Association of CD8⁺ T cell infiltration in oesophageal carcinoma lesions with human leucocyte antigen (HLA) class I antigen expression and survival

Abstract

Oesophageal cancer is one of the most aggressive tumours with a poor prognosis. However, little is known about the immune response in the tumour microenvironment. To investigate the role of immunosurveillance in the clinical course of oesophageal squamous cell carcinoma, 98 formalin-fixed, paraffin-embedded primary tumours were analysed using immunohistochemical methods for human leucocyte antigen (HLA) class I heavy chain and β2-microglobulin expression and for CD4-, CD8- and CD57-positive cell infiltration. HLA class I expression of tumour cells was correlated positively with infiltration of CD8⁺ T cells into the cancer nest, but not with the clinical course of disease. However, CD8⁺ and CD4⁺ T cell infiltration was correlated with prognosis. These results suggest that tumour antigen-specific cellular immune response plays a role in the clinical course of the disease and that HLA class I antigen expressed on tumour cells contribute to this association most probably by mediating the interactions between tumour cells and CD8⁺ T cells.

Introduction

Oesophageal cancer is one of the most aggressive tumours with a poor prognosis; it is the sixth leading cause of cancer-related death worldwide [1]. Despite recent advances in early screening and therapies, including surgical resection, radiation and chemotherapy, the prognosis remains poor mainly because of frequent distant metastases or local recurrence [2,3]. The overall 5-year survival rate of oesophageal cancer still ranges between 20 and 40% [4–7]. These findings emphasize the need to characterize the mechanisms underlying the poor prognosis of oesophageal cancer and to utilize this information to develop novel therapeutic strategies. Convincing evidence indicates that malignant transformation of tumour cells is associated with changes in the expression of molecules such as human leucocyte antigen (HLA) antigens and natural killer (NK) cell-activating ligands, which mediate the interaction of tumour cells with the host's immune system [8–11]. These changes appear to have clinical significance, as they are associated with the clinical course of the disease in at least some malignant diseases [12–17]. To the best of our knowledge, no information is available about the expression of HLA class I antigens in oesophageal cancer lesions. The lack of this information has a negative impact on our understanding of the role of immunological events in the pathogenesis and the clinical course of oesophageal cancer. Therefore, in the present study we have investigated the expression of HLA class I antigens in primary oesophageal cancer lesions, utilizing immunohistochemical methods. The results of immunohistochemical staining have been correlated with the histopathological characteristics of the lesions and with the clinical course of the disease.

Materials and methods

Patients

Ninety-eight patients (84 males and 14 females; mean age 62·9 years) with primary oesophageal squamous cell carcinoma (ESCC) who underwent radical oesophagectomy between September 1989 and May 1999 at Hokkaido University Hospital or at the affiliated hospitals (Department of Surgery, Teine Keijinkai Hospital and Department of Surgery, Hokkaido Gastroenterology Hospital) were studied. No distant metastasis was detected in any patient upon preoperative examination. No patients had received prior chemotherapy and/or radiotherapy. Cases of in-hospital death were excluded from the current study. The clinical typing of tumour was determined according to the tumour–node–metastasis (TNM) classification system of the international union against cancer. All specimens were fixed in 10% formalin and embedded in paraffin wax. The thickest part of each tumour was selected for evaluation. Serial 4-µm-thick sections were obtained from this part and examined by immunohistochemistry. All the informed consent processes for immunohistochemical staining were conducted in accordance with the guidelines of the Hokkaido University Institutional Review Board Authorization for this study.

Monoclonal antibodies (mAb)

The mAb HC-10, which recognizes a determinant expressed on β2m-free HLA-A10, HLA-A28, HLA-A29, HLA-A30, HLA-A31, HLA-A32 and HLA-A33 heavy chains and on all β2m-free HLA-B and HLA-C heavy chains, the β2m-specific mAb L368 were developed and characterized as described [18–20].

The human CD4-specific mAb [Histofine CD4 mouse immunoglobulin (Ig)G1 mAb], the human CD8⁺-specific mAb (Histofine CD8 mouse IgG1κ mAb) were purchased from Nichirei Corporation (Tokyo, Japan) and the human NK cell-specific mAb (Leu 7) were purchased from Becton Dickinson Immunocytometry System (San Jose, CA, USA) to identify cells originating from the neural crest, including NK cells [17,21,22].

Immunohistochemical staining

Immunohistochemical staining of tumour sections was performed using the streptavidin–biotin–peroxidase method. The staining methods have been described elsewhere [16]. In the case of staining with mAb HC-10 and L368, each slide was deparaffinized in xylene, washed with phosphate-buffered saline (PBS, pH 7·4) and dehydrated through a graded series of ethanol solutions. Following treatment with citrate buffer, specimens were subjected to autoclave for 20 min. After cooling for 5 min, endogenous peroxidase activity was blocked by 10-min incubation with 3% hydrogen peroxidase. The slides were then immunostained using the Ventana ES automated immunohistochemistry system (Ventana Medical Systems Japan, Yokohama, Japan). The mAb HC-10 and L368 were used at a concentration of 10 µg/ml and 25 µg/ml, respectively. The automated protocol was based on an indirect biotin–avidin system and used the Ventana diaminobenzidene (DAB) universal kit (Ventana Medical Systems Japan). Sections were then counterstained in haematoxylin for 1 min and mounted in Permount Microslides (Muto-Glass, Tokyo, Japan).

Scoring of HLA class I and β2 expression

The expression of HLA class I and β2-microglobulin were assessed by two investigators blinded to the patients' clinical information. The HLA class I heavy chain and β2-m expression by tumour cells was scored using the following scale: negative, heterogeneous and positive, when the percentage of stained tumour cells was 25, 25–75 and >75%, respectively. Normal lymphocytes and vessel endothelia were used in each specimen as internal controls.

Scoring of CD4⁺ T cell, CD8⁺ T cell and NK cell infiltrates

Immunohistochemistry and evaluation of immune cells were performed according to a previous report [16]. Briefly, the degree of immune cell infiltration was analysed in more than 10 independent high-power (×200) microscopic fields for each tissue sample. The five areas of most abundant distribution were selected. The number of CD4⁺ T cells, CD8⁺ T cells and NK cells was counted both in the mesenchymal stroma and within the cancer nest. The degree of immune cell infiltration was classified into two categories, abundant [CD4 (+), CD8 (+) and NK (+)] and scanty [CD4 (−),CD8 (−) and NK (−)], respectively. The threshold values used to demarcate infiltration levels were selected such that each group contained almost equal numbers of patients. For CD4, the threshold value was 1·4/microscopic field, for CD8, 1·8 and for NK cells, 0·8. Finally, the patients were classified into four groups: patients classified in both CD4 (+) and CD8 (+) were described as CD4/8 (+/+). In the same manner, other three groups were designated as CD4/8 (+/−), CD4/8 (−/+) and CD4/8 (−/−).

Statistical analysis

The χ² test was used to asses the significance of the association of HLA class I expression with clinicopathological parameters. Overall patient survival was calculated from the date of diagnosis to the date of last follow-up (censored) or date of patient death (event). Differences in survival times between patient subgroups were analysed using the log-rank statistical test. Survival probabilities were calculated using the Kaplan–Meier method. In all tests, statistical significance was set at 5%. All analyses were performed using a statistical software (StatView, version 5·0; SAS Institute Inc., Cary, NC, USA).

Results

HLA class I antigen expression in the ESCC lesions

Ninety-eight primary ESCC lesions were stained with HLA class I heavy chain-specific mAb HC-10 and β2-m-specific mAb L368. As shown in Table 1, 24 (24·5%) were classified as HLA class I antigen-negative, 26 (26·5%) as heterogeneous and 48 (49·0%) as positive. All the lesions showed positive expression with regard to β2-m, except for one lesion, which was scored as negative. Representative photomicrographs of each immunohistochemical staining are shown in Fig. 1.

Table 1.

Expression of major histocompatibility complex (MHC) class I and β2-microglobulin in 98 primary oesophageal squamous cell carcinoma (ESCC) patients

	Positive	Heterogeneous	Negative
HLA class I antigens	48	26	24
B2-microgrobulin	97	0	1

HLA: human leucocyte antigen.

Fig. 1.

Representative immunohistochemical staining patterns of formalin-fixed, paraffin-embedded primary oesophageal squamous cell carcinoma (ESCC) sections with human leucocyte antigen (HLA) class I heavy chain-specific monoclonal antibody (mAb) HC-10 and β2-m-specific mAb L368. (a–c) Normal epithelium, lack of staining and strong staining of carcinoma cells by mAb HC-10. (d,e) Lack of staining and strong staining, respectively, of carcinoma cells by mAb L368. Magnification ×400 for all pictures.

CD4⁺ and CD8⁺ T cell infiltration in tumour lesions

CD4⁺ T cells and CD8⁺ T cells were detected within the tumour lesions by immunohistochemistry. Two groups were classified by the number of CD4⁺ or CD8⁺ T cells in the cancer cell nest, respectively (Table 2).

Table 2.

Classification of patients by the number of T cells in the nest and natural killer (NK) cells in the stroma

	No. of immunoreactive cells/microscopic field, mean ± s.d. (range, no. of patients)
	Abundant	Scanty
CD8+ T cells	18·8 ± 17·3 (2·2–144·0, 48)	10·5 ± 0·3 (0–1·8, 50)
CD4+ T cells	7·4 ± 3·4 (1·8–25·8, 50)	0·6 ± 0·2 (0–1·4, 48)
NK cells	5·3 ± 4·7 (0·9–41·3, 49)	0·3 ± 0·1 (0–0·8, 49)

Ninety-eight patients with primary oesophageal squamous cell carcinoma (ESCC) were classified into two groups (abundant and scanty) containing almost equal number of patients. Mean ± standard deviation (s.d.), range and the number of patients are shown.

Correlation between MHC class I expression and the number of CD8⁺ T cells

We next evaluated the number of CD8⁺ T cells within the cancer cell nest and in the stroma according to the HLA class I expression. As shown in Fig. 2a, there was significant correlation (P = 0·0130) between the number of CD8⁺ T cells and the positive expression of HLA class I antigen in the cancer nest. Similarly, high expression of HLA class I was found with high infiltration of CD8⁺ T in the stroma, although the correlation was not statistically significant (Fig. 2b).

Fig. 2.

Correlation between the number of CD8⁺ T cells and human leucocyte antigen (HLA) class I expression in the nest in primary oesophageal squamous cell carcinoma (ESCC). The number of CD8⁺ T cells in the nest (a) or in the stroma (b) according to HLA class I expression were evaluated. Data are shown as the mean numbers of infiltrated cells ± standard deviation. P-values are shown for the χ² test.

Association of HLA class I expression with the histopathological and clinical characteristics of lesions

Association of HLA class I expression with clinicopathological characteristics of the analysed patients is summarized in Table 3. Expression of HLA class I was associated neither with the histopathological characteristics nor with the clinical variables we have analysed. Kaplan–Meier analysis revealed no significant association of the overall survival of the patients with HLA class I antigen expression patterns (P = 0·7197) (data not shown). Using Kaplan–Meier survival analysis, patients with abundant infiltration of CD8⁺ T cells showed a significantly higher survival rate than patients with low infiltration of CD8⁺ T cells (P = 0·0155), as shown in Fig. 3b. Patients with infiltration of abundant CD4⁺ T cells also showed improved survival (P = 0·0183), as shown in Fig. 3a. Similar results were obtained with the analysis within the mesenchymal stroma (data not shown). Furthermore, when patients were classified into four groups as CD4/8 (+/+), CD4/8 (+/−), CD4/8 (−/+) and CD4/8 (−/−), the group of CD4/8 (−/−) showed unfavourable survival, consistent with our previous report [16], as shown in Fig. 3d. Ninety-eight ESCC patients were also classified into two groups according to the numbers of NK cells in the stroma (Table 2). The number of NK cells in the nest was found to be very low, and therefore the analysis of NK cells was performed only in the stroma. In contrast to the analyses of CD8⁺ and CD4⁺ T cells, there was no significant correlation between patient survival and the infiltration of NK cells (Fig. 3c).

Table 3.

Clinicopathological factors of 98 oesophageal squamous cell carcinoma (ESCC) patients according to human leucocyte antigen (HLA) class I expression

	HLA class I antigen
	− and heterogeneous	+	P-value
Gender			0·3428
Male	45	39
Female	5	9
Median age (years)			0·9831
Range	62·9 ± 7·6	62·9 ± 9·0
PS			0·4106
0	42	44
1	7	4
2	1	0
pStage			0·7882
I	14	11
II	17	14
III	10	13
IV	9	10
Grade			0·2942
Well	9	15
Moderate	27	23
Poor	14	10
pT classification			0·3276
T1	21	18
T2	6	4
T3	16	23
T4	7	3
pN classification			0·8192
N0	23	20
N1	27	28
pM classification			0·9211
Mo	41	38
M1	9	10
Tumour size (cm)			0·9835
≥4·5	27	27
<4·5	23	21
Surgical margin			0·7752
Positive	5	4
Negative	45	44
Adjuvant therapy			0·3207
Yes	19	24
No	31	24
CD8 nest infiltration			0·0461
Abundant	21	28
Scanty	29	20

Fig. 3.

Association of patients' survival with infiltration of T cells in the nest or natural killer (NK) cells in the stroma. Kaplan–Meier analyses of overall survival of 98 oesophageal squamous cell carcinoma (ESCC) patients were performed with infiltration of CD4⁺ cells in the nest (a), CD8⁺ T cells in the nest (b) or NK cells in the stroma (c), or CD4/8 status in the nest (d). P-values are shown for the log-rank test.

Discussion

Recent studies have demonstrated that the immune system does protect the host from cancer development and also shapes the immunogenicity of the developing tumours, leading to a concept of cancer immunoediting [23,24]. In humans, several lines of evidence suggest the existence of cancer immunosurveillance process: (i) spontaneous immune responses against tumour were observed frequently in tumour-bearing patients that has led the identification of a number of tumour-associated antigens; (ii) correlation between the presence of immune cells into tumour tissue and favourable patient survival has been established in many types of tumours; (iii) defects or down-regulations of molecules involved in antigen presentation such as HLA class I molecules and transporter associated with antigen processing (TAP) as well as tumour antigens have been reported in many types of tumour [13,25,26]. To elucidate an escape mechanism from immunosurveillance in ESCC, we examined the expression of HLA class I and β2-m in the ESCC specimens, and determined their impact on the immunological response against tumour in patients with ESCC. We found that 51% of 98 ESCC samples showed down-regulation of HLA class I antigen. This frequency of HLA class I antigen down-regulation is similar to that of cervical cancer [25], breast cancer [26] and melanoma [27]. However, comparing this result of ESCC with the information in the literature on the analysis of HLA class I antigen expression in other types of cancer, it is higher than that of lung [28], renal and colon cancer and lower than that of prostate cancer [29]. Consistent with our previous report [16], the number of CD8⁺ T cells infiltrated into the cancer tissues correlated significantly with improved patient survival (Fig. 2b). Moreover, the expression of HLA class I in ESCC was correlated with the number of CD8⁺ T cells in the cancer cell nest (Fig. 2a).

The association between HLA class I expression and CD8⁺ T cell infiltration has also been found in laryngeal squamous carcinoma [30], pancreatic carcinoma [31] and ovarian carcinoma [32]. These results suggest that CD8⁺ T cells play an important role in cancer immunosurveillance against ESCC, and that the loss or down-regulation of HLA class I antigen is one of the major escape mechanisms of ESCC from immunosurveillance. In the present study, however, no direct correlation was found between HLA class I expression pattern and ESCC patient survival. HLA class I antigen down-regulation has been found to be a better prognostic indicator in uveal melanoma [12], breast cancer [14] and colon cancer [15] and, conversely, to be a poor prognostic indicator in laryngeal squamous cell carcinoma [13]. However, our result is concordant with previous reports with regard to the prognostic role of HLA class I expression pattern in oesophageal carcinoma [33]. The lack of a direct correlation between HLA class I expression and ESCC patient survival suggests that the clinical outcome is a total sum of the integrated multiple factors including other types of immune cells involved in the immunosurveillance of ESCC, and also affected by tumour-inherent factors (such as type of mutations) [34]. This is supported by our data showing that both CD4⁺ and CD8⁺ T cells play important roles in the improvement of patient prognosis (Fig. 3a,b), along with our previous report [16]. It is also possible that specific haplotypes or alleles of HLA have a predominant role in the T cell response against ESCC. Even though both haplotype loss and allelic loss of HLA can have profound consequences for T cell recognition [27,28], our study addressed only the issue of total HLA loss.

Despite the fact that 51% of 98 ESCC samples showed down-regulation of HLA class I expression, β2-microglobulin was expressed all the samples except for only one case (Table 1). It is reported that post-transcriptional regulation of β2-microglobulin, a critical component of the HLA class I-β2m-peptide complex, contributes to one mechanism which causes lack of HLA class I cell surface expression [35]. This dissociation in our study between the expression of the heavy chain and β2-microglobulin raises several possibilities: (i) several components involved in antigen presentation are being missed; (ii) β2-microglobulin loss is not a major HLA class I deficiency mechanism in ESCC; and (iii) HC-10 did not cover the total variety of HLA class I alleles, thereby the possibility remains that other alleles are being expressed. To resolve this problem, analysis utilizing more than one anti-HLA class I monomorphic antibody and also antibodies for antigen-processing machinery (APM) components should be performed. Future studies may reveal the major mechanism by which ESCC escape from immunosurveillance and may suggest strategies to overcome their escape mechanisms, including HLA down-regulation, as well as selecting patients who could benefit from cancer immunotherapy in ESCC.

Disclosure

None of the authors has any financial conflicts to disclose.