Detection of circulating tumor lysate–reactive CD4⁺ T cells in melanoma patients

Introduction

There is increasing evidence for a T-cell response against tumor antigens in some cancer patients and that tumor-reactive CD4⁺ or CD8⁺ T cells may circulate and be detectable in samples of peripheral blood [12]. Circulating tumor-reactive T cells might be a clinically useful source of information regarding the immunological response to cancer, as a diagnostic or prognostic tool, or as a monitor of treatment, and such T cells could possibly also be isolated, cultured [6], and reinfused for diagnostic or therapeutic purposes [9].

Previous studies have mainly focused on CD8⁺ T cells which are particularly interesting due to the direct tumor cell killing effect of CTLs [10]. However, it is increasingly recognized that CD4⁺ T cells may play a major role in tumor immunology. CD4⁺ T cells provide help during priming to achieve full activation and effector function of tumor-specific CTLs [7], and CD4⁺ T cells are required for CD8⁺ T cells to establish long-term specific immunity against tumors [29]. Moreover, in a mouse model, CD4⁺ effector T cells can suppress tumor completely independent of CD8⁺ T cells and in the absence of MHC class I and II molecules on the tumor cells [7].

The presence of antigen-specific circulating T cells in patients can be shown directly or indirectly. Direct detection can be done using tetramers of MHC class I and peptide [2, 4]. This approach, however, does require prior knowledge of the immunogenic peptide and its restricting allele [31]. Also, it is important to take into consideration that the tetramer technique in principle detects all T cells which bind the peptide-MHC complex, irrespective of the functional status of the cell [14, 20].

Methods for indirect detection of antigen-specific T cells usually employ culturing of T cells. However, assays involving long-term antigen and/or IL-2 stimulation are not ideal for assessing the in vivo situation. Selectively expanded T cell clones may not be representative [1, 8], and repeated or prolonged in vitro stimulation may result in quantitative and qualitative changes [28], including the possibility of in vitro priming and activation of naïve T cells. Alternatively, assays using short-term in vitro antigen stimulation followed by rapid detection of T-cell cytokine production may be applied. These include the ELISpot assay [16], quantitative PCR analysis for cytokine messenger RNA [11, 20], and flow cytometric detection of intracellular cytokines [19]. The flow cytometric technique is generally considered to be more sensitive than the ELISpot assay and enables the concurrent staining of individual cells with antibodies to multiple cytokines, surface markers, or intracellular proteins [31].

Most previous studies have used well-characterized tumor antigens for T-cell stimulation. However, tumor antigens expressed by human cancers are numerous [30] and probably for a large part unknown, and the significance of finding peripheral T cells reactive to one or a few defined tumor antigens is doubtful with respect to assessment of the individual patients’ polyclonal immune response to the tumor [31]. In the present study we used lysate made from a cultured allogeneic melanoma cell line as tumor antigen source. Allogeneic tumor lysate potentially contains a large number of specific, activating tumor antigens (known, as well as unknown), and is at the same time easily prepared and handled, not requiring access to autologous tumor. For assessment of a T-cell response, we used short-term coculture of lysate and fresh whole blood with costimulatory factors followed by flow cytometric analysis of CD4⁺ T cells for intracellular IFN-γ production.

Material and methods

Results

Clinical data and a summary of results of the flow cytometric analyses of 19 melanoma patients and 7 healthy individuals are shown in Table 1. Four patients were in AJCC stage I, eight in stage III, and seven in stage IV. The median age was 57 years (range 22–77 years). Eleven patients had clinically overt disease at the time of blood test. The remaining were clinically disease free and for those, the interval between radical operation and blood test ranged from 16 days to 63 days with an average of 38 days. For patients with metastatic disease (i.e., stages III and IV), the interval from histological diagnosis of melanoma to the blood test ranged from 14 days to 1,364 days with a median of 680 days.

The CD4⁺ T-cell response to tumor lysate for healthy donors and for melanoma patients grouped according to stage of disease is shown in Fig. 1. For each patient two samples were analyzed: one to which tumor lysate had been added and one control sample without lysate. The patient’s CD4⁺ T-cell response to the lysate was assessed by subtracting the percentage of IFN-γ⁺ cells in the control sample from the percentage of IFN-γ⁺ cells in the lysate-stimulated sample. In general, tumor lysate appeared to have an inhibitory effect, because if no or a few CD4⁺ cells were induced to produce IFN-γ by tumor lysate in a given sample, the calculated lysate response would most likely become negative. This effect was particularly evident in samples from individuals presenting a substantial frequency of spontaneous IFN-γ-producing cells (see Table 1). On the other hand, the positive responses found in patients, presumably overcame this apparent suppression.

The frequency of lysate-reactive CD4⁺ T cells in responding melanoma patients ranged from 0.0037% to 0.080% with a median of 0.024% of the CD4⁺ cell population. None of the healthy subjects showed a response, whereas 12 out of 19 melanoma patients (63%) had a response. Response was observed most often in patients with metastatic melanoma—in 11 of 15 (73%)—and especially in stage III patients (88%). One responder was found in four individuals treated by radical excision of stage I melanomas. However, the response was quantitatively weak (approximately half the value of the weakest responder with metastatic disease) and possibly of marginal significance. The responding patients with metastatic malignant melanoma had a significantly longer interval between diagnosis and blood test compared with the nonresponders (mean 840 days versus 239 days). No clear association between CD4⁺ T-cell response and age or sex was observed.

During follow-up, no recurrences or deaths occurred among stage I patients, whereas six patients with metastatic melanoma expired from their disease. Among stage IV patients, two out of three nonresponders rapidly died, whereas none of four responders has expired. The actuarial 1-year-survival for stage III and IV melanoma patients was 88% among responders and 66% among nonresponders. These differences were not statistically significant, however. The majority of patients with advanced disease received IL-2–based immunotherapy, however, resulting in only one transient partial remission of disease in the present cohort.

Discussion

The present study of a heterogeneous population of malignant melanoma patients and healthy donors demonstrates the feasibility of a simple technique for detecting circulating tumor-reactive T cells, not dependent on access to specific tumor antigens or autologous tumor material.

Most studies aiming at detecting the in vivo presence of tumor-reactive T cells in peripheral blood have employed single, well-characterized, tumor-associated peptides as activating antigen source. The results of these studies suggest that detection of peripheral tumor-reactive T cells might be used as evidence of malignancy and their numbers might be correlated with antitumor immunity [11, 14, 17, 24]. For example, using the ELISpot IFN-γ assay, Nagorsen et al. found that peptide-specific CD8⁺ T cells against Ep-CAM, HER-2/neu, or CEA could be detected in 32% of 22 patients with colorectal cancer, but not in 8 healthy subjects [17]. Schmittel et al. showed, using a similar technique, that peripheral CD8⁺ T cells against tyrosinase could be induced in melanoma patients by immunotherapy or chemoimmunotherapy and loss of these T cells was associated with clinical relapse of disease [24]. Circulating tyrosinase-reactive T cells were especially frequent in melanoma patients with clinical response to IL-2–based therapy [23]. Using tetramer staining, Lee et al. detected a circulating population of CD8⁺ T cells recognizing the HLA-A2–restricted epitopes of Melan-A/MART-1 and tyrosinase in 55% of 11 patients with metastatic malignant melanoma [14]. Similarly, Pittet et al. found high frequencies of Melan-A/MART-1–specific CD8⁺ T cells in 77% of 13 patients with melanoma, but also in 6 of 10 healthy HLA-A*0201⁺ individuals [21]. These cells were, however, of the naïve phenotype and do not therefore reflect immune activation by antigen. Recently, Speiser et al. found that a fraction of Melan-A/MART-1–specific T cells in the peripheral blood of patients with advanced melanoma were nonnaïve, revealing tumor-driven immune activation [25].

It should be stressed, however, that if the aim is to assess the polyclonal tumor-specific T-cell response in a patient, the evaluation of T-cell response to one or a few specific peptides is probably insufficient [31]. First, the biological relevance of peripheral T cells being activated upon coincubation with a specific tumor-antigenic peptide in an individual is unknown, in part because no exclusively tumor-restricted antigenic peptide has been found, and in part because some “tumor-associated” peptides activating T cells in vitro might not be presented or processed naturally [27]. Secondly, it is believed that the currently known tumor antigens only represent a minority of antigenic peptides produced by tumors [30], and even when defined antigens are known, there is no guarantee that they will be expressed by a given patient’s tumor or metastases [32].

Hence, for tumors that may not express known, defined tumor-associated antigens and for evaluation of a broader pool of in vivo tumor-reactive T-cell clones, alternative sources of tumor antigens must be employed. These include preparations of whole tumor cells [15], tumor cell apoptotic bodies [5], total tumor peptides [18], or tumor lysate [22]—and may come from autologous tumor [32], from tissue-type–matched allogeneic tumor [15], or from nonmatched allogeneic tumor [5, 18, 22].

In a previous, pioneering study, Letsch et al. used allogeneic HLA-A2- or HLA-A1-matched cultured melanoma cells from several cell lines as antigen source for rapid identification of circulating melanoma-reactive mononuclear cells. In an ELISpot IFN-γ assay, they found a significant immune response in the blood from 11 of 19 patients (58%) with metastatic melanoma, whereas healthy subjects had no or few activated cells [15]. Compared with these results, the findings of the present study seem encouraging as we—using allogeneic tumor cell lysate from a single melanoma cell line—showed a CD4⁺/IFN-γ T-cell response in 11 out of 15 patients (73%) with metastatic melanoma. It seems likely that careful selection of additional melanoma cell lines would further increase the frequency of responders.

In a few patients with high background levels of spontaneous T-cell IFN-γ production a profound negative effect was obtained, as the T-cell IFN-γ response detected in the lysate-stimulated samples was less than the spontaneous T-cell IFN-γ production. It may be argued that such samples would be less suited for assessment of a tumor lysate–specific response per se and should be excluded from analysis in future studies of the technique. That a sizeable spontaneous T-cell IFN-γ production very often leads to a negative result indicates that addition of the lysate may inhibit the cytokine production, and it is possible that such an effect also reduces the magnitude of any specific response to tumor antigen present in the lysate. The likely inhibitory effect of lysate may be reduced by optimizing the culture conditions including the composition of lysate.

Although the CD4⁺ T cells of most melanoma patients produced IFN-γ upon short-time coincubation with tumor lysate in vitro, the present study does not provide definite proof that these cells are in fact tumor-reactive in vivo. It is possible that the responses seen could be mediated by alloreactive T cells being activated by polymorphic gene products present in the melanoma cell lysate. Alternatively, the T cells could be reacting with non-tumor-associated peptides which are normally sequestered from the immune system or for other reasons do not give rise to T-cell activation in vivo. We do not find these explanations likely, however, as most stage I patients and all healthy donors showed no response at all. Moreover, the very short incubation time (6 h in total) and the treatment of cell lysate by high-speed centrifugation and sonification should diminish the risk of inducing an alloresponse.

In contrast to the frequent occurrence of response in metastatic melanoma we found only one marginal responder among four patients with stage I melanoma. Similarly, Nagorsen et al. reported no tumor-antigen-specific, circulating T cells in nine early stage colorectal cancer patients of whom eight had been operated on several months prior to the test, whereas more than half of the patients with local or distant metastases showed a response [17]. This apparent correlation between disease extent and presence of circulating tumor antigen–reactive T cells may reflect a scarcity or absence of these cells in localized cancers. Accordingly, it has been suggested that evasion of tumor cells in lymph nodes is a prerequisite for the induction of a tumor antigen–specific T-cell response in patients [17].

An alternative hypothesis linking disease stage with detectable tumor-reactive T cells in the circulation is that in vivo presence of antigen may be required for T cells to become activated in assays using short-term culture. It may follow from this that patients cured by surgery will show negative results after some time. Hence, previous studies suggest that quickly responding T cells are either effector T cells or a special category of memory T cells called “activated” or “cycling” memory T cells [26]. Both effector T cells and activated memory T cells may retain and require T-cell receptor contact with small quantities of specific antigen and disappear within days or weeks, respectively, if antigen is not present [26]. Mathematical modeling suggests that activated memory T cells comprise 95% of the memory pool with persistent, low-dose antigen exposure [4]. If this is valid then the current method may be useful as an “immunological tumor marker”. Indeed, data of the present study showing a correlation between response and disease stage and between response and the period patients might have been exposed to tumor antigens in vivo (roughly estimated by the interval between primary melanoma diagnosis and blood test) fits well with this theory. Together with the suggested more favorable prognosis of metastatic melanoma in responders this may additionally indicate a more indolent course of disease and/or an increased effect of treatment in responders. However, the limited number of patients and short follow-up of the current study do not allow any definite conclusions with respect to prognostic issues.

In the present study, we have shown that a tumor lysate produced from cultured allogeneic melanoma cells could induce a CD4⁺ T-cell response, easily detectable in peripheral blood samples in the majority of patients with metastatic melanoma, but absent in normal controls. The assay should be methodologically optimized and then thoroughly tested with respect to reliability and reproducibility. Additionally, it must be elucidated in clinical studies, whether the assay might be useful as a diagnostic tool, a marker of residual or recurrent disease, a prognostic factor, or a predictor or monitor of effect of antineoplastic therapy including immune-modulating therapy.