INTRODUCTION

Clinical reports with the use of monoclonal antibodies directed against the ligand-binding site of the epidermal growth factor receptor (EGFR) have shown promising results in the treatment of colorectal cancer; it is hoped that they will improve further the available treatment options for patients diagnosed with this deadly disease.^1-3

However, in these trials a considerable proportion of colorectal tumors highly expressing EGFR has been found to be refractory to treatment, whereas patients with tumors showing low or even absent expression of EGFR were surprisingly responsive to therapy.⁴ These considerations have dampened enthusiasm about the possibility for a reliable selection of the appropriate antineoplastic therapy based on the molecular characteristics of the tumor, which initially seemed possible with the introduction of anti-EGFR treatment modalities. Nevertheless the expanding role of anti-EGFR therapeutic modalities, along with the growing number of patients potentially requiring such a treatment approach, made the need for a correct and reliable identification of responding tumors even more relevant. Increasing preclinical and clinical evidence suggested that the nuclear factor kB (NF-kB), in addition to immune response and inflammation, may also play a crucial role in tumor cell functions such as proliferation, migration, and apoptosis.^5-8 Under physiologic conditions NF-kB dimers are predominantly inactive because of the interaction with cytoplasmic NF-kB inhibitors. Nevertheless, it has been shown that different signaling pathways may trigger NF-kB activation, which is aberrantly or constitutively present in many human malignancies.^5-8 Interestingly, an altered expression of NF-kB in solid tumors has been correlated with prognosis and resistance to antitumoral treatments such as radiotherapy and chemotherapy, presumably via the inhibition of the apoptotic response induced by these treatments.^9-14 This suggests that the inhibition of the NF-kB pathway could present an impressive therapeutic potential in the treatment of cancer. A biologic link between NF-kB and EGFR has been also suggested by the demonstration that EGFR can induce NF-kB and that high level of EGFR expression is essential for EGFR-mediated NF-kB activation.^15-17 On the basis of these findings, a preclinical study investigated the possibility of restoring apoptosis in cancer cells by targeting the NF-kB signaling pathway with an anti-EGFR antibody.¹⁸ Using this strategy, the authors were able to re-establish tumor sensitivity to chemotherapy and radiotherapy, thus suggesting a strong rationale for EGFR inhibition in NF-kB–driven chemotherapy resistance.

The aim of our analysis was to verify the role of NF-kB expression in terms of response rate and time to progression (TTP) in advanced colorectal cancer patients treated with irinotecan and cetuximab after failure of an irinotecan-based chemotherapy. This was done to identify a possible new molecular determinant of response in this group of patients.

PATIENTS AND METHODS

RESULTS

Seventy-six patients were eligible for our analysis, 45 males (59%) and 31 females (41%); median age at diagnosis was 65 years (range, 38 to 78 years). Nineteen patients (25%) received cetuximab-irinotecan therapy as a second-line treatment; the remaining 57 patients (75%) were treated with cetuximab-irinotecan after failure of at least two previous lines of chemotherapy. Overall, we observed a partial response in 16 patients (24%) and progressive disease in 32 patients (48%). Nineteen patients (28%) showed stable disease, whereas no complete remissions were obtained. Cetuximab-related skin rash was present in 54 patients (80%). TTP for the entire patient population was 3.6 months and overall survival was 10.3 months (Table 1).

Immunohistochemistry analysis for NF-kB and EGFR expression was performed in primary tumors in all patients. NF-kB nuclear expression was positive in tumor samples from 46 patients (60%) and negative in the remaining 30 patients (40%). All major clinical characteristics were comparable among NF-kB–positive and NF-kB–negative groups of patients. In particular, no differences were noticed for sex, age at diagnosis, previous lines of chemotherapy, and skin rash (Table 1). In contrast, in NF-kB–expressing and NF-kB–negative tumors, we observed a partial remission in four (10%) and 12 (48%) patients, respectively (P = .0007). In addition, survival parameters seemed influenced by NF-kB status. In fact, median time to progression was 3 months in patients showing NF-kB–expressing tumors and 6.4 months for those who had tumors lacking NF-kB (P = .021; Fig 3). Moreover, median overall survival was 9.5 v 15.8 months for NF-kB–positive and NF-kB–negative patients, respectively (P = .036; Fig 4).

According to previously published findings, the response rates for patients showing the typical cetuximab-related skin rash were higher than those in patients without skin rash. These observations were particularly evident and became statistically significant when we accounted for the severity of the skin rash. Skin rash also positively affected both median TTP and median overall survival (Table 2). These differences were statistically significant as well (Table 2). Other tested variables such as sex, age at diagnosis, and previous lines of chemotherapy did not reach the statistical significance level at univariate analysis. Multivariate analysis including those factors that were significant at univariate analysis for TTP and overall survival (skin rash and NF-kB status), confirmed that NF-kB was able to influence median TTP (HR = 0.59; 95% CI, 0.39 to 0.89; P = .012) and median overall survival (HR = 0.47; 95% CI, 0.29 to 0.76; P = .002) independently. Analogously, the severity of cetuximab-related skin rash independently correlated with median TTP (HR = 0.38; 95% CI, 0.29 to 0.51; P = .0001) and median overall survival (HR = 0.31; 95% CI, 0.22 to 0.43; P = .0001).

DISCUSSION

In addition to the suggestion of an improved therapeutic level, the introduction of biologically targeted agents in the treatment of colorectal cancer patients held the promise for a more accurate selection of patient candidates for such a treatment approach. Nevertheless, although the use of anti-EGFR monoclonal antibodies (ie, cetuximab) in the clinical practice seemed to confirm the activity profile previously suggested, since their first appearance, it was clear that a reliable predictive factor for outcome was lacking.³ The level of EGFR immunohistochemical expression, initially considered mandatory for such agents, in fact was shown to be totally inadequate for the role of predictive factor.⁴ Researchers worldwide tried then to investigate molecular and clinical mechanisms involved in the sensitivity (or resistance) of neoplastic cells to anti-EGFR monoclonal antibodies, moving, in a manner of speaking, from bedside to benchside. We previously suggested that EGFR status in metastatic sites may be different from what can be observed in the corresponding primary tumor, and therefore EGFR detection in primary colorectal cancer could be insufficient for planning therapy with EGFR-targeted monoclonal antibodies.¹⁹ However, it is also probable that immunohistochemistry itself may represent an inadequate method for a reliable determination of EGFR expression as a predictive factor and other methods should be then explored. EGFR copy number as determined by fluorescent in situ hybridization analysis seemed to correlate with response to anti-EGFR monoclonal antibodies in a study by Moroni et al.²⁰ Nevertheless, the limited number of patients analyzed and contradictory results from other authors warranted caution in the interpretation of these findings, which clearly deserve additional in-depth analysis.²¹ Although great efforts are being made in the study and analysis of EGFR in this setting, less data can be found about other putative molecular determinants, which are part of the EGFR-related biologic cascade.

We investigated the role of NF-kB, a transcription factor that is also activated by the EGFR downstream signaling pathway, in predicting response and global outcome in irinotecan-refractory colorectal cancer patients treated with cetuximab and irinotecan. In our series, patients whose tumors do not express NF-kB experienced a better response rate (48% v 10%), median TTP (6.4 v 3.0 months), and overall survival (15.8 v 9.5 months) than those patients whose tumors express NF-kB. These findings were confirmed at multivariate analysis. NF-kB expression in our study was determined in primary colorectal tumors. Although EGFR immunohistochemistry expression may change between primary colorectal tumors and corresponding metastases, we believe that that this could hardly apply to NF-kB expression. NF-kB activation may in fact result as a late event in the neoplastic transformation process, and therefore we can speculate that its status in metastases should not be dissimilar from what can be observed in primary tumors. This is especially valid when we consider that NF-kB expression can be constitutively activated in many cancer types (including colorectal tumors). However, it is important to note that NF-kB expression is also reported to be activated in response to different stimuli, including chemotherapy. It is then conceivable that NF-kB activation status in vivo, in some cases, may differ after chemotherapy. We believe that only a specifically focused analysis may be able to clarify this biologic mechanism.

Given that EGFR activation is known to trigger NF-kB nuclear translocation, we believe that the observation of EGFR-positive tumors lacking NF-kB activation should deserve more attention. Although we can hypothesize that an unbalance across multiple regulatory factors may be responsible for such an observation, we believe that EGFR immunohistochemical evaluation, as in the present analysis, should be considered inadequate, not only for definite conclusions, but also for additional speculations in this area.

Many preclinical data focused on the role of NF-kB in predicting response to various antineoplastic agents and suggested that an altered expression of this transcription factor may be related to resistance to drugs such as oxaliplatin, fluorouracil, and irinotecan. In fact, NF-kB has been implicated in the inhibition of apoptotic cellular pathways, which represent a key mechanism for chemotherapy-induced cell death. However, clinical experiences confirming these findings in vivo are lacking.

An analysis by Abdel-Latif et al¹² suggested that response to neoadjuvant chemoradiotherapy in esophageal cancer patients was correlated strongly with NF-kB expression, with response rate significantly reduced in NF-kB positive tumors. More recently, a study by Izzo et al¹⁴ addressed again the issue of resistance to chemoradiotherapy in locally advanced esophageal carcinoma patients and determined that activated NF-kB has a major role in determining disease-free and overall survival in this setting. In addition, data derived from breast cancer patients indicated that the NF-kB/bcl-2 pathway may be associated with a poor response to neoadjuvant doxorubicin-based chemotherapy.²² Nevertheless, no data can be found on the role of NF-kB in patients treated with an anti-EGFR strategy.

Cetuximab-induced skin rash is another important parameter in patients treated with cetuximab.^1,2 Its strong association with response and survival, especially when we considered severe skin rash (ie, grade 3 to 4), has been confirmed in our series. Unfortunately, cutaneous reactions can be observed obviously only after the introduction of anti-EGFR therapy, and are then of no use as a pretreatment selection criterion; such pretreatment selection criteria are urgently needed.

Globally, our findings seem to suggest that tumors that constitutively and aberrantly express NF-kB are more likely to be refractory to cetuximab and irinotecan than those that do not show nuclear expression of this transcriptional factor. This also suggests that anti-EGFR therapy does not seem to be able to overcome treatment resistance induced by NF-kB in vivo. Our observations also seem to confirm that NF-kB may play a major role in the apoptotic pathway in tumors. The aberrant activation of this transcriptional factor in our series seems able to preclude response to antineoplastic therapies that are known to trigger the apoptosis-mediated death cell.

Additional analysis in this setting, exploring the interaction between NF-kB and other key-proteins with a regulatory function on apoptotic response in cancer cells, may then represent an innovative field of research for biologic mechanisms involved in resistance to anti-EGFR treatment strategies. These conclusions are of particular relevance not only for the possibility of a more accurate selection of patients suitable for cetuximab-irinotecan treatment, but also for the possible future application with the introduction of drugs potentially able to block NF-kB. In this area, proteosome inhibition may represent an interesting perspective.²³ We believe that our data, along with those already available about the biologic mechanism putatively connected to tumor resistance to anti-EGFR monoclonal antibody, may help compose the molecular mosaic of the various aspects related to the use of this new class of antineoplastic agents.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The author(s) indicated no potential conflicts of interest.

AUTHOR CONTRIBUTIONS

Conception and design: Mario Scartozzi, Stefano Cascinu

Provision of study materials or patients: Italo Bearzi, Chiara Pierantoni, Alessandra Mandolesi, Fotios Loupakis, Alberto Zaniboni, Vincenzo Catalano, Antonello Quadri, Fausto Zorzi, Tommasina Biscotti, Roberto Labianca, Alfredo Falcone, Stefano Cascinu

Collection and assembly of data: Italo Bearzi, Chiara Pierantoni, Alessandra Mandolesi, Fotios Loupakis, Alberto Zaniboni, Vincenzo Catalano, Antonello Quadri, Rossana Berardi, Stefano Cascinu

Data analysis and interpretation: Mario Scartozzi, Rossana Berardi, Stefano Cascinu

Manuscript writing: Mario Scartozzi, Stefano Cascinu

Final approval of manuscript: Mario Scartozzi, Stefano Cascinu