dendritic cells (E) tend to concentrate in the center of the follicle, where the proliferative activity appears more marked (F). CD27+ memory T cells (G) are localized at the boundary with the T cell zone, with some dispersion into the follicle. Only sporadic FOXP3+ regulatory T cells can be found in the context of lymphoid aggregate (H), while PD-1+ lymphocytes (I) appear to be dispersed throughout the ELN. Magnification: 100×.

Citation: Endocrine-Related Cancer 26, 1; 10.1530/ERC-18-0189

Prognostic value of PD-L1 expression in stage IV SB NETs

Among 55 patients with metastatic SB NET, 16 had died and 39 were alive at the time of data cut-off. Seven deaths were known to be directly related to progressive metastatic disease, seven were possibly related to disease progression and two were not tumor related. Only one patient with duodenal NET had metastatic disease and was alive at the time of data cut-off. After a median follow-up of 60 months (range, 23–182 months), the 5-year and 10-year OS rates were 75% (±6%) and 59% (±10%) respectively. The 5-year and 10-year CSS rates were 77% (±6%) and 70% (±8%). There was no correlation between OS or CSS and tumor expression of PD-L1, irrespective of the cut-off used for defining staining positivity (Supplementary Fig. 1).

Prognostic value of immune infiltration in stage IV SB NETs

Among patients with metastatic SB NET, intratumor immune infiltration was absent in 17/55 cases, mild to moderate in 28 cases and intense in 10 cases. There was no significant difference in terms of both OS and CSS according to the degree of immune infiltration. Neither ELN presence nor intratumor infiltration by PD-1-positive cells had an impact on OS or CSS (Supplementary Fig. 2).

Discussion

Data on expression of immune checkpoint molecules in SB NETs are limited. Prior studies (Kim et al. 2016, Cavalcanti et al. 2017, Roberts et al. 2017, Lamarca et al. 2018) have been limited by small sample size, inclusion of heterogeneous cohorts of tumors and lack of IHC standardization. Yet, accurate information on intratumor immune infiltration and expression of PD-L1 by tumor cells is critically important to define the immune environment status of SB NETs and potentially identify subgroups of patients suitable for emerging immunotherapy approaches.

This study represents, to our knowledge, the largest series to examine PD-L1 expression and tumor-associated immune infiltration in a homogeneous population of patients with G1/G2, SB NETs. We included in our analysis only patients who underwent surgical resection of duodenal, jejunal or ileal primaries in order to avoid possible overestimation or underestimation of PD-L1 expression resulting from inadequate/unrepresentative biopsy sampling and evaluation of metastatic sites. To minimize the biases possibly deriving from the intratumoral heterogeneity of PD-L1 expression, all samples were analyzed at the tumor/non-tumor interface.

Results of our analysis show that PD-L1 is expressed in 39 and 14% of samples when the cut-off for IHC positivity determination is set at ≥1/5% or ≥50% of tumor cells, respectively. Interestingly, expression of PD-L1 appears to be significantly higher in duodenal NETs, where staining positivity in ≥50% of tumor cells was observed in 70% of tumors. The rate of PD-L1 expression observed in our cohort appears higher than rates preliminarily reported by other groups. A complete absence of membranous PD-L1 expression was described in a series of 64 SB NETs (da Silva et al. 2018), while a positive staining in ≥5% of tumor cells was noted in 13% of samples analyzed in a study of 62 patients with small intestinal NETs (Lamarca et al. 2018). Although different anti-PD-L1 Abs have been used in various reports, high levels of staining concordance have been recently shown when comparing such Abs (Gaule et al. 2016). This suggests that the reported discordance of PD-L1 expression relies primarily on the high levels of intratumor heterogeneity already described for other cancers (Gaule et al. 2016), rather than on Ab-based variability. In this context, the systematic analysis of PD-L1 expression at the tumor/non-tumor interface may have influenced our results, since increased levels of PD-L1 expression have been described at the invasive tumor edge (McLaughlin et al. 2016).

Based on our data, about one-third of SB NETs appear immunologically ignorant (PD-L1; TIL), and therefore, not suitable to immune checkpoint blockade. On the other hand, two-thirds of tumors were infiltrated by inflammatory cells, and intratumor immune infiltration was positively associated with PD-L1 expression. This suggests that ‘immunologically hot’ SB NETs upregulate PD-L1 in response to cytokines released by tumor-infiltrating lymphocytes as an adaptive immune resistance mechanism, rather than as consequence of constitutively activated oncogenic signaling pathways (Fig. 5). In this context, it is intriguing to note that a type I microenvironment (PD-L1+; TIL+), which is thought to be the immune subgroup most responsive to checkpoint blockade, predominates in duodenal vs jejunal/ileal NETs, making duodenal primaries preferred candidates for immunotherapy trials of SB NET patients.

Figure 5

Download Figure

Figure 5

Immune microenvironment and PD-L1 regulation in SB NETs. After tumor initiation (A), SB NETs are infiltrated by inflammatory cells that produce pro-inflammatory cytokines, thus stimulating the expression of PD-L1 on tumor cells (B). As result of the interaction between PD-L1 on the membrane of tumor cells and PD-1 on the surface of T lymphocytes, cross-primed CD8+ T cells become exhausted and lack their cytotoxic capability, leading to tumor immune escape (C). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0189.

Citation: Endocrine-Related Cancer 26, 1; 10.1530/ERC-18-0189

In our series, only a minority of samples stained positively for PD-1, suggesting that infiltrating lymphocytes lack effective priming by tumor neoantigens. This is in line with the low mutational burden of SB NETs (Francis et al. 2013). An alternative hypothesis is that co-inhibitory receptors different from PD-1 (LAG-3, TIGIT, TIM-3, etc) are activated following lymphocyte activation.

About one-fifth of cases analyzed in our study had ELNs showing germinal center maturation at the invasive edge of the tumor, and their presence correlated with immune infiltration and PD-L1 expression. It is difficult to draw firm conclusions about the true nature of ELNs, but their prevalence in the duodenum, where secondary lymphoid structures such as Peyer’s patches tend to be less common, suggests that they might be tertiary lymphoid structures (TLS). Described in a number of cancers, TLS have been involved in either immune-surveillance or immune-suppression (Coppola & Mulé 2008, Colbeck et al. 2017).

In line with prior observations in well-differentiated NETs (Kasajima et al. 2018, Lamarca et al. 2018), we found that both PD-L1 expression and immune infiltration had no impact on prognosis. Although the predictive ability of PD-L1 expression can be assessed only in prospective trials of immune checkpoint inhibitors, the heterogeneous immune landscape reported in our study for SB NETs suggests that highly variable patterns of response to immunotherapy may be encountered in the clinical setting for these malignancies.

This study has several limitations. First, our analysis was focused exclusively on primary tumors, thus hindering any possible considerations on the role of the immune microenvironment in the progression of NETs. Second, we systematically assessed the expression of PD-L1 only at the tumor/non-tumor interface. This strategy of staining standardization may have limited our ability to evaluate the intratumoral immune landscape.

Taken together, our data show that the immune landscape of SB NETs is heterogeneous, with adaptive immune resistance mechanisms prevailing in duodenal NETs rather than in jejunal/ileal carcinoids. Identification of patients more likely to respond to immunotherapy is key for the success of clinical trials of immune checkpoint inhibitors.

Supplementary data

This is linked to the online version of the paper at https://doi.org/10.1530/ERC-18-0189.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding

This work did not receive any specific grant from any funding agency in the public, commercial, or not-for-profit sector.

References

  • CavalcantiEArmentanoRValentiniAMChieppaMCarusoML 2017 Role of PD-L1 expression as a biomarker for GEP neuroendocrine neoplasm grading. Cell Death and Disease 8 e3004. (https://doi.org/10.1038/cddis.2017.401)

  • ColbeckEJAgerAGallimoreAJonesGW 2017 Tertiary lymphoid structures in cancer: drivers of antitumor immunity, immunosuppression, or bystander sentinels in disease? Frontiers in Immunology 8 1830. (https://doi.org/10.3389/fimmu.2017.01830)

  • CoppolaDMuléJJ 2008 Ectopic lymph nodes within human solid tumors. Journal of Clinical Oncology 26 43694370. (https://doi.org/10.1200/JCO.2008.17.6149)

  • CoppolaDNebozhynMKhalilFDaiHYeatmanTLobodaAMuléJJ 2011 Unique ectopic lymph node-like structures present in human primary colorectal carcinoma are identified by immune gene array profiling. American Journal of Pathology 179 3745. (https://doi.org/10.1016/j.ajpath.2011.03.007)

  • da SilvaABowdenMZhangSMasugiYThornerARHerbertZTZhouCWBraisLChanJAHodiFSet al. 2018 Characterization of the neuroendocrine tumor immune microenvironment. Pancreas 47 11231129. (https://doi.org/10.1097/MPA.0000000000001150)

  • DasariAShenCHalperinDZhaoBZhouSXuYShihTYaoJC 2017 Trends in the incidence, prevalence, and survival outcomes in patients with neuroendocrine tumors in the United States. JAMA Oncology 3 13351342. (https://doi.org/10.1001/jamaoncol.2017.0589)

  • FrancisJMKiezunARamosAHSerraSPedamalluCSQianZRBanckMSKanwarRKulkarniAAKarpathakisA 2013 Somatic mutation of CDKN1B in small intestine neuroendocrine tumors. Nature Genetics 45 14831486. (https://doi.org/10.1038/ng.2821)

  • FrillingAAkerströmGFalconiMPavelMRamosJKiddMModlinIM 2012 Neuroendocrine tumor disease: an evolving landscape. Endocrine-Related Cancer 19 R163R185. (https://doi.org/10.1530/ERC-12-0024)

  • GaulePSmithyJWTokiMRehmanJPatell-SochaFCougotDCollinPMorrillPNeumeisterVRimmDL 2016 A quantitative comparison of antibodies to programmed cell death 1 ligand 1. JAMA Oncology [epub]. (https://doi.org/10.1001/jamaoncol.2016.3015)

  • KasajimaAIshikawaYIwataASteigerKOkaNIshidaHSakuradaASuzukiHKameyaTKonukiewitzB 2018 Inflammation and and PD-L1 expression in pulmonary neuroendocrine tumors. Endocrine-Related Cancer 25 339350. (https://doi.org/10.1530/ERC-17-0427)

  • KimSTHaSYLeeSAhnSLeeJParkSHParkJOLimHYKangWKKimKM 2016 The impact of PD-L1 expression in patients with metastatic GEP-NETs. Journal of Cancer 7 484489. (https://doi.org/10.7150/jca.13711)

  • LamarcaANonakaDBreitwieserWAshtonGBarriusoJMcNamaraMGMoghadamSRoganJMansoorWHubnerRA 2018 PD-L1 expression and presence of TILs in small intestinal neuroendocrine tumours. Oncotarget 9 1492214938. (https://doi.org/10.18632/oncotarget.24464)

  • McLaughlinJHanGSchalperKACarvajal-HausdorfDPelekanouVRehmanJVelchetiVHerbstRLoRussoPRimmDL 2016 Quantitative assessment of the heterogeneity of PD-L1 expression in non-small-cell lung cancer. JAMA Oncology 2 4654. (https://doi.org/10.1001/jamaoncol.2015.3638)

  • PavelMESersC 2016 WOMEN IN CANCER THEMATIC REVIEW: systemic therapies in neuroendocrine tumors and novel approaches toward personalized medicine. Endocrine-Related Cancer 23 T135T154. (https://doi.org/10.1530/ERC-16-0370)

  • RindiGArnoldR 2010 Nomenclature and classification of neuroendocrine neoplasms of the digestive system. In World Health Organization Classification of Tumours of the Digestive System1314. Eds BosmanFCarneiroFHrubanR & TheiseN. Lyon, France: IARC Press.

  • RindiGKlöppelGCouvelardAKomminothPKörnerMLopesJMMcNicolAMNilssonOPerrenAScarpaA 2007 TNM staging of midgut and hindgut (neuro) endocrine tumors: a consensus proposal including a grading system. Virchows Archiv 451 757762. (https://doi.org/10.1007/s00428-0007-0452-1)

  • RobertsJAGonzalezRSDasSBerlinJShiC 2017 Expression of PD-1 and PD-L1 in poorly differentiated neuroendocrine carcinomas of the digestive system: a potential target for anti-PD-1/PD-L1 therapy. Human Pathology 70 4954. (https://doi.org/10.1016/j.humpath.2017.10.003)

  • TengMWNgiowSFRibasASmythMJ 2015 Classifying cancers based on T-cell infiltration and PD-L1. Cancer Research 75 21392145. (https://doi.org/10.1158/0008-5472.CAN-15-0255)

  • VikmanSGiandomenicoVSommaggioRObergKEssandMTottermanTH 2008 CD8+ T cells against multiple tumor-associated antigens in peripheral blood of midgut carcinoid patients. Cancer Immunology Immunotherapy 57 399409. (https://doi.org/10.1007/s00262-007-0382-4)

  • VikmanSSommaggioRDe La TorreMObergKEssandMGiandomenicoVLoskogATottermanTH 2009 Midgut carcinoid patients display increased numbers of regulatory T cells in peripheral blood with infiltration into tumor tissue. Acta Oncologica 48 391400. (https://doi.org/10.1080/02841860802438495)

 

An official journal of

Society for Endocrinology

Sections

Figures

  • View in gallery

    PD-L1 expression in FFPE samples of small intestine NETs. (A) Small bowel NETs showed different intensities of PD-L1 expression (1+, weak; 2+, moderate; 3+ intense). Scale bar: 20 µm. By using either the ≥1% (B) or ≥50% (C) cut-offs for positivity interpretation, expression of PD-L1 was more common in duodenal NETs as compared with jejunal/ileal NETs.

  • View in gallery

    Histopathologic examples of immune infiltration in SB NETs. (A) Absence of intratumor lymphocyte infiltrates. Examples of mild and intense immune infiltration are depicted in (B) and (C) respectively. Scale bar: 20 µm. (D) ELNs (black arrows) at the invasive tumor edge. Magnification: 200×.

  • View in gallery

    Heterogeneity of the immune microenvironment of SB NETs. (A) In the overall cohort, one-third of cases showed features suggestive of adaptive immune resistance. (B) and (C) depict the immune landscape of duodenal and jejunal/ileal NETs respectively. A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0189.

  • View in gallery

    Cellular composition of ELNs. By IHC, the cellular makeup of ELNs resembles secondary or tertiary lymphoid structures. (A) CD3+ T cells are mainly located in the parafollicular cortex, with a similar distribution for CD4+ (B) and CD8+ cells (C). CD20+ B cells (D) and CD21+ dendritic cells (E) tend to concentrate in the center of the follicle, where the proliferative activity appears more marked (F). CD27+ memory T cells (G) are localized at the boundary with the T cell zone, with some dispersion into the follicle. Only sporadic FOXP3+ regulatory T cells can be found in the context of lymphoid aggregate (H), while PD-1+ lymphocytes (I) appear to be dispersed throughout the ELN. Magnification: 100×.

  • View in gallery

    Immune microenvironment and PD-L1 regulation in SB NETs. After tumor initiation (A), SB NETs are infiltrated by inflammatory cells that produce pro-inflammatory cytokines, thus stimulating the expression of PD-L1 on tumor cells (B). As result of the interaction between PD-L1 on the membrane of tumor cells and PD-1 on the surface of T lymphocytes, cross-primed CD8+ T cells become exhausted and lack their cytotoxic capability, leading to tumor immune escape (C). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0189.

Index Card

PubMed

Google Scholar

Related Articles

Altmetrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 186 186 186
PDF Downloads 41 41 41