Prognostic Effect of Epithelial and Stromal Lymphocyte Infiltration in Non–Small Cell Lung Cancer Free

Despite medical advances, lung cancer remains the leading cause of cancer death mainly due to delayed diagnosis resulting in the low resection rate of 20% (1, 2). The major histologic subtypes of lung cancer are non-small cell lung cancer (NSCLC; 80%) and small cell lung cancer (20%; ref. 2). The only chance for cure is surgical treatment in early stages of disease. However, as much as 60% of stages IB to IIIA patients relapse after surgery and die of metastasis (2).

Activation of the adaptive immune system may suppress malignant cells, whereas activation of various types of innate immune cells may promote tumor growth (3). The adaptive immunity is orchestrated by antigen-specific T and B lymphocytes and inhibits tumor growth through direct killing by CTLs as well as a combination of cytokine- and antibody-mediated tumor cell lysis (3). Recently, it became clear that analysis of the tumor stroma is paramount, as many of these antitumor effects act mainly on the tumor stroma (4), and cancer infiltration by tumor-reactive T lymphocytes is required for efficient tumor eradication (5). However, cancer cells can escape the immune surveillance by many mechanisms such as suppression of cytotoxic T cells by regulatory T cells and by accumulation of myeloid suppressor cells (5–7).

Tumor-infiltrating lymphocytes can be divided into three groups: (a) lymphocytes within cancer cell nests (epithelial lymphocytes), (b) lymphocytes in the central cancer stroma (stromal lymphocytes), and (c) lymphocytes present along the invasive margins (peritumoral lymphocytes; ref. 8). In NSCLC, about two-thirds of inflammatory cells are lymphocytes, among these 80% are T cells (9), expressing various activation antigens (10). CD8⁺ cells in malignant tumors have been associated with a better survival in small cell lung carcinoma, carcinomas of the endometrium, bile duct, colon, esophagus, and urothelium, as well as uveal melanoma and follicular lymphoma (11–18). It has been shown that NSCLC contain higher proportions of CD8⁺ T lymphocytes than their corresponding peripheral blood samples (10), and a higher number of tumor-infiltrating CD8⁺ lymphocytes is associated with tumor cell apoptosis in NSCLC (19).

The role of CD8⁺ lymphocytes in NSCLC is, however, controversial and most of the studies contain either relatively few cases and/or neglect the stromal component. In addition, CD4⁺ T and B lymphocytes are double-edged swords in tumor immunity, and their role is controversial in many cancers including NSCLC. Pelletier et al. (20) found peritumoral B, and not T, lymphocytes to be related to a better survival. Wakabayashi et al. (8) found stromal CD4⁺ cells to be a good prognostic indicator. Hiraoka et al. (21) found only the concurrent infiltration by CD8⁺ and CD4⁺ T lymphocytes in the tumor epithelial cells to be associated with a better survival, whereas Nakamura et al. (22) showed no prognostic significance, neither for the total intraepithelial T cells nor for its helper and cytotoxic subtypes, in NSCLC.

To elucidate the prognostic significance of lymphocyte infiltrate in NSCLC, we analyzed the degree of infiltration of CD4⁺, CD8⁺, and CD20⁺ lymphocytes in 335 resected NSCLC in both epithelial and stromal compartments and studied their relations to other clinicopathologic variables as well as the proportion of lymphocytes compared to CD138⁺ and CD68⁺ cells.

Patients and clinical data. Patients diagnosed with NSCLC, pathologic stage I to IIIA at the University Hospital of Northern Norway and Nordland Central Hospital from 1990 to 2004 were used in this retrospective study. In total, 371 patients were obtained from the hospitals' database. Of these, 36 patients were excluded from the study due to (a) radiotherapy or chemotherapy before surgery (n = 10), (b) other malignancy within 5 years before NSCLC diagnosis (n = 13), and (c) adequate paraffin-embedded fixed tissue blocks not available (n = 13). Thus, 335 patients were included in this study.

The median follow-up was 96 (range, 10-179) months. Two pathologists (S.A.S. and K.A.S.) reviewed all the cases; the diagnosis of carcinoma, histologic type, angiolymphatic invasion, and pathologic stage were confirmed before including any case in the study. The tumors were staged according to the International Union Against Cancer's tumor-node-metastasis classification (23). Stage IA and IB patients were grouped together as stage I, and stage IIA and IIB patients were grouped as stage II. Histologic classification was done according to the WHO guidelines (24).

The National Data Inspection Board and the Regional Committee for Research Ethics in Northern Norway approved this study.

Microarray construction. Tissue microarrays were constructed as reported previously (25, 26). Studies suggest that punching multiple 0.6 mm cores from different regions captures the heterogeneity of the tumors more accurately than single 2 to 4 mm core (25). Hence, we chose using four 0.6-mm cores that were selected to be as representative as possible, after reviewing all the original sections of the tumor and taking the fact of heterogeneity in consideration. In addition, all the surface area of the four cores was counted. Two areas of viable invasive carcinoma tissue (neoplastic epithelium) and two from the surrounding, central tumor stroma were selected and marked on the donor blocks. The tissue microarrays were obtained using a tissue-arraying instrument (Beecher Instruments) from two epithelial areas and two stromal areas and transferred to the recipient blocks. Lung tissue without signs of pathologic changes localized distant from the primary tumor as well as lung tissue samples from 20 normal lungs were used as control. To include all the 1,340 cores plus the control cores, eight tissue array blocks were constructed. Multiple 5 μm sections were cut with a Micron microtome (HM355S) and stained by specific antibodies for immunohistochemistry analysis.

In most NSCLC cases, the malignant cells proliferate along the alveolar walls of the lung and lack stroma at the invasive margin (8); in this study, we analyzed infiltration of CD4⁺, CD8⁺, and CD20⁺ lymphocytes in the neoplastic epithelial areas and in the central stromal areas.

Immunohistochemistry. Ventana Benchmark, XT automated slide stainer (Ventana Medical System) was used for immunohistochemistry. Sections were deparaffinized with xylene and rehydrated with ethanol. Antigen retrieval was done by placing the specimens in 0.01 mol/L citrate buffer (pH 6.0) and exposed to two repeated microwave heating of 10 min at 450 W. The DAKO Envision⁺ System-HRP (diaminobenzidine) kit was used as endogen peroxidase blocking. As negative staining controls, the primary antibodies were replaced with the primary antibody diluents. Primary mouse monoclonal antibodies were incubated for 16 min (CD20, clone L26 Ventana), 20 min (CD4, clone 1F6 Novocastra, dilution 1:5), 32 min (CD8, clone 1A5 Ventana), 16 min (CD68, clone KP1, Ventana), and 32 min (CD138, clone B-A38, Ventana) in room temperature. The Ventana antibodies were prediluted from the manufacturer. As secondary antibodies, biotinylated goat anti-mouse IgG and mouse anti-rabbit IgM were used. The diaminobenzidine was used to visualize the antigens. This was followed by application of liquid diaminobenzidine and substrate-chromogen, yielding a brown reaction product at the site of the target antigen. Finally, slides were counterstained with hematoxylin to visualize the nuclei. For each antibody, including controls, staining were done in a single experiment.

Scoring of immunohistochemistry. By light microscopy, the tissue sections were scored for the degree of infiltration of CD4⁺, CD8⁺, and CD20⁺ lymphocytes. The percentages of lymphocytes compared with the total amount of nucleated cells in the epithelial and stromal compartments were assessed. Arbitrary cutoff points at 1%, 5%, 25%, or 50% for each cell/compartment according to the degree of cell densities were used, as these percentages are easy to follow and reproduce in daily practice.

For CD8⁺ cells, the epithelial infiltrate was scored as low ≤5% or high >5% in the whole surface area of the two epithelial cores and was scored as low ≤50% or high >50% of the total nucleated cells in the whole surface area of the two stromal cores. CD20⁺ (B) cells were very sparse and its infiltrate was scored as low if they represent <1% of the nucleated cells or if absent, or high otherwise, in both epithelial and stromal compartments. CD4⁺ cells were scored as high if ≥5% or ≥25% of the total nucleated cells in the epithelial and stromal cores, respectively. The interstitial tissue of the nonneoplastic/normal controls show no or sparsely scattered CD20⁺ cells (0 to <1% of the total nucleated cells) and few CD4⁺ and CD8⁺ cells (0 to <5% of the total nucleated cells).

All samples were anonymized and independently scored by two pathologists (S.A.S. and K.A.S.). In case of disagreement, the slides were re-examined and the observers reached a consensus. When assessing one marker in a given core, both observers were blinded to the scores of other markers as well as to the patient's outcome. The interobserver scoring agreement between the two pathologists was tested on the current material in a previous report (27). The mean correlation coefficient (r) was 0.95 (range, 0.93-0.98). Examples of stromal and epithelial cores stained for CD8 lymphocytes are shown in Fig. 1.

Statistical analysis. All statistical analyses were done using the statistical package SPSS version 15. The χ² test and Fisher's exact test were used to examine the association between the density of the lymphocyte infiltrates and various clinicopathologic variables. Univariate analyses were done by using the Kaplan-Meier method, and statistically significant differences between survival curves were assessed by the log-rank test. Disease-specific survival (DSS) was determined from the date of surgery to the time of lung cancer death. To assess the independent value of different pretreatment variables on survival in the presence of all other variables, multivariate analysis was carried out using the Cox proportional hazards model. Only variables with a significant P value from the univariate analysis were entered into the Cox regression analysis. Probability for stepwise entry and removal was set at 0.05 and 0.10, respectively. The significance level was set at P < 0.05.

Clinicopathologic variables. Demographic, clinical, and histopathologic variables are shown in Table 1. Patients' age ranged between 28 and 85 years (median, 67 years), and 75% of the patients were males. The NSCLC tumors comprised 191 squamous cell carcinomas (SCC), 95 adenocarcinomas, 31 large cell carcinomas, and 18 bronchioloalveolar carcinomas. Due to nodal metastasis and/or nonradical surgical margins, 59 patients received postoperative radiotherapy. There were 232 lymph node–negative and 103 positive cases (76 N₁ and 27 N₂).

Univariate analysis. Performance status, pathologic stage, T status, N status, differentiation, surgical procedure, vascular infiltration, and postoperative radiotherapy were all significant indicators for disease-free survival in univariate analysis.

Lymphocyte infiltration. Tumor-infiltrating lymphocytes were observed in both epithelial and stromal compartments and were generally more abundant in the stroma. Tumor inflammatory cells were mainly lymphocytes (∼65%) compared with 20% to 25% CD68⁺ cells and 5% to 10% plasma cells. Plasma cells were more abundantly found in the stroma of SCCs compared with other histologic types (high expression; SCCs 61%, adenocarcinomas 35%, bronchioloalveolar carcinomas 29%, and large cell carcinomas 50%; P < 0.001). Similar results were obtained for the epithelial compartment.

In univariate analyses, the group of patients with high epithelial CD8⁺ lymphocytes had a significantly better DSS than those with low CD8⁺ cells (P = 0.023; Table 2; Fig. 2). By subgroup analysis, this significance was seen only in patients without lymph node metastasis (N₀; P = 0.018), whereas patients with lymph node metastasis showed a tendency for better DSS but did not reached the significant level. Further stratifying the cases based on histology revealed that the significance was limited to SCCs (P = 0.011). Epithelial CD20⁺ lymphocytes showed a similar significance (P = 0.023). This significance was limited to SCCs (P = 0.030) and most significant in T₂ cases (P = 0.017). Epithelial CD4⁺ cells showed no significant correlation with DSS even after analysis of the cases in subgroups.

Regarding stromal lymphocytes, a high number of CD8⁺ cells was significantly associated with an improved DSS (P = 0.002; Table 2; Fig. 2). This was limited to large cell carcinomas and SCCs. Stromal CD4⁺ lymphocytes showed a highly significant correlation with DSS (P < 0.001), strongest in N₁ cases. Similarly, CD20⁺ cells in the stroma showed a significant association with DSS (P < 0.001). Both stromal CD20⁺ and CD4⁺ significance were limited to SCCs. Concurrent infiltrates of epithelial CD4⁺ and CD8⁺ cells failed to show prognostic significance, whereas concurrent infiltrates of stromal CD4⁺ and CD8⁺ cells was a significant and strong positive indicator (P < 0.001; Fig. 3). Furthermore, a high level of stromal CD8⁺ lymphocytes was associated with a lower incidence of angiolymphatic invasion (P = 0.032). However, neither epithelial nor stromal CD4⁺, CD8⁺, or CD20⁺ lymphocytes showed any correlation with smoking, performance status, sex, or tumor differentiation. In addition, there were significant correlations between the amounts of CD4⁺ and CD8⁺ cells in the stroma (r = 0.18; P = 0.001) and in the tumor epithelium (r = 0.274; P < 0.001).

Multivariate analyses. All significant demographic, clinicopathologic, and lymphocyte infiltrate variables from the univariate analyses were entered into the multivariate Cox regression analysis. These data are presented in Table 3. A high number of stromal CD4⁺ (P = 0.002) and stromal CD8⁺ (P = 0.043) lymphocytes were independent prognostic factors for DSS.

Tumor-infiltrating lymphocytes are considered to be an indication of the host immune reaction to tumor antigens (28). In this study, we evaluated whether there is an association between the amount of epithelial and stromal CD4⁺, CD8⁺, and CD20⁺ lymphocytes and the clinical course in 335 primary resected NSCLC. Our data suggest that high intensities of stromal CD4⁺ cells as well as epithelial and stromal CD8⁺ and CD20⁺ cells are significantly associated with an improved survival in NSCLC. In fact, the strong association between survival and the stromal intensities of CD4⁺ and CD8⁺ were independent of the other clinicopathologic prognostic factors. Furthermore, a high stromal infiltration of CD8⁺ lymphocytes was associated with a lower incidence of angiolymphatic invasion in our cohort. As expected, the patients' survival was highly linked to stage of disease.

The fact that our patients were diagnosed over a 15-year period (1990-2004) may have introduced a bias due to altered diagnostic techniques (e.g., introduction of spiral computed tomographic scan and positron emission tomographic scan). However, the cited disease stage is based on the pathologic tumor-node-metastasis (examination of the surgical specimen), which should be minimally altered during the period. Adjuvant chemotherapy was not introduced in Norway during this period.

Our findings regarding CD8⁺ cells suggest that these cells can modify tumor stroma and/or epithelium in a way that reduces the disease progression and metastatic potential. In studies on adaptive immunotherapy and tumor vaccination in colorectal carcinoma, the outcome was superior among patients whose primary tumor contained high level of CD8⁺ lymphocytes (29). Tormanen-Napankangas et al. (19) found that a higher number of epithelial CD8⁺ lymphocytes is associated with tumor cell apoptosis in NSCLC, but this finding did not translate to a survival benefit, whereas Trojan et al. (30) could not show any influence of epithelial CD8⁺ cells and IFN-γ/CD8 ratio on overall survival in their study of 31 NSCLC patients. However, in a laser microscopic study on 16 randomly selected stage I NSCLC, Verdegaal et al. (31) found a vigorous antitumor immune response mediated by CD8⁺ T cells despite HLA haplotype loss in 69% of the tumors. They also found 90% of the CD8⁺ cells to be activated cytotoxic cells. The positive prognostic effect of CD8⁺ cells in NSCLC despite their tendency to loose HLA antigens may be explained by Kikuchi et al. (32) in a study of 161 NSCLC tumors. A negative HLA was observed in 54, weak in 57, and strong in 50 cases. The density of cancer-infiltrating CD8⁺ cells in HLA-I-negative tumors was significantly decreased compared with that in HLA-I strong tumors (P < 0.01).

Epithelial and stromal CD8⁺ cells, in our study, correlated strongly with outcome in patients without lymph node metastases, a finding that may indicate a key role of these cells in tumors restricted to the lung tissue. In addition, the detection of a significant association between increased stromal CD8⁺ cells and low incidence of angiolymphatic invasion may indicate a protective role of these cells in preventing tumor cell infiltration into the vascular spaces. This may suggest that the positive prognostic effect of CD8⁺ cells may be mediated, at least in part, by suppressing micrometastases, which is a major cause of cancer-related death following surgical radical resection. A similar suggestion was made by Chiba et al. (33) in their study on colon cancer.

Moreover, our data showed that stromal CD4⁺ T cells are associated with a favorable DSS in NSCLC. A similar result has been reported by Wakabayashi et al. (8) who studied epithelial and stromal CD4⁺ and CD8⁺ cells in 178 NSCLC patients and observed only CD4⁺ in cancer stroma to be associated with a better prognosis. This finding points to the fact that CD4⁺ lymphocytes may be needed for initiating and maintaining anticancer immune responses. In the absence of CD4⁺ cell help, specific CD8⁺ lymphocytes can become lethargic (34) or be deleted (35). In fact, CD4⁺ cells are essential for CD8⁺ transformation into long-lived functional effector cells (36). This positive interaction between CD4⁺ and CD8⁺ lymphocytes in our cohort is shown by the observation of a significant correlation between the degree of infiltration of the two cell types and by the finding of an improved DSS in cases with concurrent high stromal CD4⁺ and CD8⁺ cell counts when compared with heterogeneous high/low or low/high and homogenous low/low counts. In addition, CD4⁺ cells can inhibit tumor growth in the absence of CD8⁺ cells by direct lyses or recruiting other cells (3). However, CD4⁺ cells are a mixture of CD4⁺/CD25⁺Foxp3⁺ (regulatory T cells) and CD4⁺/Foxp3^- cells. The first subtype is known to suppress tumor immunity and enhance cancer growth and metastases (7, 29). In animal models, removal of CD4⁺/CD25⁺ T cells improved immune-mediated tumor clearance and enhanced the response to immunotherapy (37, 38). Our results suggest that in NSCLC the net effect of CD4⁺ cells is toward supporting cancer eradication. Badoual et al. (39) found that even CD4⁺ regulatory T cells were, paradoxically, associated with a superior locoregional control (but not survival) in head and neck SCC. This may also explain our results; more studies involving CD4⁺ subtypes in NSCLC are needed to test this hypothesis.

Epithelial CD4⁺ cells failed to show a prognostic significance. Hiraoka et al. (21) studied epithelial CD8⁺ and CD4⁺ cells in 109 NSCLC patients and found only the concurrent high levels of epithelial CD8⁺ and CD4⁺ cells to be a significant independent indicator for a better survival. A similar analysis in our cases failed to reveal a significant prognostic relevance.

In our cohort, the independently positive prognostic effect of CD4⁺ and CD8⁺ cells in the stroma points to the significance of the stromal component in modulating cancer cells. Antitumor cytokines contributing to tumor suppression act mainly in the stroma, whereas killing mediated by molecules such as perforin or Fas ligand acts mainly on tumor cells and to a lesser extent on the stroma (4). Similarly, in an adoptive T-cell transfer model, Spiotto et al. (40) showed that tumor rejection requires antigen presentation to CD8⁺ cells by cancer cells and cross-presentation by bone marrow–derived and non-bone marrow–derived stromal cells. In fact, T-cell-produced cytotoxic molecules and cytokines destroy the stromal cells, thereby withdrawing essential resources and leading to tumor infarction and subsequent T-cell-mediated elimination of residual tumor cells (41). Treating well-established tumors expressing low levels of antigens with local irradiation or chemotherapy causes sufficient release of antigens to sensitize stromal cells for destruction by cytotoxic CD8⁺ cells (42). In addition, the antitumor effect of cytotoxic lymphocytes may be theoretically strengthened by strategies inhibiting both the differentiation block of CD8⁺ cells and the early Fas expression that leads to apoptosis of lymphoid cells.

Studying several prognostic factors in 113 NSCLC tumors, including “peritumoral” B (CD20⁺) and T (CD43⁺) lymphocytes, Pelletier et al. (20) found B cells to be associated with a better survival (P = 0.04). Our study showed a similar effect for stromal, as well as epithelial, B cells in univariate analysis. B cells are precursors of antibody-mediated immunity, which may contribute in tumor eradication (43). Additionally, the presence of tumor-specific B cells may be a mirror of the host's overall immunity in NSCLC (20). On the other hand, the precipitation of immunoglobulin in the tissue may have a stimulatory effect on the innate immune cells, which may enhance tumor growth by secreting mediators stimulating angiogenesis (3). In fact, de Visser et al. (44) showed in an animal model that B lymphocytes are required for establishing chronic inflammatory states that promote de novo carcinogenesis.

In conclusion, a high number of stromal CD4⁺ and CD8⁺ cells are independent positive prognostic indicators for resected NSCLC patients. It may be hypothesized that the majority of CD8⁺ lymphocytes in NSCLC tumors are activated cytotoxic cells mediating a vigorous antitumor immune response and that CD4⁺ cells play a central role in this immune response. Before these findings can be used for novel targeted therapy strategies, the underlying mechanisms need to be further characterized.

No potential conflicts of interest were disclosed.