Cancer classification using the Immunoscore: a worldwide task force

Increasing literature [9, 11, 13, 14, 19] and meeting reports [20‐22] support the hypothesis that cancer development is influenced by the host immune system. A common theme has emerged, emphasizing the critical need to evaluate systemic and local immunological biomarkers. It is in agreement that this may offer powerful prognostic information and facilitate clinical decision-making regarding the need for systemic therapy [6, 23]. Numerous data collected from large cohorts of human cancers (with sample sizes n = 415, 599 and 602, [9, 13, 14], respectively) demonstrated that the number, type and location of tumor immune infiltrates in primary tumors, are prognostic for DFS and OS. Altogether these immune parameters are designated as the immune contexture [11, 12]. Notably, two large studies (with sample sizes n = 843 and 768, [24, 25], respectively) have shown that tumor immune infiltrate patterns and subsets in colorectal cancer are significant prognostic biomarkers, even after adjusting for stage, lymph node count, and well-established prognostic tumor molecular biomarkers including microsatellite instability (MSI), BRAF mutation, and LINE- hypomethylation.

A potential clinical translation of these observations is the establishment of an Immunoscore, based on the numeration of two lymphocyte populations (CD3/CD45RO, CD3/CD8 or CD8/CD45RO), both in the CT and in the IM of tumors, as a clinically useful prognostic marker [14]. For instance, colorectal cancer (CRC) patients with local tumor, no detectable lymph node or distant metastasis are usually treated by surgery alone. However, 20-25% of these patients will have recurrence of their disease indicating that occult metastases were already present at the time of curative surgery. No tumor-associated marker predicts recurrence in these patients. The Immunoscore (“I”) utilizes the numeration of CD8 and CD45RO cells in the CT and the IM of resected tumors to provide a score ranging from Immunoscore 0 (“I”0), when low densities of both cell types are found in both regions, to Immunoscore 4 (“I”4), when high densities are found in both regions. This Immunoscore approach was applied to 2 large independent cohorts (n = 602). Only 4.8% of patients with a high “I”4, relapsed after 5 years and 86.2% were alive. In comparison, 72% of patients with a low score (“I”0 and “I”1) experience tumor recurrence and only 27.5% were alive at five years. These “I”0 and “I”1 patients potentially could have benefited from adjuvant therapy, had the Immunoscore been incorporated into the tumor staging [14].

The Immunoscore classification, demonstrating the prevalence of immune infiltrates, potentially has a prognostic significance superior to that of the AJCC/UICC TNM-classification system. For all patients with CRC stages I/II/III, multivariate Cox analysis revealed that the immune criteria remained highly significantly associated with prognosis. In contrast, the histopathologic staging system (T stage, N stage, and tumor differentiation) was no longer significant [13]. Tumor invasion was shown to be statistically dependent on the nature of the host-immune reaction. Indeed, the immune pattern remained the only significant criteria over the classical AJCC/UICC TNM-classification for DFS and OS, and led to an editorial entitled “TNM staging in colorectal cancer: T is for T cell and M is for memory” accompanying the publication by Mlecnik and Broussard et al. in the Journal of Clinical Oncology [13, 26]. It has thus been suggested that the prevalence of post-surgical immune infiltrates, and not tumor status, is the key indicator for reoccurrence, metastasis and therefore clinical outcome.

These results suggest that once human cancer becomes clinically detectable, the adaptive immune response may play a critical role in preventing tumor recurrence. The ability of effector-memory T cells to recall previously encountered antigens leads to a protective response. Following primary exposure to antigen, memory T cells disseminate and are maintained for long periods of time [27]. The trafficking properties and the long-lasting antitumor capacity of memory T cells could result in long-term immunity in human cancer.

Although first described in CRC, the impact of the immune cytotoxic and memory T cell phenotype has been demonstrated in many other human tumors and appears to be a general phenomenon [23, 28]. It is interesting to note that the implications of this immune phenotype apply not only to various organs of cancer origin (such as breast, colon, lung, head and neck, kidney, bladder, ovary, prostate), but also to various cancer cell types (adenocarcinoma, squamous cell carcinoma, large cell cancer, melanoma, etc).

A recent Nature Cancer Review meta-analysis [12] summarizes the impact of immune cells including B cells, NK cells, myeloid derived suppressor clls MDSC, macrophages, and all subsets of T cells on clinical outcome from more than 120 published articles. Beyond colorectal cancer, a strong T cell infiltration associated with good clinical outcome has been reported in many different tumours, including melanoma, head and neck, breast, bladder, urothelial, ovarian, esophageal, renal, prostatic, pancreatic, cervical, medulloblastoma, merkel cell carcinoma, hepatocellular, gastric, and lung cancers [12]. Thus, high densities of T cells (CD3+), of cytotoxic T cells (CD8+), and of memory T cells (CD45RO+) were clearly associated with a longer DFS (after surgical resection of the primary tumour) and/or OS.

The prognostic impact of other immune cells such as B cells, NK cells, MDSC, macrophages, and subset of T-helper populations, (T_H2, T_H17, T_REG cells) may differ depending on the type of cancer, and on the cancer stage [12]. In contrast, T cells, cytotoxic T cells, T_H1 cells, and memory T cells were strongly associated with good clinical outcome for all cancer types [12]. Thus, general characteristics emerge in which cytotoxic T cells, memory T cells, and T_H1 cells are associated with prolonged survival.

Considering the important role of the host immune signature in controlling tumor progression, it is now imperative to initiate the incorporation of the Immunoscore as a component of cancer classification [13, 14] and a prognostic tool [23]. This strategy has a dual advantage: firstly, it appears to be the strongest prognostic factor for DFS and OS, particularly in early stage cancers and secondly, it could allude to potential targets for novel therapeutic approaches, including immunotherapy. Current immunohistochemical technologies allow the application of such analyses by laboratories concerned with routine diagnostic and prognostic assessment of tumors.

The inherent complexity of immunohistochemistry, in conjunction with protocol variability, analysis of different immune cell types, inconsistent tissue region selection criteria, combined with differences in conjunction with qualitative and semi-quantitative criteria to measure immune infiltration, all contribute to the variability of the results obtained, and raise the concern that specialized protocols and training may be required. It is therefore essential to pursue assay uniformity to reduce these limitations. Many markers, signatures, and methods have been described to evaluate the prognosis of cancer patients. Yet, very few such markers and laboratory assays are used in clinical practice. Thus, we believe that harmonization of an assay evaluating the “inflammation”, i.e. the Immunoscore of the tumor is essential. Analytical and clinical validation of the assay is required before the Immunoscore will reach clinical applicability for individual patients. However, current immunohistochemical technologies allow the application and cross-validation of such analysis in laboratories performing routine diagnostic and prognostic assessment of tumors. In order to be able to compare results in the future, and for the development of more effective prognostic and predictive markers to improve clinical decision-making, it is important to perform a standardized set of experiments. Assay harmonization should minimize data variability and allow worldwide correlations of Immunoscore results with clinical outcomes. Harmonization guidelines resulting from this process are expected to be simple to implement and will improve assay performance. Effective large-scale assay harmonization efforts have already been conducted for commonly used immunological assays of peripheral blood immune cell populations [29, 30].

A fundamental parameter to determine the Immunoscore will include the immune cell density, calculated by numerical quantification of two lymphocyte populations, cytotoxic and memory T cells at the CT and the IM of tumors. This core criterion will establish prognosis of patient clinical outcome, regardless of the absence of other cancer associated prognostic markers, such as in early tumor stage (I/II) patients [14]. In human cancers, a high density of T_H1/cytotoxic memory T lymphocytes, located both in the CT and IM of the primary tumor, is associated with long DFS and OS, in addition to low risk of relapse and metastasis. This was particularly illustrated in CRC [5, 9, 13, 14, 19], and should be applicable to most human tumors [23]. Thus, this Immunoscore classification may help identify the high-risk patients who would benefit the most from adjuvant therapy.

Whether the immune contexture of the primary tumor predicts therapeutic responses is of paramount importance for patient clinical management. Data based on immune signatures have established that a strong immune component is predictive of good response to chemotherapy in breast cancer [31‐33], a tumor in which a high lymphocyte infiltrate is associated with higher response rate in neo-adjuvant therapy [34, 35]. In hepatic metastases of CRC, high CD8 infiltrates in the IM predicts better response to chemotherapy and prolonged survival [36]. In melanoma, an immune signature displaying high expression of T_H1 and cytotoxicity-associated genes, correlates with favorable clinical outcome to several different therapeutic vaccines [8]. In addition, high numbers of CD8 T cell infiltrates within metastatic melanoma correlated with prolonged survival [37]. However, the high T_H1 and cytotoxic immune response associated with prolonged survival in patients receiving adjuvant therapies may not be a prediction of response to the therapy, but rather the fact that the host-immune response within the tumor protects the patient and prolongs patient life. To assess the impact of the Immunoscore as a predictive marker, it should be evaluated prospectively in randomized clinical trials.

Over the past few years, the area of immune regulation at the level of the tumor microenvironment has gained a forefront position in cancer research, in CRC [9, 12‐14], in melanoma [38] and all other cancer types [6]. The Immunoscore was initially described several years ago [9], and more recently advances have been made in the development of the Immunoscore as a prognostic factor [13, 14] that could be used in routine testing [39]. In an effort to promote the utilization of such Immunoscore in routine clinical settings worldwide, the Society for Immunotherapy of Cancer (SITC), the European Academy of Tumor Immunology (EATI), and “La Fondazione Melanoma Onlus”, initiated a task force on “Immunoscoring as a New Possible Approach for the Classification of Cancer” that took place in Naples, Italy, February 13^th, 2012 [39]. This perspective represents a follow-up on this initiative, originally announced in a J Transl Med. editorial in January 2012 [39]. The working group, composed of international expert pathologists and immunologists, identified a strategy for the organization of worldwide participation by various groups for the validation of the Immunoscore. The objectives of the meeting included discussing: the role of immune system in cancer; a review of the AJCC/UICC-TNM classification of CRC; the role of the microenvironment in melanoma biology; the review of the AJCC classification of melanoma; the relevance of HLA-A2 in cancer prognosis and tumor malignancy; data utilizing the Immunoscore and a proposal for standardizing the operating procedures for the Immunoscore quantification. Furthermore, the international working group evaluated the feasibility of using the Immunoscore for the classification of cancer. Evidence-based selection of specific markers and their combinations for the Immunoscore was discussed including biological rationale, clinical use, synthetic meta-analysis of the Immunoscore, analytical performance, reagents availability and testing, metrics for decision making, cross-laboratory validation of methodology and identification of potential problems during development of other markers. Practical aspects of the validation of the assay by participating centers were proposed including consideration of cancer types, cancer stages, and the definition of a working group of pathologists for the validation phase.

CRC has been most comprehensively studied and the prognostic significance of immunologic parameters has been best validated, thus special emphasis will be placed in this disease for this formal validation. As neo-adjuvant treatments are nowadays recommended for rectal cancer, it may be advisable to separate the validation of colon cancers and rectal cancers. Other cancer types, including melanoma and breast cancers were additionally discussed and their validation will follow. An independent international consensus panel of expert laboratories discussed cross-laboratory assay validation for the development of an Immunoscore prognostic method. As evaluation of cytotoxic memory CD8⁺ T cells (CD3⁺, CD8⁺, CD45RO⁺, Granzyme B⁺ (GZMB)) provides the best method to discriminate patient outcome, any combination of two of these aforementioned markers should have similar statistical power. Because of technical difficulties including background noise (CD45RO) and granular staining (GZMB), it was decided to employ the two easiest membrane stains, CD3 and CD8. Thus, the combination of two markers (CD3⁺ and CD8⁺) in two regions (CT and IM) was agreed for validation in standard clinical practice. Precise quantification will be performed on whole slide sections (Figure 1). For harmonization of the assay and reproducibility of the method, all laboratories agreed to test the prognostic value of specific immune cell infiltration following the recommended initial guidelines. The inherent complexity of quantitative immunohistochemistry underscored the urgent need to reach assay harmonization. The components of the Immunoscore are listed in Table 1. Additional markers could be added subsequently to refine the methodology even further if required. After worldwide validation, a consensus detailed protocol will be available.

To be used globally in a routine manner, evaluation of a novel marker should have the following characteristics: pathology-based, feasible in routine settings, simple, inexpensive, rapid, robust, reproducible, quantitative, standardized, and powerful. The Immunoscore fulfills all these keys aspects summarized in Table 2.

The purpose of the Immunoscore worldwide task force is to validate these points.

The goals of the first ongoing initiative are the following:

Thus, the benefit of the Immunoscore worldwide study would be to validate the feasibility, reproducibility, and prognostic value of the routine Immunoscore on colon cancer patients.

The goals of the next initiatives will be the following:

In the inaugural World Immunotherapy Council meeting (February 21^st - 24^th 2012, Curacao), the Immunoscore task force, led by the Society for Immunotherapy of Cancer (SITC), received the support from several additional cancer immunology societies including; Biotherapy Development Association (BDA); Canadian Cancer Immunotherapy Consortium (CCIC); Cancer Immunotherapy Consortium (CIC) of the Cancer Research Institute (CRI); Association for Cancer Immunotherapy (CIMT); Committee for Tumor Immunology and Bio-therapy (TIBT); European Academy of Tumor Immunology (EATI); European Society for Cancer Immunology and Immunotherapy (ESCII); Italian Network for Tumor Biotherapy (NIBIT); Japanese Association of Cancer Immunology (JACI); Nordic Center for Development of Antitumor Vaccines (NCV-network); Progress in Vaccination Against Cancer (PIVAC); Adoptive engineered T cell Targeting to Activate Cancer Killing (ATTACK) and the Tumor Vaccine and Cell Therapy Working Group (TVACT). These groups share a clinical or basic interest in the immunobiology of the tumor microenvironment and will collaborate with worldwide expert pathologists to assess the validity of this new approach. Following the Immunoscore Workshop and the World Immunotherapy Council meeting, 22 international expert centers agreed to participate in this visionary enterprise. These participants represent 22 Centers Worldwide from 16 countries including Asia, India, Europe, North America, Australia, and Middle East (Figure 2). Additionally, pathologist associations and other medical specialty groups have been invited to participate.

A preliminary summary of this effort will be presented during the “Workshop on Tumor Microenvironment” prior to the SITC annual meeting (October 24^th - 25^th 2012, Maryland, USA). Finally a “Workshop on Immunoscore” (December 5^th 2012, Naples, Italy), will lead to the preparation of a summary document providing recommendations for the harmonization and implementation of the Immunoscore as a new component for the classification of cancer TNM-I (Immune).