Conflict of Interest: B. Clinchy, A. Håkansson, B. Gustafsson, R. Sjödahl and L. Håkansson may have a potential conflict of interest in a future commercial application of some of the material included in this article; patent pending. The other authors declare no conflict of interest.
Colorectal cancer is one of the most common forms of cancer in the Western world. Staging based on histopathology is currently the most accurate predictor of outcome after surgery. Colorectal cancer is curable if treated at an early stage (stage I-III). However, for tumors in stages II and III there is a great need for tests giving more accurate prognostic information defining the patient population in need of closer follow-up and/or adjuvant therapy. Furthermore, tests that provide prognostic information preoperatively could provide a guide both for preoperative oncologic treatment and the surgical procedure.
Peripheral blood mononuclear cells (PBMCs) were isolated preoperatively, within a week before primary surgery, from 39 patients undergoing surgery for colorectal cancer. The PBMCs were cultured in vitro for 24 hours in the presence of autologous serum and lipopolysaccharide (LPS). Interleukin-6 (IL-6) production was measured with enzyme-linked immunosorbent assay (ELISA). Staging based on histopathology was performed in all patients. Patients were followed for at least 54 months.
A production of >5000 pg/mL of IL-6 identified colorectal cancer patients with a poor prognosis. Eight out of 13 patients with >5000 pg/mL IL-6 died from cancer within the follow-up period, whereas no cancer-related deaths were recorded among 21 patients with 5000 pg/mL IL-6 or less. A multivariate Cox regression analysis, stratified for T- and N-stage, identified IL-6 production as an independent prognostic factor.
Several factors of prognostic significance for colorectal cancer have been identified. Staging based on the TNM classification or Dukes classification gives good prognostic information, but still there is a great need for tests giving more accurate prognostic information, especially for patients with stage II and III (Dukes B and C) cancers, as each of these stages still include patients with highly divergent prognosis.
The prognostic value of several molecular and genetic factors has been investigated.1 Analyses of a multitude of such markers (eg, alterations in p53, k-ras, c-myc, and chromosome 18q21 locus) are contradictory and the clinical impact is still low.2, 3 Neither has the improved ability to detect occult disease, for example, by reverse-transcriptase polymerase chain reaction (RT-PCR) for tumor associated antigens, led to a more accurate prediction of treatment outcome for the individual patient.4 Global gene expression profiling of primary tumors has revealed patterns that are associated with metastases and poor prognosis.4 However, these techniques are still far from being implemented in the clinical setting. Circulating carcinoembryonic antigen (CEA) is often elevated in colorectal cancer and high levels of CEA predict a more advanced stage and worse clinical outcome.5 A correlation between infiltration of lymphocytes and prolonged survival has also been described in colorectal cancer.6 In particular, infiltrating CD8+ T cells may be a prognostic factor after surgery with an impact similar to Dukes staging.7 So far no single parameter that allows individual monitoring of colorectal cancer patients has been described.
Cancer is often associated with a systemic chronic inflammation resulting in production of cytokines, eg, interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α) and induction of acute phase reactants, such as C-reactive protein (CRP). High serum levels of CRP has been reported to correlate with poor prognosis, but still a fairly high recurrence rate is found in patients with normal serum levels of CRP,8, 9 thus reducing the value of this parameter for individual monitoring of cancer patients. Therefore, scores based on several parameters, eg, serum level of CRP and TNM stage, have been proposed.10
IL-6 is a pleiotropic, cytokine of importance for immunoregulation.11 An increased serum concentration of this cytokine is often found in cancer patients, especially in patients with advanced disease, and has been reported to correlate with a poor prognosis in various types of cancer, eg, multiple myeloma,12 chronic lymphocytic leukemia (CLL),13 renal cell carcinoma,14 prostate cancer,15 ovarian cancer,16 metastatic breast cancer,17, 18 and pancreatic cancer.19
Colorectal and other gastrointestinal carcinoma patients often have elevated serum levels of IL-6 as compared with healthy controls and several of these studies have reported a correlation with tumor stage.20–27 Belluco et al.23 found that an IL-6 concentration of more than 10 pg/mL was an independent prognostic factor of survival. The recurrence rate at IL-6 levels <10 pg/mL was 31% as compared with 55% at levels >10 pg/mL. Similarly, an elevated serum concentration of IL-6 was found to correlate with tumor size or stage22, 27 or prognosis.24 However, in several of these reports serum IL-6 was not found to be an independent prognostic indicator.24, 27 In contrast to these results, no significant difference in disease-free survival among 50 radically resected gastric and colon cancer patients was noted between patients with high vs low preoperative IL-6 serum concentrations.25 Thus, serum level of IL-6 is clearly of limited value at present for predicting recurrences on an individual basis.
The source of serum IL-6 in cancer patients is still unclear, but it is generally assumed to be derived from the tumor.22, 28–30 However, a broad range of normal cells such as fibroblasts, monocytes, T lymphocytes, endothelial cells, and keratinocytes can produce IL-6.11 Blood mononuclear cells from patients with gastric and colon cancer have been reported to produce cytokines like TNF-α and IL-6 in vitro.31, 32
In the present study lipopolysaccharide (LPS)-induced IL-6 production by peripheral blood mononuclear cells (PBMCs) from colorectal cancer patients was determined preoperatively and a possible correlation to the prognosis of these patients was analyzed.
MATERIALS AND METHODS
Thirty-nine patients with colorectal cancer were included in the present study. Five of these patients had synchronous distant metastases detected by ultrasonographic investigation of the liver and x-ray of the lung and are therefore excluded from the following analyses. Further clinicopathologic factors are presented in Table 1. The median follow-up time was 58.6 months and all patients were followed for at least 54 months.
Table 1. Clinicopathologic Factors of Patients With Colorectal Carcinoma (n = 34)
Median age [range]
Location of tumor
Lymph node metastasis
All tumors were staged based on the TNM classification system and the histopathologic investigation of the tumors was reviewed by a panel of surgeons and oncologists.
Five patients with rectal cancer were given preoperative radiotherapy according to standard procedures (25 Gy). One patient with rectal cancer had locally advanced disease and was therefore given preoperative radiotherapy to a dose of 44 Gy. One of these patients with locally advanced disease also received adjuvant treatment with 5-fluorouracil/leucovorin. Six patients with colon cancer received adjuvant treatment with 5-fluorouracil/leucovorin, 1 patient received capecitabine.
Isolation of PBMCs
Blood samples were collected from healthy volunteers or from colorectal carcinoma patients within 1 week before surgery. Blood was drawn in glass vacuum tubes with acid dextrose citrate solution A as anticoagulant (Vacutainer, Becton Dickinson, Franklin Lakes, NJ). Erythrocytes were removed by sedimentation on 2% dextran T500 solution (Amersham Pharmacia Biotech AB, Uppsala, Sweden) in 0.9% NaCl. Mononuclear cells (PBMC) were then isolated by Ficoll-paque Plus (Pharmacia) density gradient centrifugation after which the cells were washed twice in RPMI 1640 Dutch modification (Gibco BRL, Life Technologies, Paisley, Scotland) with 2% human serum albumin (HSA, Pharmacia & Upjohn, Stockholm, Sweden) (RPMI/2%HSA). Cell viability was assessed by exclusion of 0.05% Trypan Blue and was always above 95%. The cell suspension was stained with Turk's solution and the number of lymphocytes and monocytes in the PBMC preparation were counted in a hemocytometer. PBMCs were resuspended in RPMI/2%HSA and the cell concentration adjusted to 5 × 105 lymphocytes/mL.
Serum was collected in serum collection glass tubes without additives (Vacutainer, Becton Dickinson) at the same time as blood samples for isolation of PBMC. Serum, obtained from healthy volunteers or from patients with colorectal carcinoma, was heat-inactivated at 56°C for 30 minutes and used fresh the same day.
Culture of PBMCs for generation of cell culture supernatants
PBMC were cultured in round-bottomed, tissue culture microtiter plates (Costar 3799, Corning, Corning, NY); 100 μL per well of culture medium consisting of RPMI 1640 supplemented with 200 IU/mL penicillin, 200 μg/mL streptomycin, 4 mM L-glutamine (all from Sigma Chemical, St. Louis, Mo), and 20% fresh, heat-inactivated serum was then added to the microtiter plates followed by 100 μL per well of PBMC suspension in RPMI/2%HSA. LPS from Escherichia coli 026:B6 (Sigma Chemical) was also added at a final concentration of 0.05 ng/mL. The cells were cultured in a humidified 5% CO2 atmosphere at 37°C. Supernatants (SNs) were harvested after 24 hours and residual cells removed by centrifugation in a refrigerated centrifuge at 2600g for 5 minutes. SNs were frozen and stored at −70°C until IL-6 concentrations were evaluated by enzyme-linked immunosorbent assay (ELISA).
ELISA for IL-6
IL-6 in cell culture SN or in serum was measured by ELISA using the DuoSet ELISA development kit or Quantikine ELISA kit for human IL-6 (Both from R&D Systems Europe, Abingdon, UK) following the manufacturer's recommended procedures. The lower limit of detection was 3.1 pg/mL. All samples were analyzed as duplicates. Both SNs and sera had been kept frozen at −70°C before determination of IL-6.
Survival curves were plotted using the method of Kaplan and Meier and time to progression and survival comparisons between subgroups were performed using the log-rank test. In addition, the prognostic significance of the level of LPS-stimulated IL-6 production was also calculated using Cox regression. A multivariate analysis, stratified for T- and N-stage, was also performed.
This study was approved by the local Ethics Committee. Oral informed consent was obtained from all participating patients and control subjects.
PBMCs were isolated from 39 patients with colorectal cancer preoperatively and were analyzed for spontaneous and for LPS-induced IL-6 production in short-term cultures with autologous sera in the culture medium.
The spontaneous IL-6 production by PBMCs was very low in both cancer patients (mean, 223 ± 48; range, 31–1120) and controls (mean, 162 ± 52; range, 20–1620) and was detectable in 9 out of 39 (23%) patients and 12 out of 35 (34%) controls. After LPS stimulation, PBMCs from colorectal carcinoma patients produced more IL-6 (mean, 6442 ± 1510; range, 31–32,400) than PBMCs from healthy controls (mean, 3899 ± 779; range, 57–24,000), P = .116 (Fig. 1). An analysis of the distribution of IL-6 producers among healthy controls and cancer patients showed that a production of 5000 pg/mL discriminated between the 2 groups, which in the following analysis are called low and high producers. Spontaneous IL-6 production was at most 15% of LPS-stimulated IL-6 production in high producers. Adjuvant treatment was given to 4 patients in each of these groups. Preoperative radiotherapy was given to 6 patients, 4 low and 2 high producers. This treatment does not seem to have any influence on the present results as both high and low IL-6 production was found in the irradiated group of patients.
No significant difference was found in IL-6 production by PBMCs due to stage, occurrence of lymph node metastases, ulceration, vascular invasion of the tumor tissue, or degree of differentiation in the tumor tissue. There was a slight trend toward a higher IL-6 production by PBMCs from patients with left-sided tumors as compared with right-sided (P = .3) and PBMCs from the few patients (n = 3) with T4 tumors had a significantly higher IL-6 production than those from T2 (P < .044) or T3 tumors (P < .0005).
The effect of tumor stage, tumor extension, and lymph node metastases on survival is shown in Figure 2. These results are in good agreement with previously published data.
A possible correlation between the overall survival of colorectal cancer patients and IL-6 production by their PBMCs was analyzed. When patients with radical resection were analyzed, low producers of IL-6 had a significantly better survival compared with high producers (Fig. 3A). Twenty-one patients had low IL-6 production. No cancer-related deaths were found in this group. Three patients, 80 years old, had died due to other causes. In contrast, 8 out of 13 patients with high production of IL-6 had died from cancer (P < .001) and 2 patients in this group due to other causes. The median survival was not reached in the group of low producers but was 32 months for high producers. To demonstrate that the cutoff between high and low producers of IL-6 at 5000 pg/mL is by no means critical for the present results, a Cox regression analysis was also performed with very similar outcome.
Two of the 21 patients, who are alive, have developed regional recurrences. One had a local recurrence after 58 months and 1 had retroperitoneal lymph node metastases 30 months after primary surgery. Interestingly, PBMCs from the latter patient did not respond at all to LPS.
In the group of patients with lymph node metastases (N1, N2), IL-6 production identified those with a poor prognosis. As shown in Figure 3B, all 6 patients with low IL-6 production were alive, whereas 7 out of 8 with high production had died from their cancer. Similarly, patients with T3 tumors and low IL-6 production had a significantly better survival than high producers (Fig. 3C). In the former group, 1 patient had died for other reasons than cancer; the other 11 were alive compared with high producers, where 5 out of 8 patients had died from their cancer. Even in the subset of T3 and N1 or N2 patients, low IL-6 production predicts a long-term event-free survival (6 out of 6 patients), whereas 4 out of 5 patients with high production have died from cancer (Fig. 3D). Multivariate Cox regression analysis stratified for T- and N-stage showed LPS-stimulated IL-6 production to be an independent prognostic factor, both when the patients were analyzed as 2 groups, high and low producers, and as 1 group (P < .001).
Accurate prediction of treatment outcome after microscopically radical surgery of primary malignant tumors is of great importance in order to ensure proper follow-up procedures and identification of high-risk patients for whom adjuvant treatment should be considered. Prognostic information based on clinical or histopathologic investigations or molecular markers is insufficient. Both within stage II and III the survival of colorectal cancer patients may show a wide range. Several molecular markers have shown correlation to the prognosis of patients, but their sensitivity or specificity are generally not good enough to allow individual monitoring of cancer patients.1–3 Therefore, there is a need for other prognostic instruments in addition to the TNM classification and molecular or genetic markers. Optimally, the prognostic information should be available even before surgery of the primary tumor, which possibly then could have an impact both on preoperative oncologic treatment and the extent of the surgical procedures.
In the present investigation we demonstrated the prognostic value of LPS-stimulated IL-6 production analyzed 1 week before surgery. Eight out of 13 patients with an increased IL-6 production in this test died from their cancer, whereas no cancer-related deaths were found in 21 patients with low IL-6 production. A multivariate Cox regression analysis, stratified for T- and N-stage, identified IL-6 production as an independent prognostic factor.
The aim of the present investigation was to create an in vitro test model, where the results correlate to in vivo conditions such as risk of recurrence of the resected cancer, overall survival, and prognosis. The basic hypothesis behind the development of this in vitro test model was that the immune system in cancer patients is dysregulated and that the often enhanced serum concentration of IL-6 in cancer patients is due to the influence of malignant tumors on the immune system. Therefore, PBMCs were stimulated with LPS at a low concentration in order to find out if PBMCs from cancer patients are influenced by the malignant disease in such a way that it allows discrimination between patients with different prognosis. However, the use of LPS-stimulation in culture models, where the intention is to obtain data of prognostic value, has to be interpreted with great care, as some 10% of the population carries a mutation resulting in no response to LPS.33
In setting up in vitro culture models, there is always the risk that the regulatory mechanisms operating in vivo are changed or lost. Therefore, a culture model with minimal manipulation of the mononuclear blood cells was created. The end results of the immune system, such as monokine production, often depend on interaction between different cell types such as monocytes and lymphocytes.34 Thus, no attempts were made to further purify monocyte or lymphocyte subsets, as their interaction might be crucial to maintain the in vivo regulation of cytokine production.
Furthermore, many studies investigating cytokine production by PBMCs from cancer patients have used fetal calf serum rather than autologous serum in the culture medium. This is advantageous from the point of standardizing the culture conditions, but it may introduce new unknown factors in the culture model and cancer-related autologous, possibly immune regulatory serum factors are removed. Thus, all cultures in the present investigation used autologous sera in the culture medium.
Although an increased serum concentration of IL-6 has been found to correlate with a poor prognosis in several types of cancers,12–19, 22–24, 27 the information provided by serum concentrations of factors such as IL-6 can be questioned. The cellular origin of these serum factors is often unknown, they are immensely diluted in serum, and their half-life is not accounted for. The information obtained from serum concentrations of such factors is therefore most likely just a faint reflection of the mechanisms behind their appearance in serum.
Usually, when the prognostic value of the serum IL-6 concentration in colorectal cancer patients was analyzed a cutoff around 8–13 pg/mL was used.22–25, 27 It is possible that some of the IL-6 detected by us in the culture supernatants was already present endogenously in the autologous serum in the culture medium. However, in the absence of LPS, levels of IL-6 in the culture supernatants from high producers never exceeded 15% of the LPS-stimulated level. Hence, serum concentrations of IL-6 or production of IL-6 in the absence of LPS can only marginally affect our results, which have a cutoff value of 5000 pg/mL.
IL-6 in serum from cancer patients is often assumed to be derived from the tumor cells. Malignant cells, including colorectal cancer tumor cells, have been found to produce IL-6.22, 28–30 In colorectal carcinoma, IL-6 has been found localized in the tumor cytoplasm by immunohistochemical staining, indicating that these cells are capable of IL-6 production in situ.22, 30 The IL-6 concentration in tumor tissue was significantly higher than in normal mucosa and correlated with the serum IL-6 concentration.22 In a study by Piancatelli et al.35 IL-6 mRNA was expressed in tumor tissue in 83% of patients with colorectal carcinoma. However, the majority of tumor tissue specimens also contained lymphocytic infiltrates, making it impossible to determine the origin of the IL-6. Furthermore, no correlation between the serum IL-6 level and tumor IL-6 mRNA expression could be found in that study.35
Cells of the immune system have the capacity to produce IL-6, as have a broad spectrum of other cell types including endothelial cells.11 Based on the present data on LPS-induced IL-6 production by PBMC, cancer patients as well as healthy controls can be divided into 2 groups, low-producers (≤5000 pg/mL) and high-producers (>5000 pg/mL). As also PBMCs from some healthy controls showed an increased LPS-induced production of IL-6, this phenomenon is most likely involved in normal immune regulation, possibly occurring during a short period when the immune system has been activated due to infections or inflammatory processes. In cancer patients down-regulation of this type of protective immunity is dangerous, because the inflammatory reaction is due to the presence of a malignant tumor, which sustains long-lasting down-regulatory mechanisms. This results in a poor protection against systemic dissemination of tumor cells.
Two patients with low LPS-induced IL-6 production had intra-abdominal recurrences, 1 a local recurrence diagnosed after 58 months and the other retroperitoneal lymph node metastases diagnosed after 30 months. This is hardly surprising, as the test used in the present investigation obviously does not correlate with the local aggressiveness of the tumor, shown as vessel invasion or lymph node metastases. However, the overall survival was significantly better for low-producers of IL-6. This indicates that the production of cancer-related factors supporting LPS-induced IL-6 production correlate with its propensity to establish distant metastases. Alternatively, there might be a regional/systemic gradient of these cancer-related factors modulating the immune function so that the regional tumor control is broken down while systemic protection is still active.36
It is interesting to note that PBMCs from the patient who had lymph node metastases after 30 months did not respond at all to LPS. He might thus carry the mutation discussed above and therefore cannot be identified using the present prognostic test.
In summary, the present investigation shows that LPS-induced IL-6 production by PBMCs from colorectal cancer patients determined preoperatively correlates with the prognosis of these patients. This method also identifies patients with good prognosis among those who according to the TNM classification have poor prognosis. The fact that this test is performed before surgery is of particular value, as this prognostic information can then be used for planning the surgical procedure. In addition, these results indicate that the immune system plays a major role in the control of malignant tumors.
We thank Prof. John Carstensen, Department of Health and Society, Tema Research Institute, Linkoping University, for assistance with statistical analysis.