FAS/FAS Ligand Ratio: A Marker of Oxaliplatin-Based Intrinsic and Acquired Resistance in Advanced Colorectal Cancer

Colorectal cancer is the most common cancer in western Europe, with ∼70 new cases a year per 100,000 inhabitants. In spite of advances in screening, 15% to 20% of patients show initially advanced disease, and 30% to 50% are destined to metastasize. Recently, oxaliplatin/5-fluorouracil (5-FU)/leucovorin or irinotecan-5-FU/leucovorin have increased responses in first-line therapy up to 40%, with median survival between 16 and 20 months, but 2-year overall survival still remains <20% (1, 2). Unfortunately, those patients who will benefit from first-line chemotherapy are unknown. Furthermore, the initially sensitive patients become rapidly resistant to current therapies.

The FAS (CD95) receptor is a cell surface protein that mediates apoptotic cell death on triggering by FAS ligand (FASL). This interaction causes FAS receptor homo-oligomerization and this leads to activation of the caspase cascade (apoptotic extrinsic pathway). A FASL-independent activation of the FAS receptor has also been described (3).

Whereas proapoptotic role of FAS and FASL are well known, more conflicting data come from functionality of soluble forms. Various forms of soluble FAS (sFAS) have been described derived from alternative splicing phenomena (4). The majority of these spliced forms have an oligomerization domain, which allows them to form homotrimers (between soluble forms) and heterotrimers (when joining to transmembrane FAS receptor). A dual antiapoptotic or proapoptotic function has been advocate for these soluble forms. When they form heterotrimers, they are counteracting the apoptotic signaling (5). While forming homotrimers, they are capable of interacting with transmembrane FASL leading to a proapoptotic effect (6). More convincing data of an antiapoptotic function of soluble FASL (sFASL; resulting from the cleavage of FASL by metalloproteinase-7; ref. 7) or a marginal proapoptotic function (8) has been proposed.

The extrinsic apoptotic pathway seems to be physiologically compromised during colorectal cancer progression. It has been shown that adenoma through carcinoma step leads to FASL up-regulation and FAS down-regulation (9). sFAS levels have been proven to be elevated in serum of patients with colorectal cancer (10), whereas some colorectal cell lines have turned to be releasing sFASL (11). Together, all these data support the hypothesis of the acquisition of a FAS/FASL apoptotic resistance profile as well as an immunoescape capacity during colorectal cancer progression (12).

Cytotoxicity due to 5-FU/leucovorin treatment in p53 wild-type colorectal cell lines can be mediated via FAS (13, 14). It has also been described that this stimulus can produce apoptosis in p53 mutant cells (15). In colorectal cancer cells, there was an increased level of FASL and apoptosis induction during thymineless death after 5-FU treatment, via activation of nuclear factor-κB and activator protein-1 transcription factors (16), as well as in thymidylate synthase–deficient cells, after treatment with DNA-damaging agents (17). Other drugs, such as camptothecin, seem to induce cell death through recruitment of the FAS-FADD adaptor in a FASL-independent fashion (18, 19). Therefore, it seems that 5-FU (20), capecitabine (21), and antimetabolite therapies can restore the lost of apoptotic capacity of colorectal cancer cells in vitro, either p53 wild-type or mutant, through the extrinsic pathway by regulating FAS and FASL expression and/or function.

Because of the mentioned chemotherapy capacity to modulate FAS/FASL, we hypothesize that these drugs could also modulate the soluble forms and therefore its role in regulating the apoptotic response through the extrinsic pathway as well as the immunologic “counterattack.” Because soluble forms (sFAS and sFASL) can have opposite effects, the ratio between them (sFAS/sFASL) could be a way to measure the final balance of apoptotic and immunoescape effect. This ratio and its variations along chemotherapy treatment could be therefore a useful variable to measure colorectal cancer chemosensitivity and chemoresistance.

Patients. Blood samples were obtained from 68 patients treated for advanced colorectal cancer from July 2001 to September 2003. Patients received 85 mg/m² oxaliplatin on day 1, 200 mg/m² leucovorin on day 1, and 3 g/m² 5-FU on day 1 in 48-hour continuous infusion every 2 weeks for a maximum of 12 cycles (n = 55) as standard treatment in our institution. Thirteen patients were treated with other oxaliplatin-fluorouracil combinations in multi-institutional clinical trials: 85 mg/m² oxaliplatin on day 1 and 2.25 g/m² 5-FU on day 1 in 48-hour continuous infusion weekly every 2 weeks (n = 3), 130 mg/m² oxaliplatin on day 1, and 1,000 mg/m² capecitabine on days 1 to 14 every 3 weeks (n = 3) or FOLFOX-4 (n = 7). Eligible criteria were stage IV histologically proven colorectal cancer, measurable metastatic lesions by Response Evaluation Criteria in Solid Tumors criteria, Eastern Cooperative Oncology Group performance status score of 0 to 2, no previous neoplasm in the last 10 years, normal liver and renal function, and no previous chemotherapy for advanced disease.

All patients had chest X-ray and a helical computed tomographic (CT) abdominal scan before entry into study and underwent repeated evaluations at least every 3 months. Tumor response was assessed according to Response Evaluation Criteria in Solid Tumors criteria (22) as complete response, partial response, stable disease, and progressive disease. Each tumor measurement by CT scan was compared with previous CT scan. Therefore, patients with initial partial response in first evaluation (second CT versus initial CT) and with stabilization on the second evaluation (third CT versus second CT) were defined as stable disease instead of confirmed partial response. Only those patients with new partial response in second evaluation were defined again as partial response. Patients gave signed informed consent before treatment and the study was approved by the institutional ethics of research committee.

Samples and assay. Venous blood samples were drawn into sterile vacuum tubes before the initial treatment and every 3 months until disease progression for a maximum of five extractions (month 12). We have limited the number of extractions to 5 because >90% of the patients have progressed at that time. Blood samples were kept at 4°C, centrifuged at 10,000 rpm for 15 minutes, and then immediately frozen at −80° until assayed.

FAS and FAS ligand–specific ELISA. A double-antibody sandwich ELISA was constructed to detect sFAS and sFASL in sera using a sFAS and sFASL ELISA kit (Oncogene Research Product, San Diego, CA). This assay uses FAS and FASL antibodies against two epitopes. Standard curves were constructed using serial dilutions of recombinant sFAS and sFASL. The maximum detectable concentration of sFAS was determined as 100 ng/mL. The maximum and minimum detectable concentrations of sFASL were determined as 1.25 and 0.01 ng/mL, respectively.

Statistical methods. The Mann-Whitney test was used to assess significant associations between continuous variables (FAS and FASL levels) and dichotomous variables [sex, upper limit of normal lactate dehydrogenase (>1 versus <1), number of organs involved (1 versus >1), disease location (liver versus other than liver), adjuvant chemotherapy, previous radiotherapy, and initial Dukes stage (synchronic versus metachronic)]. The Wilcoxon test was also used to ascertain FAS and FASL variations during chemotherapy treatment. The Kruskal-Wallis test was used to assess significant differences in FAS and FASL levels within multiple groups (i.e., Eastern Cooperative Oncology Group performance status). Complete response and partial response were considered as “sensitive” and stable disease and progressive disease were considered as “refractory.” A random effect regression statistical model evaluated the effects of time/therapy on response. A univariate and multivariate analysis for all variables influencing on response was also done. Only variables with a borderline significance (P < 0.1) at univariate analysis were include in the multivariate regression model.

Patients and tumor characteristics. Demographic details on the 68 patients included in the study and tumor stage are shown in Table 1. The median of received treatment cycles was 9 (range, 1-12). Twelve of the 68 patients had undergone radical procedures after chemotherapy treatment (11 underwent surgical resection and 1 radiofrequency thermal ablation) but were fully evaluable for response. Two patients were not evaluable for response due to complications after first cycle (1p with pulmonary embolism and 1p with intestinal occlusion).

FAS/FAS ligand levels. Sera were obtained from 68 patients diagnosed with advanced colorectal cancer during the study period with a total of 160 extractions. From 66 patients assessable for response, the average of extractions was 2.4 (range, 1-5). Reasons for extraction discontinuation were per protocol (n = 0.21; median, 3.4; range, 2-5), radical treatment after chemotherapy (n = 0.12; median, 2.2; range, 1-3), patient withdrawal consent (n = 0.1; median, 2), poor medical condition after rapid progression disease (n = 0.4; median, 1), and finished study period (n = 0.28; median, 1.8; range, 1-4).

There were no significant associations between sFAS and sFASL levels and any of the following variables: sex (P = 0.24 and 0.38, respectively), previous chemotherapy treatment (P = 0.32 and 0.35, respectively), lactate dehydrogenase levels (P = 0.43 and 0.77, respectively), previous radiotherapy (P = 0.39 and 0.9, respectively), synchronic or metachronic disease (P = 0.37 and 0.21, respectively), number of organs involved (P = 0.45 and 0.31, respectively), and liver involvement (P = 0.42 and 0.39, respectively). There were also no significant differences between sFAS and sFASL levels among patients with different performance status grades (P = 0.10 and 0.51, respectively; see Table 2).

We found a significant increase in sFAS levels and a significant decrease in FASL at 3 months compared with baseline (13.2 versus 10.02 ng/mL; P = 0.0001; 0.07 versus 0.14 ng/mL; P = 0.007, respectively). The median of FAS/FASL ratio increment was 1.2-fold. A significant increase in the sFASL levels up to 9 months (fourth to fifth extractions; 0.26 ng/mL) of therapy compared with first to third extractions (0.11 ng/mL; P = 0.003) was also found (see Table 3).

Response to chemotherapy. The overall response rate was 45.6%. The levels of the FAS/FASL ratio increment in the group of complete response and partial response (i.e., “responding” tumors; mean, 14.2; range, 0.06-188.4) were significantly different from the levels in the stable disease and progressive disease group (i.e., “nonresponding” tumors; mean, 2.2; range, 0.02-29.2; P = 0.005, Wilcoxon test; Table 4). A random effect regression statistical model evaluated the effects of time/therapy on response. We determined that a >1.2-fold increase in sFAS/sFASL ratio was a marker of chemosensitivity (P = 0.001). In addition, we have found a predictor of chemoresistance in a subgroup of patients who, despite presenting a high ratio and an initial CT response, rapidly developed a decreased ratio during treatment, indicating the appearance of chemoresistance. In the univariate analysis of response, only performance status (P = 0.05) and age (P = 0.1), but not lactate dehydrogenase (P = 0.7), previous adjuvant treatment (P = 0.38), carcinoembryonic antigen (P = 0.33), and number of organs involved (P = 0.93), had a borderline significance. A multivariate regression analysis of response with the relevant clinical variables (age, performance status, and sFAS/sFASL ratio) was done, and only sFAS/sFASL ratio (P = 0.003) and age (P = 0.025) remain as independent factors predicting response.

In the present study, the mean of sFAS/sFASL basal levels (sFAS, 10.02 pg/mL; sFASL, 0.14 pg/mL) is similar to that reported previously (23–25). In accordance to some authors, we have not seen any significant relation between sFAS and/or sFASL levels and variables such sex, age, or performance status (23, 24). We have also observed a lower basal level sFAS (8.2 ng/mL), but without reaching significance (P = 0.37), in patients with metachronic compared with synchronic disease (10.6 ng/mL), in accordance with the well-known chemoresistance of this group of patients in randomized advanced colorectal cancer trials (2, 26). We also noted a higher basal levels of sFAS (12.2 ng/mL) in those patients with serum lactate dehydrogenase >1 upper limit of normal compared with 8.5 ng/mL (P = 0.43), also a well-defined, poor prognosis factor of survival in colorectal cancer (1, 2) and described previously in advanced melanoma (27). These data could explain previous reports associating poor prognosis with sFAS levels in gynecologic malignancies and melanoma (24, 25).

Also in our knowledge, for the first time in the literature, we have shown a significant increase in sFAS levels after chemotherapy treatment (P = 0.0001). In addition, we have also noted that a ratio increment correlates with tumor response and the subsequent decrease is related to chemoresistance (P = 0.001). Despite these data, it is unclear how chemotherapy regulates, if it does, sFAS and sFASL functions. We hypothesize that, in advanced colorectal cancer, tumor production of soluble splicing variants (amount and type) leads to a proapoptotic action (through transmembrane FAS interaction), much more than to an antiapoptotic one. Because in the advanced stages of the disease the matrix metalloproteinases (like matrix metalloproteinase-7) are more active (27–29) and lead to an increase of the sFASL fractions, it is plausible that these events may have a global antiapoptotic and immunoevading action. Supporting this theory, high levels of sFASL have been observed in metastatic pancreatic carcinoma, a notorious resistant neoplasm (30). In mammary tissues from multiparous matrilysin (matrix metalloproteinase-7)–expressing mice, there was decreased FASL expression, suggesting that loss of FASL expression is at least one mechanism of matrilysin-induced resistance to apoptosis (31). Furthermore, CTLs trigger FAS-mediated apoptosis only after treatment with metalloproteinase inhibitors (matrix metalloproteinase-1). Matrix metalloproteinase-1 induces apoptotosis by increasing the surface expression of FASL and disappearance of sFASL (32).

There have been multiple reports in the literature measuring basal levels of sFAS and sFASL in different neoplasms. However, this is the first study that reflects the dynamics of these soluble fractions during chemotherapy treatment. We conclude that a 1.2-fold increase of FAS/FASL ratio, after receiving chemotherapy, indicates chemosensitivity in colorectal cancer. In addition, a ratio decrease during chemotherapy treatment, despite the initial values, is related to acquired chemoresistance. We suggest that sFAS/sFASL ratio can be useful as a dynamic response predictor in colorectal cancer patients following chemotherapy.