Putative contribution of CD56 positive cells in cetuximab treatment efficacy in first-line metastatic colorectal cancer patients

Background Activity of cetuximab, a chimeric monoclonal antibody targeting the epidermal growth factor receptor, is largely attributed to its direct antiproliferative and proapoptotic effects. Antibody-dependent cell-mediated cytotoxicity (ADCC) could be another possible mechanism of cetuximab antitumor effects and its specific contribution on the clinical activity of cetuximab is unknown. Methods We assessed immune cells infiltrate (CD56, CD68, CD3, CD4, CD8, Foxp3) in the primary tumor of metastatic colorectal cancer (mCRC) patients treated with a first-line cetuximab-based chemotherapy in the framework of prospective trials (treatment group) and in a matched group of mCRC patients who received the same chemotherapy regimen without cetuximab (control group). The relationship between intra-tumoral immune effector cells, the K-ras status and the efficacy of the treatment were investigated. We also evaluated in vitro, the ADCC activity in healthy donors and chemonaive mCRC patients and the specific contribution of CD56+ cells. Results ADCC activity against DLD1 CRC cell line is maintained in cancer patients and significantly declined after CD56+ cells depletion. In multivariate analysis, K-ras wild-type (HR: 4.7 (95% CI 1.8-12.3), p = 0.001) and tumor infiltrating CD56+ cells (HR: 2.6, (95%CI:1.14-6.0), p = 0.019) were independent favourable prognostic factors for PFS and response only in the cetuximab treatment group. By contrast CD56+ cells failed to predict PFS and response in the control group. Conclusions CD56+ cells, mainly NK cells, may be the major effector of ADCC related-cetuximab activity. Assessment of CD56+ cells infiltrate in primary colorectal adenocarcinoma may provide additional information to K-ras status in predicting response and PFS in mCRC patients treated with first-line cetuximab-based chemotherapy.


Background
Cetuximab is a chimeric immunoglobulin G 1 (IgG1) monoclonal antibody (mAb) which binds the epidermal growth factor receptor (EGFR) with high affinity and inhibits ligand binding [1]. Cetuximab is active in chemotherapy resistant metastatic colorectal cancer (mCRC) [2,3] and enhances response rate and progression-free survival (PFS) in first-line therapy in combination with Folfiri and Folfox [4,5]. Clinical studies of cetuximab therapy in mCRC have failed to show a significant correlation between EGFR-staining intensity and patients' response to cetuximab treatment [2,3]. Therefore, identifying molecular markers that can select patients who are likely to benefit from cetuximab is crucial to avoid chemotherapy toxicity and reduce treatment cost. Recently the absence of K-ras mutation appears to be a reliable marker in predicting cetuximab efficacy, both in first-line and in third-line of the anti-EGFR therapies [4][5][6][7][8]. Other factors such as, EGFR amplification [9][10][11], epiregulin and amphiregulin expression [12], nuclear factor-kB tumor expression [13], PTEN [14], BRAF [15] or PIK3CA [16] were also suggested to predict response to cetuximab but these additional biomarkers require further validation before incorporation into clinical practice.
The activity of cetuximab has largely been attributed to the direct antiproliferative and proapoptotic effects of the antibody. However, another possible mechanism of its antitumor effects is mediated through antibody-dependent cell-mediated cytotoxicity (ADCC). The ADCC mediated through Fc receptors (FcγR) carried by NK cells, macrophages and polymorphonuclear leukocytes, is a well-recognized immune effector mechanism in the antitumor effect of IgG1 [17]. Of these cells, NK cells represent the principal ADCC effector cells [18,19]. Recently, some polymorphisms on genes encoding for activating receptors FcγRIIa and FcγRIIIa were found to affect the clinical efficacy of cetuximab [20,21]. The recruitment of Foxp3-positive regulatory T cells (Treg) into tumor likely represents one of the mechanisms by which malignant cells evade host immune response. The intratumoral density of foxp3 as been reported to be associated with overall survival [22]. Once activated, Tregs can inhibit the function of dendritic cells, NK cells, B cells and other immune cells [23][24][25] and consequently alter ADCC activity.
Based on the potential value of ADCC in cetuximab activity, we assessed in mCRC patients, the role of peripheral blood mononuclear cells (PBMC) and their CD56+ subpopulation in ADCC activity and we evaluated the relationship between the intratumoral immune cells and the efficacy of first-line cetuximab-based chemotherapy.

Patients and tissue samples
a retrospective study review was conducted from data in our institution's prospectively collected gastrointestinal cancer database. Chemonaive patients with mCRC who underwent surgical resection of their primary tumor and diagnosed with synchronous metastases were included in the analysis. We analyzed the primary tumor of 33 chemonaïve mCRC patients treated with first-line cetuximab (Erbitux ® , Merck, Darmstadt, Germany) containing chemotherapy regimen in the framework of phase II and III studies. Patients received mostly standard FOLFIRI regimen or FOLFOX accordingly to study recommendations [4,26,27]. Thirty-five chemonaïve mCRC patients with synchronous metastasis, who underwent resection of the primary tumor before starting a similar chemotherapy regimen that did not contain cetuximab were used as control group. The treatment and the control group were case-matched for the following parameters: sex, age, primary tumor location, tumor stage, performance status, metastatic sites, type of chemotherapy administered and treatment duration.
Formalin-fixed and paraffin-embedded (FFPE) samples of the primary tumor were obtained for immunohistochemical and PCR analysis.
The study was approved by the ethic committee of the Erasme University Hospital and all patients provided written informed consent.

Clinical evaluation and tumor response criteria
We considered PFS to assess the cetuximab-based chemotherapy efficacy in first line and not OS which is influenced by second and third-lines chemotherapy and liver mets surgery. Tumor response was evaluated by computerized tomodensitometry according to the Response Evaluation Criteria in Solid Tumors (RECIST) [28] and classified in complete response (CR), partial response (PR), stable disease (SD) and progressive disease (PD). The overall best response (OR) was defined as the best response recorded from the start of the treatment until disease progression, recurrence or start of the new therapy. Response was centrally confirmed in the setting of the referenced trials. For the analysis, CR and PR patients were grouped in responders; patients with SD and PD were grouped in non-responder patients.

Immunohistochemical assessment
Hematoxylin & eosine sections of the tumors were examined by a pathologist for confirmation of the histologic diagnosis and the optimal block donors selected. The FFPE tissues were deparaffinized in xylene, and rehydrated in graded alcohols and water. Colorectal tumor sections were incubated with monoclonal antibodies against CD3, CD4, CD8, CD56, CD68, Foxp3 and EGFR and DAB-chromogen were applied (Dako, Copenhagen, Denmark) (Additional file 1: supplemental Table S1). Appropriate negative and positive controls were used.

Evaluation of Immunohistochemical variables
The slides were examined independently by two observers (RM, NN) blinded to both clinical and pathologic data. Twenty representative fields of the tumour invasive margin (IM) were chosen from each slide, and the stained cells were counted by means of a 10×10 ocular grid at × 200 magnification (observed area 0.25 mm 2 ) using confocal microscopy. For each case, the total number of CD3, CD8, CD4, FOXP3, CD56, CD68 positive cells (representing lymphocytes T (LT), LTCD8, LTCD4, Treg, natural killers (NK) and macrophages, respectively) per square millimeter was calculated. Variations in the enumeration within a range of 5% were re-evaluated and a consensus decision was made.

DNA extraction and mutation analysis
Presence of tumor cells (> 75%) in each tumor block was firstly histologically controlled by H & E coloration. Thereafter, DNA was extracted from FFPE samples after macrodissection. The presence of K-ras was determined by allelic discrimination assay on a 7500HT Real Time PCR System. K-ras mutations located within the codon 12 (n = 6) and 13 (n = 1) were screened for. All mutations were confirmed by direct sequencing (8).

Cell Lines and cell culture
The EGFR overexpressing DLD1 colorectal glandular carcinoma cell line was obtained from the American Type Culture Collection (Manassas, VA, USA). DLD1 has been previously found to carry mutated K-ras. To confirm that DLD1 indeed harbour mutated K-ras alleles, we extracted the corresponding genomic DNA and sequence the K-ras locus. We confirmed that DLD1 display Gly13Asp K-ras mutation (data not shown).The DLD1 cells were maintained in RPMI 1640 (Sigma, St Louis, MO, USA) supplemented with 10% heat-inactivated foetal bovine serum (FBS, Gibico BRL, Grant Island, NY, USA), sodium pyruvate 1%, 100 units/ml penicillin and 100 μ g/ml streptomycin. The expression of EGFR by the DLD1 cell line was confirmed by immunocytochemistry (data not shown)).

Purification of effector cells and interleukine-2(IL-2) treatment
Highly depleted PBMC from CD56 + cells (CD56 depleted PBMC), were obtained by magnetic activated cells sorting (MACS) using the system from Miltenyi Biotec according to the manufacturer's instructions. The CD56 depleted PBMC and non depleted PBMC were used sepa-rately in ADCC assays. Between 92% to 94% of the CD56 + cells were sorted after MACS depletion. As IL-2 is known to activate PBMC, we tested the effect of human recombinant IL-2 (R & D Systems) on cetuximab-mediated ADCC in CD56 depleted and non depleted PBMC. The two populations were cultured in medium alone or enriched with rhuIL-2 (10 ng/ml) for 18 h prior to use in ADCC assays. Accordingly to Heo et al., cytotoxicity was evaluated using a 3-(4,5 dimethyltiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) colorimetric assay [29,30] (additional file 2). Experiments conducted in a preliminary phase to select optimal conditions for the ADCC effect showed that (a) target cells were not killed after exposure to cetuximab in the absence of PBMC (b) 10 μ g/ml was the optimal cetuximab concentration for saturating ADCC assay by PBMC.

Statistical Analysis
The PFS was estimated by the Kaplan-Meier method and the two groups were compared with by the log-rank test. For all immunohistochemical markers, the cut off for definition of subgroups was the median value. The nonparametric χ 2 test and the Fisher's exact tests were carried out as appropriate to compare categorical variables. The Mann-Witney U test was used for the comparison of the treatment and the control groups regarding the tumor infiltrating immune cells and ADCC activity. Correlations were analyzed by Spearman's correlation coefficient test. Multivariate analyses used a step-down procedure based on the likelihood ratio test. A p-value ≤ 0.1 in univariate analysis was required to consider the variable for multivariate analysis. The Two-tailed p < 0.05 was judged to be significant. All analysis were performed with SPSS 10.0 software (SPSS, Chicago, Il).

Patients characteristics
The baseline and treatment characteristics of patients from treatment and control group are summarized in table 1. The two groups were similar regarding all variables matched including the proportion of patients treated with the FOLFIRI

Immune cells infiltrate in primary CRC and treatment efficacy
Positive CD4, CD3, CD8, CD68 and Foxp3 cells were detected in 33/33 (100%) tumor samples (figure 1) and their mean number were similar between the two patient groups (table 2). The number of intratumoral macrophages (CD68 + cells) and Lymphocytes (CD3 + ,CD4 + ,CD8 + , Foxp3 + cells) were similar between responders and non responders and we found no association with the PFS both in the treatment and the control group (table 3).
For CD56+ cells, two distinct immunologic pattern were clearly observed consisting in, either tumors with strongly positive CD56 staining (CD56 positive tumors) or tumors with undetectable CD56 staining (CD56 nega- tive tumors) (figure 1).These two pattern were found both in the treatment and the control groups (table 2) and were compared for OR and PFS and correlated with the K-ras status. In the control group, no difference has been detected in OR and PFS between patients with CD56 negative tumor cells and those with CD56 positive tumor (table 3, figure 1). By contrast, in the treatment group, CD56 positive tumors were more frequent in responder than in non-responder patients (

EGFR expression
as previously observed, no significant correlation was found between EGFR expression, OR rate and PFS.

Multivariate analysis
CD56+ cells infiltrate and WT K-ras status were independent predictors of OR. Both CD56 negative tumors (Haz- ard ratio: 2.6 (95% CI: 1.14-6.00); p = 0.019) and K-ras mutations (Hazard ratio: 4.74 (95% CI: 1.8-12.3); p = 0.001) contribute as significant independent negative prognostic factors for the PFS (table 3). Interestingly, in the group of WT K-ras tumors (n = 20), there is a trend for a higher OR and a prolonged PFS in patients with CD56 positive tumor as compared with CD56 negative tumor patients and the whole group (CD56 + and CD56 -) (additional file 3, Supplemental Table S2).

Cetuximab-mediated ADCC in vitro activity is maintained in mCRC patients
CD56 + cells represented a minor fraction (range between 5 and 14%) of the whole PBMC population. Adjunction of cetuximab enhanced the cytotoxicity of PBMC (ADCC) as compared to the PBMC activity (PBMC alone). At the higher E:T ratios of 10:1 and 20:1, cetuximab enhanced tumoral cell lysis, as compared to the baseline activity of PBMC; both in healthy volunteers (p = 0.008 at an E:T ratio of 10:1,from 26% to 60%; and p = 0.009 at an E:T ratio of 20:1, from 30% to 74%) and in mCRC patients (p = 0.001 at an E:T ratio of 10:1,from 19% to 56%; and p = 0.002 at an E:T ratio of 20:1 from 25% to 71%) (figure 2 and 3B). PBMC activity was not increased by the control antibody. Interestingly, PBMC and ADCC activity was similar between mCRC patients and healthy volunteers.

CD56 + cells are responsible for cetuximab-mediated ADCC in vitro activity and its majoration by IL-2
While CD56 + cells represent a minor fraction of the PBMC, depletion of CD56 + cells significantly reduced the ADCC activity at the different E:T ratio (p < 0.05 at all E:T ratios) even after IL-2 stimulation (figure 3C, 3D) underlying the major role of this immune cells in ADCC activity.

Discussion
One of the most important mechanisms of cetuximab and panitumumab, the monoclonal antibodies that target the EGFR, is through the inhibition of the EGF receptor/ ligand interaction. The other pathway through which cetuximab only, as an IgG1, may exert its antitumor effect is ADCC [12,14]. Several in vitro and in vivo studies have shown that this chimeric IgG1 mAb binds to the antigen, the EGFR, on the surface of tumor cells while its Fc portion binds to the immune effector cells through FCγ R. Consequently, this binding activates the immune cells leading to tumor cells killing. ADCC activity of mAbs has been well described for trastuzumab, a human IgG1 anti human EGFR 2 (Her-2) antibody and for rituximab, a chimeric IgG1 mAb for B-cell differentiation antigen CD20 [31][32][33].
The array of cellular effectors potentially exerting ADCC includes the CD3 -CD56 + NK lymphocytes [34] but also CD3 + CD16 + T-cell subset [35], CD16 + CD33 + macrophages [36] and CD16 + granulocytes [37]. Among the mononuclear cells (MNC), the NK cells (CD3 -CD56 + ) play a predominant role in the ADCC through low affinity type II (Fcγ RIIc) and type IIIA (Fcγ RIIIa) Fc receptors present on their surface [38,39]. The polymorphonuclear cells are also able to induce ADCC by the engagement of their CD32 receptor but needs higher concentration levels of antibody compare with the MNC.
Since only less than 50% of WT-Kras mCRC patients seem to benefit from cetuximab-based treatment, it may be highly relevant to identify other biomarkers than K-ras status to predict cetuximab efficacy. To this end, the eval- Figure 2 PFS according to tumor CD56 status in the treatment group (A) and the control group (B).

A. B.
Time (month) 30  uation of the ADCC actors as putative surrogate markers of cetuximab-based therapy seems to be attractive. As there is still debate whether ADCC plays a role in metastatic cancer patients who mostly have suppressed immune function, we choose to evaluate cetuximabmediated ADCC activity in front-line chemonaïve mCRC patients. We only used an EGFR + cell line and no control EGFRcell since the activation of ADCC activity by cetuximab requires the expression of its target, the EGFR [40]. Our functional assay provided evidences that CD56 + cells are the main effectors of cetuximab-mediated

Healthy volunteers Colorectal cancer patients
ADCC in mCRC patients. Furthermore, the PBMC of mCRC were capable to initiate ADCC comparable to healthy donors suggesting that their function is maintained. Furthermore, we observed a significant in vitro ADCC activity despite the mutated K-ras status of the DLD1 cell line. Our results are thus in accordance with a recent report showing that cetuximab-related ADCC is irrespective of the K-ras status [40]. This property could be exploited particularly in patients with K-ras-mutated tumors, who otherwise have a low probability of responding to cetuximab [4,6,7]. In this way, approaches to enhance ADCC activity, such as immunostimulation and new generation of EGFR-directed mAbs, may be promising in the management of this subset of patients.
Whether ADCC contributes to the clinical efficacy of cetuximab, remains to be determined. Indeed, cetuximab and panitumumab displayed similar clinical activity despite the fact that panitumumab is a fully human IgG2 monoclonal antibody with low ability to recruit immune cells compared to IgG1. The adding value of cetuximabmediated ADCC might deserve direct comparison between cetuximab and panitumumab. On the other hand, two studies have demonstrated that FCγ R gene polymorphisms (Fc R3A-V158F and Fc R2a-H131R) are associated with clinical outcome (response rate and PFS) in mCRC patients treated with cetuximab administered in combination with irinotecan or as a single-agent [20,21]. These findings may support the contribution of ADCC in cetuximab efficacy. In this line, it would be helpful to identify which surrogate markers may help us to predict this contribution to efficacy.
Complementary to our in vitro assay, we observed that the OR and PFS were significantly higher in patients with CD56 positive tumors as compared with CD56 negative tumors. This was not observed in a matched control group. Nevertheless, only evaluation in prospective randomized trials may confirm that CD56 tumor status is specifically predictive for ADCC contribution and not only prognostic.
When adjusted for K-ras status in multivariate model, the CD56 tumor status remains an independent marker of response to treatment (OR) and of better PFS and could be therefore of additional interest in predicting enhanced cetuximab activity in the WT population.
Our data raise additional questions regarding the role of NK cells. The phenotype and properties of tumor-infiltrating NK cells seems to be different from that of peripheral NK cells. Peripheral NK cells are CD3 -CD56 low CD16 + while intratumoral NK cells are usually CD3 -CD56 bright CD16and unable to do ADCC [41,42]. The number of intratumoral CD56 + cells was found to be low (mean number: 6.4/mm 2 and 5.6/mm 2 in the treatment and the control group respectively) and it may be thus questionable how this small number can truly impact on CTX efficacy. However, while the circulating CD56 + cells represent a minor fraction of PBMC, we observed in vitro that their low number does not preclude for ADCC intensity. Additionally, some studies have evidenced significant immune cells recruitment into the tumors after trastuzumab or rituximab based therapy [38,43,44] even when trastuzumab was associated with immunosuppressive chemotherapy [38]. This recruitment was more significant in tumors exhibiting higher in situ infiltration lymphocytes infiltrate at the baseline [44]. All together, these data suggest that intratumoral CD56 + cells might be a surrogate marker of subsequent recruitment.

Conclusions
In summary, the presence of tumor-infiltrating CD56 + cells is an independent predictor for PFS and OR in mCRC patients treated with first line cetuximab basedchemotherapy. This suggests that ADCC, mainly through its central effector, the NK cells, could influence response to cetuximab-based chemotherapy. The exact relationship between TIL of the primary cancer and peripheral NK cells activity should be further explored, notably in assessing the role of ADCC in the adjuvant setting in patients treated with cetuximab.