Forkhead box P3 regulatory T cells coexisting with cancer associated fibroblasts are correlated with a poor outcome in lung adenocarcinoma

Abstract

Recently, an association between tumor infiltrating Forkhead box P3 regulatory T cells (T_reg) and an unfavorable prognosis has been clinically shown in some cancers, but the mechanism of T_reg induction in the tumor microenvironment remains uncertain. The aims of the present study were to examine the relationship between T_reg and patient outcome and to investigate whether T_reg induction is influenced by the characteristics of cancer‐associated fibroblasts (CAF) in lung adenocarcinoma. The numbers of T_reg in both the tumor stroma and the tumor nest were counted in 200 consecutive pathological stage I lung invasive adenocarcinoma specimens. To examine whether the characteristics of CAF influence T_reg induction, we selected and cultured CAF from low T_reg and high T_reg adenocarcinoma. The number of T_reg was much higher in the stroma than in the nest (P < 0.01). Patients with high T_reg had a significantly poorer prognosis than those with low T_reg (overall survival: P = 0.03; recurrence‐free survival: P = 0.02; 5‐year overall survival: 85.4% vs 93.0%). Compared with the CAF from low T_reg adenocarcinoma, culture supernatant of the CAF from high T_reg adenocarcinoma induced more T_reg (P = 0.01). Also, CAF from high T_reg adenocarcinoma expressed significantly higher mRNA levels of transforming growth factor‐β (P = 0.01) and vascular endothelial growth factor (P = 0.01), both of which are involved in T_reg induction. Our studies suggest the possibility that CAF expressing immunoregulatory cytokines may induce T_reg in the stroma, creating a tumor‐promoting microenvironment in lung adenocarcinoma that leads to a poor outcome.

Lung cancer is the most common cause of cancer‐related death worldwide.1 Surgery currently plays an important role in the treatment of clinical stage I–IIIA non‐small lung cancer (NSCLC). Because of local recurrence and distant metastasis, however, patient outcome remains poor even after complete resection.2 Effective therapies are needed for individual patients after surgery, and a new prognostic marker for the selection of patients with a high risk of cancer recurrence is required.

Both tumor cell characteristics and patients' immune responses have been shown to affect tumor development and metastasis.3 Recent studies have shown that the accumulation of immunosuppressive lymphocytes, represented by regulatory T cells (T_reg) that suppress autoreactive T cells to maintain immunological self‐tolerance and inhibit autoimmunity, is associated with advanced tumor growth and a poor outcome in several types of malignant tumors, including lung cancer.4, 5, 6, 7, 8 Forkhead box P3 (Foxp3) is a member of the forkhead/winged‐helix family of transcriptional factors that is critically involved in the development and function of T_reg.9, 10

Cancer cells coexist with several stromal cell types that together create a cancer microenvironment. The main constituents of the stromal cell types are inflammatory cells, including lymphocytes and fibroblasts. Several recent reports have provided compelling experimental evidence indicating that the progression of tumors toward a malignant phenotype does not depend exclusively on the cell‐autonomous properties of the cancer cells themselves; it is also deeply influenced by cancer associated fibroblasts (CAF).11, 12, 13 Activated CAF contribute not only to inducing tumor progression, but also to creating the tumor microenvironment and inducing endothelial cells and other stromal cells via extracellular matrix proteins, proteases, cytokines and growth factors such as transforming growth factor (TGF)‐β, human growth factor, vascular endothelial growth factor (VEGF), and fibroblast growth factor.14 However, the correlation between tumor‐infiltrating T_reg cells and CAF that express immunoregulatory cytokines has not been thoroughly investigated.

The aims of this study were to investigate the relationship between the T_reg number and the outcome of patients with p‐stage I lung adenocarcinomas and to examine the possible correlation between T_reg induction in the tumor microenvironment and the characteristics of CAF.

Materials and Methods

Patients

The present study group comprises 200 consecutive patients with adenocarcinoma of the lung who underwent a complete resection at the National Cancer Center Hospital East, Kashiwa, Japan. All patients were diagnosed as having pathological stage I disease between January 2004 and December 2005, and all had a solitary lesion. Patients who had received preoperative chemotherapy or preoperative thoracic radiotherapy and whose tumor was diagnosed as a pure bronchioloalveolar carcinoma were excluded. All patients underwent a lobectomy or pneumonectomy for the resection of the primary lesion. We surveyed the patients at 3‐month intervals for the first 2 years and at 6‐month intervals thereafter.

The present research was approved by the Internal Review Board of the National Cancer Center Hospital East. The research consisted of a retrospective chart review in March 2012. No personally identifiable information was included.

Histopathological analysis and evaluation of clinicopathological factors

The available pathology slides from all 200 surgical specimens were coded, masked for identity, and then reviewed by two pathologists (T.K. and G.I.). The cases were reviewed according to the current (third) edition of the World Health Organization's histological classification and were staged according to the TNM classification of the seventh edition of the Union for International Cancer Control. Furthermore, per the WHO classification, we organized all the cases into the four subtypes focused on their histological predominance: bronchioloalveolar carcinoma, acinar, papillary, or solid adenocarcinoma with mucin production.

Antibodies and immunohistochemistry

After the pathologic assessment of the H&E‐stained slides of the surgical specimens, 4‐μm thick sections were made from formalin‐fixed, paraffin‐embedded specimens. Each slide was then incubated overnight at 4°C with mouse antihuman Foxp3 antibodies (ab20034 clone 236A/E7 diluted at 1:200; Abcam Inc, Cambridge, MA, USA). The slides were incubated with EnVision™ (Dako, Glostrup, Denmark) for 30 min at room temperature for visualization of bound primary antibody. They were visualized in 2% 3,3′‐diaminobenzidine in 50‐mM Tris buffer (pH 7.6) containing 0.3% hydrogen peroxidase.

Immunohistochemical scoring

All the stained tissue sections were semiquantitatively scored and evaluated independently under a light microscope by two pathologists (T.K. and G.I.) who did not know any clinicopathological information regarding the cases. Among tumor‐infiltrating lymphocytes, T_reg were detected based on the presence of positive nuclear staining for Foxp3. The absolute numbers of T_reg in the stroma and in the nest were counted in five different high‐power fields (HPF, ×400 magnification). We counted T_reg in the clear hot spot, and average cell count was defined as the T_reg number.

Statistical analysis

Overall survival (OS) was measured from the date of surgery until the date of death from any cause or the date on which the patient was last known to be alive. The recurrence‐free survival (RFS) time was measured as the interval between the date of surgery and the date of recurrence, the date of death from any cause, or the most recent date on which the patient was last known to be disease‐free. Survival curves were plotted according to the Kaplan–Meier method and were compared using the log‐rank test. Two‐category comparisons were performed using the Pearson chi‐squared test or the Mann–Whitney U‐test for quantitative data. All the tests were two sided, and P‐values <0.05 were considered statistically significant. The statistical analysis was performed using StatView version 5.0 for Windows (SAS Institute Inc., Cary, NC, USA).

Fibroblast culture

Both CAF and non‐cancer‐associated fibroblasts were prepared from human lung cancer tissue and non‐cancerous lung tissue obtained from the same specimen as previously reported.15, 16 Briefly, an approximately 5‐mm³ sample of carcinoma from each tissue specimen was cut into about 15 subdivisions and placed in an minimum essential medium alpha modification (α‐MEM; Sigma, St. Louis, MO, USA) culture containing 10% heat‐inactivated FBS and antibiotics (penicillin and streptomycin). The medium was changed every other day until the tissue was surrounded by adherent fibroblasts. After 10–20 days of growth, the fibroblasts were separated from the epithelial and endothelial cells using differential trypsinization. When the cells reached 80% confluence, they were harvested with 0.25% trypsin and 1‐mmol/L ethylene‐diamine‐tetra‐acetic acid and replated at a density of 1 × 10⁴ cells/cm².

CD4+CD25− cell purification and induction of Foxp3

For induction of CD4+Foxp3+ Treg cells, CD4+CD25−CD45RA+ naive conventional T cells were purified from the peripheral blood cells obtained from two healthy volunteers with a naive CD4+ T cell isolation kit II (Miltenyi Biotec, Bergisch Gladbach, Germany). Then, 5 × 10⁵ naive T cells were cultured in the culture supernatant from CAF with 4 × 10⁷ beads/mL CD3/CD28 beads‐T cell expander (Invitrogen, Carlsbad, CA, USA) and 10‐U/μL recombinant human interleukin (hIL)‐2 (Roche, Penzberg, Germany) for 144 h at 37°C in a humidified atmosphere of 5% CO₂. For the control experiments, naive CD4+ T cells were cultured in RPMI1640 containing 10% FBS and 1% penicillin/streptomycin/glutamine with 50‐ng/μL recombinant human transforming growth factor‐&1 (Peprotech, Rocky Hill, NJ, USA), CD3/CD28 beads‐T cell expander, and recombinant hIL‐2. The fresh medium was replaced every other day. After 6 days of culture, living T cells were collected from them.

Real‐time RT‐PCR

The cultured T cells, CAF, and non‐cancer‐associated fibroblasts were washed with PBS, suspended in 1 mL TRIzol (Invitrogen), and then stored at −80°C. Total RNA was purified from the thawed samples with standard techniques, and cDNA was synthesized with the PrimeScript RT Reagent Kit (TaKaRa, Shiga, Japan), according to the manufacturer's instructions. Real‐time RT‐PCR was performed using a Smart Cycler System (TaKaRa) and SYBR Premix Ex Taq (TaKaRa), according to the manufacturer's instructions. Next, Foxp3 mRNA expression in the incubated T cells was analyzed with RT‐PCR. We compared each Foxp3 expression level in proportion to the total cell number. To normalize the mRNA expression of cytokines such as TGF‐β, VEGF, interleukin‐10, and COX‐2, we calculated the expression ratio (CAF/non‐cancer‐associated fibroblasts) and defined this ratio relative to the cytokine expression in CAF.

Results

Immunohistochemical staining

We detected T_reg among the tumor‐infiltrating lymphocytes in the tumor stroma and the tumor nest based on the presence of positive nuclear staining for Foxp3. The average number of T_reg ranged from 0.0 to 23.0 (mean: 6.0, median: 5.5) per HPF in the stroma. The typical staining results for T_reg in the stroma are shown in Figure 1(a,b). In contrast, the average number of T_reg in the nest ranged from 0.0 to 6.4 (mean: 0.4, median: 0) per HPF. The typical staining results for T_reg in the nest are shown in Figure 1(c,d). Clearly, the absolute number of T_reg was much higher in the stroma than in the nest (P < 0.01) (Fig. 1e).

Figure 1.

Representative immunohistochemical findings for Forkhead box P3 (Foxp3+) expression in lymphocytes: (a) low T_reg in the stroma, (b) high T_reg in the stroma, (c) low T_reg in the nest, and (d) high T_reg in the nest. (e) Comparison of T_reg counts in the nest and the stroma. All analyses were performed using the Mann–Whitney U‐test. T_reg, regulatory T cells.

Correlations between number of T_reg and clinicopathological features

The study cohort included 88 men and 112 women, with a median of age of 65 years (range: 20–84 years, SD: 9.5 years). Five (2.5%) underwent a pneumonectomy and 195 (97.5%) underwent a lobectomy. The follow‐up periods ranged from 3 to 87 months (median follow‐up for surviving patients: 73 months).

We examined the clinicopathological characteristics of the cases according to the T_reg number in the tumor stroma and tumor nest. To assess the correlation between the clinicopathological characteristics and the T_reg number, we divided the patients into two groups according to their mean T_reg count (6/HPF). The group in which the T_reg count in the stroma was lower than 6/HPF (n = 107) was the low T_reg group, and the group with a T_reg count of 6/HPF or higher (n = 93) was the high T_reg group. The mean number of T_reg in the nest was 0.4/HPF, which was much lower than that in the stroma. The group in which the T_reg count in the nest was lower than 0.4/HPF (n = 124) was the low T_reg group, and the group with a T_reg count of 0.4/HPF or higher (n = 76) was the high T_reg group. For the stroma data, a large tumor diameter (P = 0.04), a high serum carcinoembryonic antigen level (P = 0.03), the presence of vascular invasion (P < 0.01), and the presence of pleural invasion (P < 0.01) were significantly more common among the high T_reg group than among the low T_reg group. For the nest data, a significant difference in vascular invasion (P = 0.03) was noted between the two groups, but no apparent differences in the tumor diameter, lymphatic permeation, or pleural invasion were seen (Table 1).

Table 1.

Correlation between T_reg number and clinicopathological factors

	Stroma			Nest
	Low Treg	High Treg	P‐value	Low Treg	High Treg	P‐value
	n = 107 (%)	n = 93 (%)		n = 124 (%)	n = 76 (%)
Sex
Men	48 (44.9)	40 (43.0)	0.89	54 (43.5)	34 (44.7)	0.88
Women	59 (55.1)	53 (57.0)		70 (56.5)	42 (55.3)
Age (year)
<65	51 (47.7)	55 (59.1)	0.12	67 (54.0)	39 (51.3)	0.77
≥65	56 (52.3)	38 (40.9)		57 (46.0)	37 (48.7)
Smoking status
Never	50 (46.7)	47 (50.5)	0.67	60 (48.4)	36 (47.4)	>0.99
Ever	57 (53.3)	46 (49.5)		64 (51.6)	40 (52.6)
Tumor diameter (cm)
≤3	84 (78.5)	60 (64.5)	0.04	90 (72.6)	54 (71.1)	0.87
>3	23 (21.5)	33 (35.5)		34 (27.4)	22 (28.9)
Serum CEA (ng/mL)
<5	83 (77.6)	59 (63.4)	0.03	95 (76.6)	47 (61.8)	0.07
≥5	24 (22.4)	34 (36.6)		29 (23.4)	29 (38.2)
Lymphatic permeation
Negative	101 (94.4)	80 (86.0)	0.05	114 (91.9)	67 (88.2)	0.45
Positive	6 (5.6)	13 (14.0)		10 (8.1)	9 (11.8)
Vascular invasion
Negative	93 (86.9)	59 (63.4)	<0.01	101 (81.5)	51 (67.1)	0.03
Positive	14 (13.1)	34 (36.6)		23 (18.5)	25 (32.9)
Pleural invasion
Negative	99 (92.5)	69 (74.2)	<0.01	109 (87.9)	59 (77.6)	0.07
Positive	8 (7.5)	24 (25.8)		15 (12.1)	17 (22.4)

Two‐category comparison was performed using the Pearson's χ². CEA; carcinoembryonic antigen; T_reg, regulatory T cells.

Comparison of T_reg number in the stroma according to predominant histological subtypes

To examine the difference in the T_reg number according to the predominant histologic subtype, we divided the cases into four groups: bronchioloalveolar carcinoma, papillary, acinar, and solid (Fig. 2). A difference in the T_reg number between the predominantly papillary tumors (n = 83, range: 0.2–27.4, mean: 6.7, median: 5.4) and the predominantly acinar tumors (n = 14, range: 1.0–14.4, mean: 6.9, median: 6.4) was not apparent (P = 0.79). However, the T_reg number in the predominantly bronchioloalveolar carcinoma tumors (n = 79, range: 0.2–22.9, mean: 4.0, median: 2.6) was much lower than that in the other histological types (P < 0.01), and that in the predominantly solid tumors (n = 24, range: 3.6–22.4, mean: 11.8, median: 11.6) was significantly higher than those in the other groups (P = 0.01). In the tumor nest, these differences could not be observed (data not shown).

Figure 2.

Comparison of T_reg number in the stroma between predominant histological subtypes. All analyses were performed using the Mann–Whitney U‐test. BAC, bronchioloalveolar carcinoma; T_reg, regulatory T cells.

Survival analysis according to number of T_reg in the tumor stroma

Figure 3(a) shows the OS curves according to the results of the T_reg count (high T_reg vs low T_reg) in the tumor stroma. The 5‐year OS rates of the high T_reg and low T_reg groups were 85.4% and 93.0%, respectively. The OS of the high T_reg group was significantly shorter than that of the low T_reg group (P = 0.03). Also, RFS curves were plotted according to the results of the T_reg count (Fig. 3b). The 5‐year RFS rates of the high T_reg and low T_reg groups were 76.5% and 87.0%, respectively. The RFS of the high T_reg group was significantly shorter than that of the low T_reg group (P = 0.02).

Figure 3.

Kaplan–Meier (a) overall survival curve and (b) recurrence‐free survival curve for patients with p‐stage I invasive lung adenocarcinoma according to the T_reg number in the stroma. The log‐rank test was used for all the survival analyses. OS, overall survival; RFS, recurrence free survival; T_reg, regulatory T cells.

Induction of T_reg from naive CD4+ T cells by CAF

Because we found that the tumor stroma, which is mainly composed of CAF, was the main location of the T_reg, we investigated whether soluble factors secreted by CAF influenced the induction of T_reg. We selected CAF from 12 cases: six from low T_reg adenocarcinomas and six from high T_reg adenocarcinomas (Fig. 4a). Next, we cultured the naive T cells according to the above‐described method using supernatant samples from these CAF. The results are shown in Figure 4(b). Compared with the low T_reg CAF, a significantly higher number of Foxp3+ T cells were induced by the supernatant from the high T_reg CAF (P = 0.01).

Figure 4.

(a) CAF and NCAF were prepared from human lung cancer tissue and non‐cancerous lung tissue obtained from the same samples. After 10 to 20 days of growth, the fibroblasts were cultured as previously reported. (b) Induction of T_reg from naive CD4+ T cells by CAF. We selected CAF and cultured the naive T cells according to the above‐described method using supernatant samples from 12 cases: six from low T_reg adenocarcinomas and six from high T_reg adenocarcinomas. CAF, cancer‐associated fibroblasts; NCAF, non‐cancer‐associated fibroblasts; Sup., supernatant; TGF‐β, transforming growth factor‐β; T_reg, regulatory T cells.

Correlations between number of T_reg and mRNA levels of immunoregulatory cytokines in CAF

We examined whether a correlation was present between the T_reg number in the tumor stroma and the CAF‐induced expression levels of immunoregulatory cytokines. RNA samples from cultured fibroblasts isolated from lung cancer tissue (CAF) in 12 adenocarcinoma patients were analyzed by RT‐PCR. The results are shown in Figure 5. The relative TGF‐β expression levels (median ± SD) induced by the CAF were 1.8 ± 0.4 in the high T_reg group and 0.9 ± 0.5 in the low T_reg group. In the same way, the relative VEGF expression levels induced by the CAF were 2.6 ± 1.1 in the high T_reg group and 1.0 ± 0.4 in the low T_reg group. The relative interleukin‐10 expression levels induced by the CAF were 1.9 ± 3.2 in the high T_reg group and 1.0 ± 1.2 in the low T_reg group. The relative COX‐2 expression levels induced by the CAF were 1.9 ± 3.2 in the high T_reg group and 1.0 ± 1.3 in the low T_reg group. No apparent difference in the expression levels of interleukin‐10 (P = 0.52) or COX‐2 (P = 0.78) were seen, but significant differences in the TGF‐β (P = 0.01) and VEGF (P = 0.01) levels were detected between the two groups.

Figure 5.

Relative expression levels of immunoregulatory cytokines expressed by CAF (CAF/NCAF). Total RNA was purified from thawed samples, and cDNA was synthesized. All the analyses were performed using the Student's t‐test. CAF, cancer‐associated fibroblasts; IL‐10, interleukin‐10; NCAF, non‐cancer‐associated fibroblasts; TGF‐β, transforming growth factor‐β; T_reg, regulatory T cells; VEGF, vascular endothelial growth factor.

Discussion

The accumulation of tumor‐infiltrating T_reg has been reported to be an unfavorable prognostic marker in several types of carcinomas.4, 5, 6, 7, 8 Petersen et al. reported that infiltrating T_reg were associated with the recurrence of pathological stage I NSCLC, but they did not mention the influence of T_reg on overall survival.7 We first found that the T_reg number in the tumor stroma was a significant indicator of a poor outcome with regard to overall, recurrence‐free and disease‐specific survival in p‐stage I lung adenocarcinoma. Although the T_reg number was not an independent prognostic factor in multivariate analysis (data not shown), the presence of T_reg in the tumor stroma may encourage an unfavorable prognosis in patients with lung adenocarcinoma.

Growing evidence suggests that T_reg play an important role in suppressing T cell‐mediated immunity in patients with cancer.17, 18 The number of T_reg in the tumor stroma was much larger than in the tumor nest. Commonly, CAF are located in the stroma and are distinctively detected in invasive carcinomas, including NSCLC. Therefore, we focused on the relationship between T_reg and the characteristics of CAF. We first hypothesized that cytokines secreted by CAF may have an important role in T_reg induction in the stroma, and we examined whether T_reg could be induced by soluble factors secreted by CAF derived from high T_reg and low T_reg adenocarcinomas. Compared with the low T_reg CAF, the high T_reg CAF induced a significantly larger number of T_reg from CD4+ naive T cells. Furthermore, we examined the expression of four kinds of immunoregulatory cytokines from 12 adenocarcinoma cases. The expressions of TGF‐β and VEGF, which reportedly induce T_reg from naive T cells in the periphery,19, 20 were significantly higher in the CAF from high T_reg adenocarcinomas than in the CAF from low T_reg adenocarcinomas. CAF are known to produce mainly TGF‐β and VEGF, compared with cancer cells.21, 22 We suggest that the CAF in high T_reg adenocarcinomas may have a higher immunoregulatory cytokine‐secreting capacity, leading to T_reg induction. Although we did not evaluate the influence of cytokine expression from tumor cells and tumor‐associated macrophages in the present study. Saji et al. confirmed that tumor‐infiltrating stromal cells were major sources of TGF‐β based on the results of an immunohistochemical analysis in NSCLC.23 Thus, the characteristics of CAF may have a great influence on tumor progression via the recruitment of other types of tumor‐promoting stromal cells. There have been no reports demonstrating the correlation between the T_reg induction/recruitment and characteristics of CAF. It is well‐known that cytokines, other than TGF‐β and VEGF, are also concerned with the T_reg induction. Examining the gene expression profile through microchip analysis will be helpful for elucidating which cytokines are upregulated in high T_reg CAF.

Recently, Miyao et al. reported that there was a minor population of non‐regulatory Foxp3+ T cells exhibiting promiscuous and transient Foxp3 expression that were induced from the naive T cells in the peripheral lymph nodes.24 In order to demonstrate whether these induced Foxp3+ T cells have immunoregulatory function, we would need to divide induced T_reg into groups according to CD25 expression levels and compare the immunosuppressive ability thereafter. In the current study, we did not examine the influence of chemokines. In the tumor microenvironment, there are many kinds of stromal cells such as macrophages and monocytes. It is possible these stromal cells educate CAF, which could create some chemokines and the T_reg‐abundant microenvironment.

In lung cancer, Tao et al. reported the influence of T_reg in tumor stroma on OS and RFS, but did not mention the correlation between the T_reg count and tumor malignant parameters.6 In the current study, we found that the high T_reg group, more frequently than the low T_reg group, had predictors of a poor outcome such as a large tumor diameter, vessel invasion, and pleural invasion. These differing results may have occurred because we examined T_reg in all stage of NSCLC, including squamous cell carcinomas and large cell carcinomas.

Additionally, the impact of the histological features of lung adenocarcinomas on T_reg accumulation has not been previously reported. In this study, the T_reg count was highest among patients with a predominantly solid adenocarcinoma subtype. Lung adenocarcinoma patients with a solid adenocarcinoma component are known to have a poorer prognosis than patients without this component.25 The tumor microenvironment of a solid component probably recruits and induces more T_reg than other histological subtypes, enabling both the tumor cells to evade the immune system and the tumor to progress easily. Thus, we suggest that tumor cells acquire the ability to survive and metastasize as a result of a tolerance in antitumor immunity induced by T_reg in the tumor stroma as the tumor progresses.

In conclusion, we showed that lung adenocarcinoma with a large number of T_reg in the tumor stroma was associated with a poor outcome among patients with p‐stage I lung adenocarcinoma after complete resection. Additionally, CAF overexpressing immunoregulatory cytokines, such as TGF‐β and VEGF, may play an important role in T_reg induction. Recently, numerous reports have described the treatment of patients with colorectal cancer, breast cancer, or lung cancer using humanized monoclonal anti‐VEGF antibody therapy (bevacizumab). To confirm the effectiveness of anti‐VEGF antibody as an adjuvant therapy in patients with abundant T_reg in the tumor stroma, further studies are needed to elucidate the relationship between T_reg and CAF. Understanding this relationship will enable the creation of efficacious follow‐up plans and improved therapeutic options for patients.

Disclosure Statement

The authors have no conflict of interest.

(Cancer Sci 2013; 104: 409–415)