Fax: (011) 81-7134-6865
Predominant infiltration of macrophages and CD8+ T Cells in cancer nests is a significant predictor of survival in stage IV nonsmall cell lung cancer
Article first published online: 31 JUL 2008
Copyright © 2008 American Cancer Society
Volume 113, Issue 6, pages 1387–1395, 15 September 2008
How to Cite
Kawai, O., Ishii, G., Kubota, K., Murata, Y., Naito, Y., Mizuno, T., Aokage, K., Saijo, N., Nishiwaki, Y., Gemma, A., Kudoh, S. and Ochiai, A. (2008), Predominant infiltration of macrophages and CD8+ T Cells in cancer nests is a significant predictor of survival in stage IV nonsmall cell lung cancer. Cancer, 113: 1387–1395. doi: 10.1002/cncr.23712
- Issue published online: 4 SEP 2008
- Article first published online: 31 JUL 2008
- Manuscript Accepted: 14 APR 2008
- Manuscript Revised: 10 APR 2008
- Manuscript Received: 31 OCT 2007
- stage IV;
- nonsmall cell lung cancer;
- CD8+ T cell;
- mast cell
The purpose of this study was to investigate whether tumor-infiltrating immune cells in biopsy specimens can be used to predict the clinical outcome of stage IV nonsmall cell lung cancer (NSCLC) patients.
The authors performed an immunohistochemical study to identify and count the number of CD68+ macrophages, c-kit+ mast cells, and CD8+ T cells in both cancer nests and cancer stroma in pretreatment biopsy specimens obtained from 199 patients with stage IV NSCLC treated by chemotherapy, and then analyzed for correlations between the number of immune cells and clinical outcome, including chemotherapy response and prognosis.
There was no correlation between the number of immune cells in either cancer nests or stroma and chemotherapy response. Patients with more tumor-infiltrating macrophages in cancer nests than in cancer stroma (macrophages, nests > stroma) had significantly better survival than nests < stroma cases median survival time (MST 440 days vs 199 days; P < .0001). Patients with more tumor-infiltrating CD8+ T cells in cancer nests than in cancer stroma (CD8+ T cells: nests > stroma) showed significantly better survival than in nests < stroma cases (MST 388 days vs 256 days; P = .0070). The proportion of tumor-infiltrating macrophages or CD8+ T cells between cancer nests and stroma became independent prognostic factors in the multivariate analysis. Neither the number of mast cells in nests nor in stroma correlated with the clinical outcome.
Evaluation of the numbers of macrophages and CD8+ T cells in cancer nests and stroma are useful biomarkers for predicting the prognosis of stage IV NSCLC patients treated with chemotherapy, but could fail to predict chemotherapy response. Cancer 2008. © 2008 American Cancer Society.
Lung cancer is the leading cause of cancer deaths throughout the world, and nonsmall cell lung cancer (NSCLC) accounts for approximately 80% of lung cancer. The prognosis of NSCLC is poor, and patients with advanced NSCLC are candidates for systemic chemotherapy.1 During the 1990s, 5 new drugs became available for the treatment of metastatic NSCLC: paclitaxel, docetaxel, vinorelbine, gemcitabine, and irinotecan. Each of them has since been evaluated in combination regimens with cisplatin or carboplatin, and the median survival time of patients with metastatic NSCLC treated with such regimens is approximately 8 to 10 months.2, 3 However, some patients with metastatic NSCLC exhibit long-term survival, and their tumors progress slowly after chemotherapy, or even in the absence of treatment.4
Tumor cells are surrounded by infiltrating inflammatory cells, such as lymphocytes, neutrophils, macrophages, and mast cells. Infiltration of CD8+ T cells has been shown to be an important phenomenon for a specific immune response in several tumor systems, and CD8+ T cells have been reported to play an important suppressive role in cancer progression, including ovarian cancer, esophageal cancer, pancreatic cancer, bile duct cancer, gallbladder cancer, and colorectal cancer.5–12 Immunologists have long considered the presence of tumor-infiltrating macrophages as evidence of a host response against the growing tumor, and the presence of tumor-infiltrating macrophages has been reported to be associated with anticancer immunomechanisms of the tumor-bearing host and a favorable prognosis. However, recently, tumor-associated macrophage infiltration has been found to be correlated with angiogenesis and an unfavorable prognosis in several kinds of cancer, including gastric cancer, endometrial cancer, and breast cancer.13-15 Furthermore, it has been reported that mast cells produce many angiogenic factors and a variety of cytokines, including transforming growth factor-beta, tumor necrosis factor-α (TNF-α), interleukin-8 (IL-8), fibroblast growth factor-2, and vascular endothelial growth factor, which are implicated in both normal and tumor-associated neoangiogenesis.16 In fact, mast cell density has been reported to be highly correlated with the extent of both normal and pathologic angiogenesis, such as the angiogenesis observed in chronic inflammatory diseases and tumors, including gastric cancer and endometrial cancer.17, 18
Tumor-infiltrating immune cells are thought to play important roles in disease progression and therapeutic efficacy. The effect of chemoradiotherapy has been found to be correlated with the presence of CD8+ T cells in esophageal cancer,11 and cervical cancer patients with T-cell infiltration showed improved local response to radiation therapy.19
In the current study, we evaluated the status of tumor-infiltrating immune cells in tumor biopsy specimens obtained from stage IV NSCLC patient and analyzed the numbers of immune cells and clinical outcome, including chemotherapy response and prognosis, for correlations.
MATERIALS AND METHODS
Patients and Tissue Specimens
The tumor specimens analyzed in this study were obtained from a total of 199 patients with stage IV NSCLC who received platinum-based combination chemotherapy (cisplatin plus paclitaxel, docetaxel, gemcitabine, vinorelbine, or irinotecan, or carboplatin plus paclitaxel), which is considered to be the standard regimen20, 21 at the National Cancer Center Hospital East in Kashiwa, Chiba, Japan, between January 1996 and December 2004, with performance status (PS) 0 or 1 on the Eastern Cooperative Oncology Group scale. Of the 199 patients, 184 had died by the time of the analysis. All patients had adequate tumor biopsy specimens obtainable before chemotherapy and were analyzed in this study. The tumor specimens were obtained by bronchoscopy in 152 patients, and by percutaneous needle biopsy in 47 patients. The histological classification was based on a World Health Organization report. Clinical staging was based on an initial evaluation consisting of a clinical assessment, chest radiography, computed tomography of the chest and abdomen, computed tomography or magnetic resonance imaging of the brain, and bone scintigraphy. The current International Staging System was used to stage clinical disease.22 All target lesions were evaluated for response by computed tomography or magnetic resonance imaging after completion of the first-line chemotherapy, and all patients underwent tumor biopsy and chemotherapy, after obtaining informed consent in accordance with institutional guidelines.
Immunohistochemistry and Cell Counts
All specimens were fixed in 10% formalin and paraffin embedded. Four-micrometer-thick sections were mounted on silanized slides and deparaffinized with xylene and ethanol. To retrieve the antigen for macrophages, sections were pretreated in 0.05% trypsin and incubated for 20 minutes at 37°C in a humidity chamber. Endogenous peroxidase was blocked with 0.3% H2O2 in methanol for 15 minutes. We used mouse antihuman CD68 antibody (Dako, Kyoto, Japan) to detect macrophages, mouse antihuman CD8 antibody (Novocastra, Newcastle, UK) to detect T cells, and mouse antihuman c-kit antibody (Dako) to detect mast cells; immunostaining was performed with Envision (Dako). To retrieve the antigen for CD8 and c-kit, sections were immersed in 10 mM citric buffer solution (pH 6.0) and heated to 95°C by exposure to microwave irradiation for 20 minutes.
We performed an immunohistochemical study to identify and count the number of CD68+ macrophages, c-kit+ mast cells, and CD8+ T cells and confirmed that cancer cells and mesenchymal cells such as endothelial cells were not immunostained with these antibodies.
Immunostained cells counts were blinded to the patients' clinical data. Macrophages, CD8+ T cells, and mast cells in the specimen were counted in 2 locations: in the “cancer nests” and in the “cancer stroma.” Cancer nests were defined as “cancer nests without fibroblasts and vasculatures” and cancer stroma as “connective tissues surrounding cancer nests without any cancer cells.” Every biopsy specimen had both cancer nest and stroma, and it was possible to distinguish these lesions. We counted CD68+ round cells as macrophages, c-kit+ round cells as mast cells, and CD8+ round cells as T cells. By using a high-power microscopic field (×400; 0.0625 mm2), we separately counted the number of macrophages, CD8+ T cells, and mast cells in each 2 most intensive areas. Two pathologists (O.K. and G.I) reviewed all slides and counted the cells.
Statistical analysis was performed using the Scientific Package for Social Sciences (SPSS, Chicago, Ill) software. We used median values to calculate category correlations between macrophages, CD8+ T cells, mast cells, and survival rate by the Kaplan-Meier method, and performed univariate analyses by means of log-rank test. The chi-square test was used to test for relationships between categorical variables. Multivariate analysis was performed by means of the Cox proportional hazards model. Student t test was used to test for correlation between macrophage counts, CD8+ T cell counts, mast cell counts and response to first-line chemotherapy. We evaluated test results as significant if the P value was P < .05.
The clinicopathological characteristics of all patients are listed in Table 1. Their median age at the time of diagnosis was 62 years (range, 39 years-79 years), and 139 of the 199 patients were men. There were 134 patients with adenocarcinoma, 41 patients with squamous cell carcinoma, and 24 patients with NSCLC that could not be specified by biopsy specimen.
|Patients (N = 199)|
|Median, y (range)||62 (39-79)|
|Squamous cell carcinoma||41||20.6|
|ECOG performance status|
|<20 pack years||73||36.6|
|≥20 pack years||126||63.3|
|Median survival time, d (range)||317 (19-1969)|
|Response to first-line chemotherapy|
Macrophages, Mast Cells, and CD8+ T Cells, in Cancer Nests and Cancer Stroma
Macrophages were observed in cancer nests (Fig. 2A) in 194 of the 199 tumors, and the mean number was 18.0 ± 2.4 (median, 13; range, 0-76). Macrophages were also observed in cancer stroma (Fig. 2B) in 195 of the 199 tumors, and the mean number was 15.3 ± 1.9 (median, 12; range, 0-105). Mast cells were observed in cancer nests (Fig. 2C) in 149 tumors and in the cancer stroma (Fig. 2D) in 158 tumors, and the mean number was 4.5 ± 0.8 (median, 2; range, 0-52), and 5.4 ± 0.8 (median, 3; range, 0-28), respectively. CD8+ T cells were observed in cancer nests (Fig. 2E) in 197 tumors, and the mean number was 16.9 ± 2.2 (median, 12; range, 0-89). CD8+ T cells were observed in the cancer stroma (Fig. 2F) in 198 tumors, and the mean number was 15.7 ± 1.8 (median, 13; range, 0-88).1
Relationships between the number of infiltrating macrophages, mast cells, CD8+ T cells, and clinicopathological variables
The numbers of infiltrating macrophages, mast cells, and CD8+ T cells were divided into 2 groups at the median value. The relationships between these groups in cancer nests or stroma and the individual clinicopathological characteristics (sex, age, smoking history, PS, histological type) were examined by the chi-square test. More macrophages were present in cancer nests in the nonadenocarcinomas than in the adenocarcinomas (data not shown).
Correlations between the numbers of macrophages, CD8+ T cells, mast cells, and first-line chemotherapy response
We analyzed the number of macrophages, mast cells, and CD8+ T cells in cancer nests and stroma and first-line chemotherapy response for correlations by Student t test (Table 2), but the results showed no correlations between numbers of any of the infiltrating immune cells and response to first-line chemotherapy.
|Macrophages in cancer nests||−0.577||−7.173-3.946||.556|
|Macrophages in cancer stroma||0.119||−4.094-4.617||.905|
|Mast cells in cancer nests||−0.413||−2.310-1.512||.680|
|Mast cells in cancer stroma||1.476||−0.427-2.929||.143|
|CD8+ T cells in cancer nests||−1.045||−7.114-2.201||.298|
|CD8+ T cells in cancer stroma||−0.586||−5.162-2.807||.559|
Correlations between the numbers of tumor-infiltrating macrophages, mast cells, and CD8+ T cells and patient survival
Kaplan-Meier survival analyses and the log-rank test were performed to compare survival with the number of infiltrating cells (Fig. 2, Table 3). Patients with more macrophages in cancer nests than the median value had the same survival rate as patients with fewer macrophages. By contrast, patients with more macrophages in the cancer stroma had significantly poorer survival than those with fewer macrophages (P = .0001). The median survival time was 243 days in the group with higher numbers of macrophages in the stroma, versus 391 days in the group with fewer macrophages in the stroma (1-year survival rate, 33.9% and 55.2%, respectively). Patients were divided into 2 groups, according to the distribution of infiltrating macrophages; a High Nests Macrophage (HNM) group, in which the number of macrophages in the cancer nests was higher than in the cancer stroma (macrophages, nests > stroma) and a Low Nests Macrophage (LNM) group (nests < stroma). The HNM group had significantly better survival than the LNM group (P < .0001) (Fig. 2C). Median survival time was 440 days in the HNM group versus only 199 days in the LNM group, and the 1-year survival rate was 60.8% and 21.4%, respectively. Although mast cells in the cancer nests have been reported to contribute to a favorable outcome,23 there was no significant relationship with patient survival in this study (Fig. 2D-F). Figure 2G-I shows the relation between the number of CD8+ T cells and patient survival; there was a positive association between survival and the number of CD8+ T cells in cancer nests (Fig. 2G, P = .0008). Median survival was 388 days in the group with the higher number of CD8+ T cells in the cancer nests, versus 242 days in the group with the lower number (1-year survival rate, 52.8% and 34.3%, respectively). According to the distribution of infiltrating CD8+ T cells, patients in the High Nests CD8+ T cell (HNT) group, in which the number of tumor-infiltrating CD8+ T cells was higher in the cancer nests than in the cancer stroma (nests > stroma), had significantly better survival than those in the Low Nests CD8+ T cell (LNT) (nests < stroma) group (P = .0070) (Fig. 2I). Median survival time was 440 days in the HNT group, versus only 199 days in the LNT group, and 1-year survival rate was 53.4% and 31.3%, respectively.
|Macrophages in cancer nests||.3536|
|Macrophages in stroma||.0001|
|Nests < stroma||84||199||178-220|
|Nests > stroma||115||440||370-505|
|Mast cells in cancer nests||.0551|
|Mast cells in stroma||.6246|
|Mast cell distribution||.1778|
|Nests < stroma||98||250||188-324|
|Nests ≥ stroma||101||370||304-436|
|CD8+ T cells in cancer nests||.0008|
|CD8+ T cells in stroma||.5597|
|CD8+ T cell distribution||.0070|
|Nests < stroma||83||256||224-288|
|Nests ≥ stroma||116||388||316-460|
We then classified the patients into 4 groups according to macrophage and CD8+ T cell distribution: 1) the HNM and HNT group (macrophages, nests > stroma; CD8+ T cells, nests > stroma), 2) the HNM and LNT group (macrophages, nests > stroma; CD8+ T cells, nests < stroma), 3) the LNM and HNT group (macrophages, nests < stroma; CD8+ T cells, nests > stroma), and 4) the LNM and LNT group (macrophages, nests < stroma; CD8+ T cells, nests < stroma). Median survival time was 495 days in the HNM and HNT group, versus only 196 days in the LNM and LNT group, and the 1-year survival rate was 68.4% and 12.5%, respectively. Median survival time was 342 days, and 1-year survival rate was 45.0% in the HNM and LNT group; median survival time was 221 days, and the 1-year survival rate was 27.2% in the LNM and HNT group. Patients in the HNM and HNT group had significantly better survival than patients in the other groups (Fig. 3, Table 3)
Multivariate Regression Analysis of Survival in NSCLC Patients
As immune cells in cancer nests and cancer stroma would have different biological activity in regard to tumor progression, it would be meaningful to evaluate immune cell distribution. Considering that the distributions of macrophages in cancer nests and cancer stroma may impact clinical outcome, multivariate analysis of macrophage and CD8+ T cell distribution and clinicopathological predictors of survival was performed by means of the Cox proportional hazards model (Table 4), and both macrophage distribution (P < .001) and CD8+ T cell distribution (P = .040) emerged as independent favorable prognostic indicators. Smoking status also emerged as an independent prognostic indicator (P = .033).
|Parameter||Hazard Ratio||95% CI||P|
|Age (<70 y vs ≥70 y)||1.093||0.740-1.613||.655|
|Sex (men vs women)||1.166||0.772-1.760||.897|
|PS (0 vs 1)||1.41||0.971-2.049||.071|
|Smoking (< pack years vs > pack years)||1.561||1.037-2.348||.033|
|Histology (adeno vs nonadeno)||1.031||0.742-1.432||.856|
|Macrophage distribution (nests < stroma vs nests > stroma)||0.439||0.320-0.602||<.001|
|CD8+ T cells distribution (nests < stroma vs nests > stroma)||0.723||0.530-0.985||.040|
This is the first study to investigate the relationship between the number of tumor-infiltrating macrophages, mast cells, and CD8+ T cells in tumor biopsy specimens and the clinical outcome of patients with stage IV NSCLC. Patients with higher numbers of tumor-infiltrating macrophages and CD8+ T cells in cancer nests than in cancer stroma had significantly better survival. These factors were also independent prognostic factors in multivariate analysis.
Immunologists have long considered the presence of tumor-infiltrating immune cells as evidence of a host response against the growing tumor. However, recently reports have shown that macrophages and mast cells in cancer stroma secrete several growth factors and proteases involved in angiogenesis, thereby promoting cancer progression. An experimental study has demonstrated that interaction between lung cancer cells and macrophages promotes the invasiveness and matrix-degrading activity of cancer cells,24 and infiltration by these cells has been reported to be associated with an unfavorable outcome in several kinds of cancers.25-27 Conversely, macrophages in cancer nests produce cytotoxic cytokines, such as IL-1α, IL-1β, IL-6, and TNF-α, which may protect against tumor progression.28 Considering the results of this study showing that the distributions of macrophages in cancer nests and cancer stroma impacted outcome of stage IV NSCLC, the macrophages in cancer nests and cancer stroma may have different biological activity in regard to tumor progression. Welsh et al demonstrated that higher numbers of macrophages in cancer stroma and lower numbers of macrophages in cancer nests were unfavorable prognosis factors in surgically resected NSCLC,23 and their findings are in part consistent with the results of our study. No relationship between the numbers of macrophages in cancer nests and patient survival was found in our study. This can be explained by the difference between the specimens from operable cases of NSCLC (stage I-III) and stage IV cases.
CD8+ T cells with cytotoxic activity play an important role in antitumor immunity. CD8+ T cells can circumvent many of the barriers inherent in cancer-induced stroma, while optimizing T-cell specificity, activation, homing, and antitumor function.29 The presence of tumor-infiltrating CD8+ T cells has previously been reported to be associated with a favorable outcome, the same as in our own study.5-12, 30 Patients in the HNM and HNT group had significantly better survival (median survival was 495 days; 1-year survival rate was 68.5%) than patients in the HNM group (median survival, 440 days; 1-year survival rate, 60.8%; Fig. 2C) and patients in the HNT group (median survival, 388 days; 1-year survival rate, 53.4%; Fig. 2I). There were also many long-term survivors in the HNM and HNT group, which notably had a 3-year survival rate of 19.1%. Because aggregation of tumor-infiltrating macrophages in cancer nests has been reported to have a beneficial effect by activating cytotoxic T cells,31 the macrophages and CD8+ T cells in cancer nests should exert synergistic antitumor effects. Infiltration of CD8+ T cells in gastric carcinoma is actually directly correlated with macrophage infiltration, suggesting that macrophages play an important part in the activation of T cells and subsequent tumor cell destruction.31
Whether there is any correlation between the presence of tumor-infiltrating mast cells and cancer progression is a matter of controversy. In previous studies, mast cells were found to have antitumor functions, including serving as natural cytotoxic effectors32, 33 and antitumor compounds,34 and to be a favorable prognostic factor in surgically resected NSCLC, breast cancer, and colorectal cancer.35-37 Although mast cells produce histamine, basic fibroblast growth factor, heparin, chymase, and tryptase, which have been shown to promote cancer progression, including in surgically resected NSCLC, gastric cancer, and endometrial cancer,18, 38 no significant relation to survival was found in this study.
Accumulation of immune cells in tumor tissue either before or during chemoimmunotherapy has been reported to be associated with a better clinical response and improved survival.39-41 The effect of chemoradiotherapy in esophageal cancer is correlated with the number of CD8+ T cells in the tumor of each patient, and the patterns of gene expressions for T cell activation and for tumor vessel formation may become good markers for identifying potential long-term survivors.11 However, in the present study, no correlations between numbers of macrophages, CD8+ T cells, or mast cells and response to first-line chemotherapy were found (Table 2). The results of our study suggested that patients with a favorable or unfavorable prognosis could be identified by the status of tumor-infiltrating macrophages and CD8+ T cells in tumor biopsy specimens before receiving chemotherapy regardless of chemotherapy response. Cancer patients can mount cellular immune responses against their own tumor cells, and hosts can respond to a large compendium of tumor-associated antigens and epitopes. The natural immune system within the cancer microenvironment may affect its ability to control malignant disease beyond the response to chemotherapy. The only treatment currently available for metastatic NSCLC is chemotherapy, but patients with a poor prognosis, and patients with a predominant distribution of macrophages and CD8+ T cells in the cancer stroma, require some additional therapy to prolong life. For example, elimination of macrophages from the cancer stroma or transfer of CD8+ T cells to cancer nests might be beneficial in prolonging the life of stage IV NSCLC patients in these unfavorable groups.
In conclusion, we found that predominant distribution of macrophages and/or CD8+ T cells in cancer nests as opposed to cancer stroma was correlated with a favorable prognosis in stage IV NSCLC patients. Patients with advanced NSCLC require additional therapy, because the response rate to chemotherapy has been poor (only 30%-40%), and the median survival time of patients with metastatic NSCLC is approximately 8 months to 10 months.20, 21 The results of our study indicate the possibility of using macrophages and CD8+ T cells to treat advanced NSCLC in the future.42 Decreasing the number of tumor-associated macrophages in the tumor stroma in an animal model of breast cancer effectively altered the tumor microenvironment involved in tumor angiogenesis and progression and markedly suppressed tumor growth and metastasis.43
Thus, a more accurate insight into the role of macrophages and CD8+ T cells in tumors and consideration of the local microenvironment in regulating the functions of these cells is needed and has important implications for the design of future clinical trials of adjuvant therapy, as well as for our understanding of the immunopathobiology of stage IV NSCLC.
We thank Tetsuya Nakatsura for Immunological advice and Mai Okumoto and Hiroko Hashimoto for technical assistance.