Cytokines in pancreatic carcinoma

Correlation with phenotypic characteristics and prognosis

Authors

  • Behnam Ebrahimi M.D.,

    1. Department of Gastrointestinal Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Susan L. Tucker Ph.D.,

    1. Department of Biomathematics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Donghui Li Ph.D.,

    1. Department of Gastrointestinal Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • James L. Abbruzzese M.D.,

    1. Department of Gastrointestinal Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Razelle Kurzrock M.D.

    Corresponding author
    1. Department of Gastrointestinal Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    2. Phase I Program, Division of Cancer Medicine, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    • Phase I Program, Division of Cancer Medicine, The University of Texas M. D. Anderson Cancer Center, Box 422, 1515 Holcombe Boulevard, Houston, TX 77030
    Search for more papers by this author
    • Fax: (713) 745-2374


Abstract

BACKGROUND

Cytokines have been implicated in diverse processes that are relevant to pancreatic carcinoma, including cachexia, asthenia, and tumor growth. The objective of this study was to examine the association between serum levels of proinflammatory and antiinflammatory or angiogenic cytokines and the outcomes of patients with pancreatic carcinoma.

METHODS

Serum cytokine levels were measured by enzyme-linked immunosorbent assay from 51 patients with pancreatic carcinoma and from 48–62 healthy volunteers. Cytokine levels were compared with disease manifestations and overall survival.

RESULTS

Circulating levels of vascular endothelial growth factor, tumor necrosis factor alpha, interleukin-1α (IL-1α), and IL-1β were not elevated significantly in patients with pancreatic carcinoma, but levels of IL-6, IL-8, IL-10, and IL-1 receptor antagonist (IL-1RA) were elevated significantly (P < 0.05). Cytokine levels were dichotomized based on an analysis of null Martingale residuals. Patients who had IL-6 levels > 5.2 pg/mL or IL-10 levels > 9.8 pg/mL had significantly worse survival compared with patients who had lower IL-6 or IL-10 levels (P < 0.05). IL-8 levels were not associated with survival differences. Patients who had IL-1RA levels < 159 pg/mL had significantly worse survival compared with patients who had higher IL-1RA levels (P < 0.05). Higher IL-6, IL-10, and IL-8 levels were associated with poor performance status and/or weight loss. In multivariate analysis, only T4 tumors and high IL-6 levels were selected as independent prognostic factors for poor survival.

CONCLUSIONS

Circulating levels of several cytokines were high in patients with pancreatic carcinoma, and their association with weight loss and poor performance status suggested that they may be involved in these disease manifestations. Furthermore, serum cytokine levels, in particular IL-6, may be a useful prognostic marker. Cancer 2004. © 2004 American Cancer Society.

Pancreatic adenocarcinoma is the fifth leading cause of cancer-related deaths in the United States, with approximately 30,000 new patients diagnosed per year, virtually all of whom succumb to the disease. This type of pancreatic carcinoma almost always is refractory to conventional therapies.1 The major cause of death in patients with pancreatic adenocarcinoma is metastases. Patients also suffer from a variety of disconcerting symptoms, including profound cachexia and deterioration in performance status, even when their tumor burden is low. The etiology of these paraneoplastic manifestations is unclear but may be related to the production of cytokines by the tumor or the host. An understanding of the biology and pathogenesis of pancreatic carcinoma is crucial to improving treatments for patients with this disease.

Endogenous cytokines are produced aberrantly in many malignances and serve as autocrine growth factors or as indicators of immune response to the tumors. Hence, it is likely that cytokine deregulation participates in the development or evolution of the malignant process. Over the last few years, endogenous cytokine levels have been correlated with phenotypic manifestations of carcinoma and with prognosis both in patients with hematologic malignancies and in patients with solid tumors.2–11 For instance, serum interleukin-6 (IL-6) levels are elevated in both recurrent and newly diagnosed lymphoma and correlate with established prognostic features.4–8 Furthermore, in diffuse large cell lymphoma, the serum IL-6 level is an independent prognostic variable for both complete remission and failure-free survival.12, 13 Serum IL-10 levels also are elevated in lymphoid malignancies and predict outcome.14, 15 IL-8 may promote tumor growth through its proangiogenic effects. In some patients, the balance between endogenous cytokine agonists and antagonists may be disrupted. For instance, in chronic myelogenous leukemia, high cellular levels of IL-1β and low levels of the naturally occurring antagonist IL-1 receptor antagonist (IL-1RA) are seen in advanced disease and correlate with reduced survival.16, 17

There is a paucity of biologic prognostic factors for pancreatic carcinoma. Because cytokines have potent immune-regulatory and growth-regulatory effects that may be involved in the pathogenesis of pancreatic carcinoma, we examined circulating levels of several proinflammatory, antiinflammatory, and angiogenic cytokines in patients with this disease. Our observations suggest a significant association between aberrant levels of several cytokines (IL-8, IL-10, IL-6, and IL-1RA) and phenotypic manifestations of pancreatic carcinoma, including cachexia and asthenia, as well as an association with outcome.

MATERIALS AND METHODS

Patients

The study group consisted of 51 patients with a diagnosis of pancreatic adenocarcinoma, Frozen sera (at − 70 °C) were obtained at the time patients presented at The University of Texas M. D. Anderson Cancer Center. All patients either had not started treatment or had been off treatment for at least 1 month. The samples were derived from all consenting patients who met the criteria for analysis and who were seen over a 16-month period. All histologic material was reviewed by a pathologist at The University of Texas M. D. Anderson Cancer Center. The clinical factors assessed in this study represent patients' status at the time of sample draw (which occurred at the time they were referred to the University of Texas M. D. Anderson Cancer Center). All patients provided informed consent in accordance with institutional policy.

Control Group

Cytokine levels were measured in 48–62 healthy normal donors, depending on the cytokine assessed. Samples were collected from control participants only if they had not had fever within 1 week, were not receiving any medications, were not known to be pregnant, and did not have a history of any chronic or acute illnesses. Control samples were frozen and stored in a manner identical to the handling of patient samples. Control participants provided informed consent in accordance with institutional policy.

Cytokine Assays

All cytokine levels were measured by enzyme-linked immunosorbent assay (ELISA). Samples were assayed in duplicate with all values expressed as a mean of the two determinations. A standard curve was generated using known concentrations of recombinant cytokine. The concentration of a particular cytokine was then determined from the standard curve. IL-1RA, IL-1α, IL-1β, IL-6, IL-8, tumor necrosis factor alpha (TNF-α), and vascular endothelial growth factor (VEGF) levels were assayed using a validated commercial ELISA (Quantikine R & D Systems, Minneapolis, MN). The IL-10 levels were measured by using the Endogen Inc. assay (Cambridge, MA). The lower limits of assay sensitivity are as follows: IL-1RA (22 pg/mL), IL-1α (0.5 pg/mL), IL-1β (0.3 pg/mL), IL-6 (0.7 pg/mL), IL-8 (10 pg/mL), IL-10 (3 pg/mL), TNF-α (4.4 pg/mL), and VEGF (9 pg/mL), respectively.

Statistical Analysis

All statistical analyses were performed by our biomathematician (S.L.T.). Patient survival was computed from the date of the serum sample draw to the date of death, with surviving patients censored at the date of last follow-up. Survival curves were calculated using the method of Kaplan and Meier. Univariate analyses of survival were performed using the log-rank test, and multivariate analyses were performed using a Cox proportional hazards model. For all multivariate analyses, forward stepwise procedures were used. Cut-off points for dividing patients into subgroups with respect to cytokine levels were identified by minimizing the sum of the variances of the null Martingale residuals in each subgroup. Potential differences in the levels of clinical factors measured on a continuous or ranked scale in different patient subgroups were assessed using the Mann–Whitney test. In addition, correlations between such factors and measured cytokine levels were investigated using the Spearman test, as were levels of different cytokines in the same patients. Differences in the incidence of dichotomous clinical factors as a function of cytokine levels were investigated using logistic analysis, and differences in the incidence of such factors in patient subgroups with high versus low cytokine levels were examined using the Fisher exact test.

RESULTS

Control Group

The majority of normal volunteers (n = 48 to 62 participants, depending on the cytokine) had undetectable serum levels of TNF-α, IL-1α, IL-1β, IL-6, IL-8, and IL-10. In contrast, IL-1RA and VEGF levels were detectable in all normal individuals (Table 1).

Table 1. Comparison of Cytokine Levels between Patients with Pancreatic Carcinoma and Healthy Volunteers
CytokineMedian cytokine level (range) (pg/mL)P valueaHealthy volunteers with undetectable levels (%)Lower limit of assay sensitivity (pg/mL)
Patient groupControl group
  • TNF-α: tumor necrosis factor alpha; UND: undetectable; VEGF: vascular endothelial growth factor; IL-1RA: interleukin 1 (IL-1) receptor antagonist; IL: interleukin.

  • a

    Determined using the Mann–Whitney test.

TNF-αUND (< 4.4–12.7)UND (< 4.4–10.5)0.828894.4
VEGF117.0 (50.0–449.0)110 (52.0–464.0)0.69209.0
IL-1RA405.0 (123.0–3759.0)275.0 (65.0–1484.0)< 0.05022.0
IL1-1αUND (UND)UND (UND)0.991000.5
IL-1βUND (UND)UND (UND)0.991000.3
IL-63.4 (< 0.7–38.0)UND (< 0.7–4.3)< 0.0001520.7
IL-811.0 (< 10.0–3106.0)UND (< 10.0–33.0)< 0.00019310.0
IL-103.4 (< 3.0–141.0)UND (< 3.0–18.0)0.001803.0

Cytokine Levels in Patients with Pancreatic Carcinoma

In total, 51 patients with pancreatic carcinoma were studied. Their clinical characteristics are shown in Table 2. The following numbers of patients were studied for each cytokine: TNF-α (n = 48 patients), VEGF (n = 50 patients), IL-1RA (n = 51 patients), IL-6 (n = 50 patients), IL-8 (n = 51 patients), IL-10 (n = 50 patients), IL-1α (n = 51 patients), and IL-1β (n = 51 patients). (The number of patients from whom samples were assayed was dependent on the availability of serum.)

Table 2. Patient Characteristics (n = 51)
CharacteristicValue
  • ECOG: Eastern Cooperative Oncology Group; CA19-9: carbohydrate antigen 19-9; U: units.

  • a

    Normal range for CA19-9 levels is < 1.0–47.0 U/mL.

Male:female ratio30:21
Age (yrs) 
 Median65
 Range43–79
ECOG performance status 
 010
 137
 22
 32
Disease status 
 No evidence of disease4
 Resectable20
 Locally advanced15
 Metastatic12
CA19-9 (U/mL) (n = 48)a 
 Median217
 Range< 8–33,700
Median follow-up for survivors (mos)10
No. of deaths (%)63
Median survival (mos)10

Serum TNF-α and VEGF levels in patients with pancreatic carcinoma did not differ from those in normal individuals (Table 1). Most patients with pancreatic carcinoma (91%) had undetectable levels of TNF-α. Similarly, serum levels of IL-1α and IL-1β were undetectable in all controls and in all patients. In contrast, circulating levels of IL-1RA, IL-6, IL-8, and IL-10 all were elevated significantly compared with the levels in healthy volunteers (Table 1).

Correlation among Cytokine Levels

Elevation in one cytokine often correlated with elevation in others. The Spearman test demonstrated a correlation between IL-1RA and IL-8 (P < 0.001; Spearman ρ = 0.465), between IL-6 and IL-10 (P = 0.004; Spearman ρ = 0.403), and between IL-8 and IL-10 (P = 0.006; Spearman ρ = 0.385). Other correlations between cytokines did not reach statistical significance.

Correlation between Clinical Prognostic Features and Survival: Univariate Analysis

The following prognostic factors were examined in the univariate analysis: age (≤ 60 years vs. > 60 years); Zubrod performance status (< 2 vs. ≥ 2); weight loss (> 10% vs. ≤ 10%); tumor location (pancreatic head vs. pancreatic body/tail); and abdominal pain, appetite loss, hemoglobin, total bilirubin, lactate dehydrogenase (LDH), albumin, protime, transaminases, alkaline phosphatase, carcinoembryonic antigen (CEA), CA19-9, disease stage, resectability, superior mesentery artery/vein invasion, tumor size, and lymph node status. The following clinical prognostic factors were associated significantly with inferior survival: T4 tumors (extend beyond the pancreas into nearby arteries or veins; these tumors are unresectable), superior mesentery artery/vein invasion, unresectability, metastases, tumor in the pancreatic body/tail versus the pancreatic head (all with P < 0.001), and weight loss > 10% (measured over 2 months; P = 0.01; log-rank test).

Correlation between Serum Cytokines and Survival: Univariate Analysis

With a median follow-up of 10 months among the survivors, 32 of 51 patients died. The median survival was 10 months. Analysis for predictive value of cytokine level was performed on cytokines for which levels in patients with pancreatic carcinoma differed from the levels in normal volunteers (IL-1RA, IL-6, IL-8, and IL-10 but not TNF-α or VEGF). Patients who had low IL-1RA levels (< 159 pg/mL) had a decreased 1-year survival (0%; median survival, 5.4 months) compared with patients who had high IL-1RA levels (1-year survival, 43%; 95% CI, 27–58%; median survival, 10.8 months; log-rank P = 0.013) (Fig. 1). Patients who had elevated serum IL-6 levels (> 5.2 pg/mL) had decreased 1-year survival (0%; median survival, 4.6 months) compared with patients who had low IL-6 levels (1-year survival, 49%; 95% CI, 31–65%; median survival, 10.9 months; log-rank P = 0.002) (Fig. 2). Elevated serum IL-10 levels (> 9.8pg/mL) also were associated with worse survival (1-year survival, 0%; median survival, 5.5 months) compared with lower levels (1-year survival, 44%; 95% CI, 27–60%; median survival, 10.9 months; log-rank P = 0.006) (Fig. 3). Serum IL-8 levels were not associated with survival differences.

Figure 1.

Survival of patients with pancreatic carcinoma stratified according to interleukin-1 receptor antagonist (IL-1RA) levels. Distributions were estimated using the Kaplan–Meier method. Tick marks represent the time of last follow-up for patients who remained alive. Patients with low IL-1RA levels (< 159 pg/mL) had worse survival (P = 0.013).

Figure 2.

Survival of patients with pancreatic carcinoma stratified according to interleukin-6 (IL-6) levels. Distributions were estimated using the Kaplan–Meier method. Tick marks represent the time of last follow-up for patients who remained alive. Patients with high IL-6 levels (≥ 5.2 pg/mL) had worse survival (P = 0.002).

Figure 3.

Survival of patients with pancreatic carcinoma stratified according to interleukin-10 (IL-10) levels. Distributions were estimated using the Kaplan–Meier method. Tick marks represent the time of last follow-up for patients who remained alive. Patients with higher IL-10 levels (> 9.8 pg/mL) had worse survival (P = 0.002).

When serum IL-1RA, IL-6, and IL-10 levels were used to predict survival, 2 groups were formed according to the number of cytokines elevated (or decreased for IL-1RA). Patients with either 2 or 3 cytokines elevated (or decreased for IL-1RA) had decreased survival (1-year survival, 23%; median survival, 4.6 months) compared with patients who had, at most, 1 elevated cytokine or reduced IL-1RA levels (1-year survival, 47%, median survival, 10.9 months; log-rank P = 0.022) (Fig. 4).

Figure 4.

Survival stratified according to number of abnormal cytokine levels (interleukin-1 receptor antagonist [IL-1RA], interleukin-6 [IL-6], and IL-10). Because IL-1RA, IL-6, and IL-10, but not IL-8, predicted survival, the first 3 were examined. Elevated IL-6 levels (> 5.2 pg/mL), elevated IL-10 levels (> 9.8 pg/mL), and low IL-1RA levels (< 159 pg/mL) predicted poor survival (see Figs. 1–3) and were used in this analysis. Distributions were estimated using the Kaplan–Meier method. Tick marks represent the time of last follow-up for patients who remained alive. Patients who had abnormal levels of two or three cytokines had worse outcomes compared with patients who had abnormal levels of zero or one cytokine (P = 0.002).

Multivariate Analysis of Correlations between Prognostic Factors, including Serum Cytokines, and Overall Survival

In the multivariate analysis using the clinical factors that had univariate significance, the three adverse factors selected were tumor location in the body or tail of the pancreas, superior mesentery vessel invasion, and the presence of metastatic disease. The hazard ratios estimated for these risk factors are shown in Table 3. For the multivariate analysis that included only the cytokines as potential factors, the model selected only high IL-6 levels (> 5.2 pg/mL). The hazard ratio also is shown in Table 3. When the multivariate analysis included the cytokines as well as the clinical factors with univariate significance, the model selected T classification T4 (vs. T1–T3) and high IL-6 levels as independent factors. The hazard ratios are shown in Table 3.

Table 3. Prognostic Factors in Multivariate Analysis
Prognostic factorHRaSEP value95% CI
  • HR: hazard ratio; SE: standard error; 95% CI: 95% confidence interval; IL-6: interleukin 6.

  • a

    The HR at time t is the probability of death at time t conditional on survival up to time t. HR represents the estimated ratio of hazards between patient subgroups. For example, the data in the Clinical factors only rows indicate that the risk of near-term death for patients with tumors in the body/tail of the pancreas is approximately four times greater than the risk for patients with tumors in the pancreatic head.

  • b

    Only clinical factors that were significant in the univariate analysis were considered for inclusion in the multivariate analysis.

  • c

    Only cytokines were considered in the multivariate analysis.

  • d

    Both clinical factors and cytokines were considered in the multivariate analysis.

Clinical factors onlyb    
 Body/tail4.1592311.791540.0011.788002–9.675161
 Superior mesenteric vessel invasion2.9723361.2206550.0081.329026–6.647564
 Metastases2.4516320.99674180.0271.105067–5.439036
Cytokine levels onlyc    
 High IL-6 level (> 5.2 pg/mL)3.2165411.2963250.0041.459954–7.086618
Cytokine levels and clinical factorsd    
 T4 disease2.3797442.5591880.0012.117572–13.66737
 High IL-6 level (> 5.2 pg/mL)2.6791231.1036640.0171.194925–6.006822

Correlation between Serum Cytokines and Clinical Factors

For clinical factors that were measured on a continuous scale (hemoglobin, tumor size, CA19-9, alkaline phosphatase, CEA, albumin, alanine aminotransferase, aspartate aminotransferase, protime, LDH, total bilirubin, weight loss in pounds, and age) or on a ranked scale (T classification, Karnofsky performance status, and ECOG performance status), correlations with cytokine levels were investigated using the Spearman test. The data also were analyzed using the Mann–Whitney test to compare the distribution of values of each factor among patients in the high cytokine groups versus the low cytokine groups, which were defined using the following cut-off values: IL-1RA, < 159 versus ≥ 159 pg/mL; IL-6, > 5.2 versus ≤ 5.2 pg/mL; IL-8, > 14 versus ≤ 14 pg/mL; and IL-10, > 9.8 versus ≤ 9.8 pg/mL (Table 4).

Table 4. Significance Levels According to the Mann–Whitney Test
Clinical factorCytokine
IL-1RAIL-6IL-8IL-10
  • IL-1RA: interleukin 1 receptor antagonist; IL: interleukin; CA19-9: carbohydrate antigen 19-9; CEA: carcinoembryonic antigen; ALT: alanine aminotransferase; AST: aspartate aminotransferase; LDH: lactate dehydrogenase; KPS: Karnofsky performance status; ECOG PS: Eastern Cooperative Oncology Group performance status.

  • a

    Statistically significant.

  • b

    No patients in the low IL-1RA group had AST data available. Consequently, this analysis was not possible.

Hemoglobin0.004a0.0960.5000.655
Tumor size0.2870.0510.8480.969
CA19-90.7700.4180.0720.406
Alkaline phosphatase0.5170.0710.5940.615
CEA0.3200.7720.7650.143
Albumin0.6100.008a0.0780.028a
ALT0.3710.0590.1300.354
ASTb0.009*0.5160.736
Prothrombin time0.3680.7230.5060.119
LDH0.0980.3510.5280.436
Total bilirubin0.9160.1790.3700.313
Weight loss0.9170.1110.008a0.379
KPS0.9070.032a0.6810.017a
ECOG PS0.5910.033a0.7510.005a
T classification0.2760.0540.7950.034a
Age at sample draw0.7130.5170.6480.442

For dichotomous clinical factors, including tumor location (head vs. body/tail), T classification (T4 vs. T1–3, with T1–T3 tumors confined to the pancreas or spread to other tissue directly around pancreas and T4 tumors spread beyond the pancreas to the stomach, spleen, colon, and large blood vessels), weight loss > 10% (yes vs. no), loss of appetite (yes vs. no), Karnofsky performance status (≥ 90 vs. < 90), ECOG performance status (0 vs. > 0), and gender, logistic regression was used to investigate whether there was a difference in the incidence of each factor with increasing cytokine levels. In addition, these data were analyzed using the Fisher exact test to examine the distribution of data in a 2 × 2 table based on the cytokine cut-off levels described above.

All tests reported are two-tailed, with no corrections made for multiplicity of testing. The associations observed between clinical factors and cytokine levels are summarized below (Table 4).

Albumin

Albumin levels were correlated inversely with several cytokines (IL-6, IL-8, and IL-10) (Spearman correlation coefficient, − 0.420; P = 0.002). The Spearman correlation coefficient was − 0.420 (P = 0.002) for IL-6, − 0.289 (P = 0.040) for IL-8, and − 0.380 (P = 0.007) for IL-10.

Weight loss

The correlation between weight loss at the time of sample draw and the IL-8 level was significant (Spearman correlation coefficient, 0.438; P = 0.002). The median weight loss among patients with low IL-8 was 6.5 pounds (range, 0–55 pounds) compared with a median weight loss of 20 pounds (range, 0–80 pounds) among patients with high IL-8 (P = 0.008; Mann–Whitney test). There was also a trend toward a correlation between weight loss and IL-6 level (P = 0.11; Mann–Whitney test). This trend also was observed in the logistic regression analysis for IL-6 (P = 0.075), but not for IL-8 (P = 0.245), suggesting that there was a trend toward a consistent change in the incidence of weight loss with increasing IL-6 levels. There was also a correlation between weight loss and IL-10 levels (Spearman correlation coefficient, 0.311; P = 0.032). However, this was not significant according to the Mann–Whitney test (P = 0.379).

Performance Status

There was an inverse correlation between Karnofsky performance status1 and IL-6 levels (Spearman correlation coefficient, − 0.338; P = 0.016). The median Karnofsky performance status among patients with low IL-6 levels was 90 (range, 70–100) compared with a median Karnofsky performance status of 80 (range, 50–90) among patients with high IL-6 levels (P = 0.032; Mann–Whitney test).

There also was a trend toward an inverse correlation between Karnofsky performance status and IL-10 levels in the Spearman test (correlation coefficient, − 0.241; P = 0.091). The median Karnofsky performance status among patients with low IL-10 levels was 90 (range, 50–100), compared with a median Karnofsky performance status of 80 (range, 50–90) among patients with high IL-10 levels (P = 0.017; Mann–Whitney test).

Similarly, there was an inverse correlation between ECOG performance2 and IL-6 levels. The median ECOG performance status among patients with low IL-6 was 0 (range, 0–2), compared with a median ECOG performance status of 1 (range, 0–3) among patients with high IL-6 levels (P = 0.033; Mann–Whitney test).

T Classification

There was a correlation between tumor classification and IL-10. The median T classification among patients with low IL-10 was T3 (range, T0–T4), compared with a median T classification of T4 (range, T2–T4) among patients with high IL-10 levels (P = 0.034; Mann–Whitney test). There also was a correlation between IL-6 levels and T classification that approached statistical significance (P = 0.054).

DISCUSSION

Our patients with pancreatic carcinoma often had elevated circulating levels of IL-6, IL-10, IL-8, and IL-1RA (but not TNF-α, VEGF, IL-1β, or IL-1α) compared with the levels in healthy individuals. Furthermore, elevation in one cytokine often correlated with elevation in others. For instance, high IL-10 levels were correlated with high IL-8 and high IL-6 levels. These results are not surprising in light of the fact that many cytokines share transcriptional activators, such as neuronal factor κB,18 which is known to be up-regulated in pancreatic carcinoma.10 In addition, cytokines can induce each other. For instance, it is known that IL-6 induces IL-10 production by T cells.19 Elevations in IL-6, IL-10, and IL-8 levels were correlated with a poor performance status and/or weight loss (Table 4), whereas high IL-6 and IL-10 levels and low IL-1RA levels were correlated with a worse outcome (Figs. 1–3). In addition, patients with abnormal levels of 2 of the 3 cytokines (IL-6, IL-10, and IL-8) fared worse than patients with abnormal levels of 1 or 0 cytokines (Fig. 4). The source of the circulating cytokines in our patients is not clear, but it may be host immune cells reacting to the tumor or the tumor itself. With regard to the latter, Blanchard et al.20 reported production of IL-8 and IL-6 by pancreatic carcinoma cell lines, and we noted high levels of IL-1RA (but not IL-10) in several such cell lines (R.K., unpublished data).

The Role of IL-6

IL-6 is a pleiotropic cytokine that has been implicated in the pathogenesis of several lymphoproliferative disorders, including multiple myeloma, lymphoma, and Castleman disease.21–29 Serum IL-6 levels are increased in diffuse large cell lymphomas, are associated with adverse prognostic features, and are predictive of poor failure-free and overall survival in multivariate analysis.12, 13 Serum IL-6 levels also may be elevated and correlated with poor prognostic features and an inferior outcome in patients with Hodgkin disease, indolent non-Hodgkin lymphomas, renal cell carcinoma, prostate carcinoma, ovarian carcinoma, multiple myeloma, chronic lymphocytic leukemia, and colorectal carcinoma.2, 4–8, 14, 15, 30–34

Several observations suggest that IL-6 is involved in the manifestations of pancreatic carcinoma. Okada et al.35 found high IL-6 levels in patients with pancreatic carcinoma, and these levels were correlated with weight loss. Furthermore, Falconer and colleagues36 demonstrated that at least a component of cachexia, which is a hallmark of pancreatic carcinoma, is due to high resting energy expenditure and that the presence of an acute-phase response identifies a group of patients who are markedly hypermetabolic. These patients have high spontaneous IL-6 production from their peripheral blood mononuclear cells.36 Other investigators also have shown that the production of C-reactive protein (which reflects acute-phase response) by peripheral blood mononuclear cells correlates with IL-6 expression in patients with pancreatic carcinoma and that anti-IL-6 antibodies reduced C-reactive protein levels.37 In our study, there was also a trend, albeit without statistical significance, toward a correlation between weight loss and IL-6 levels. High IL-6 levels correlated significantly with poor performance status (Karnofsky or ECOG) and low albumin levels (Table 4). Finally, high IL-6 levels predicted a shorter survival in both univariate and multivariate analyses (Fig. 2, Table 3).

Preclinical and clinical investigations of IL-6 suggest that this molecule has properties that are consistent with a causative role in the phenotypic correlations noted above. For instance, IL-6 can induce cachexia and fever,38–41 and its presence is highly correlated with B symptoms (fever, weight loss) in patients with recurrent lymphomas.5 It also was demonstrated previously that IL-6 inhibits the hepatic synthesis of albumin, a possible cause of hypoalbuminemia.42 Furthermore, clinical studies in which recombinant human IL-6 has been administered to patients show that this molecule increases C-reactive protein levels and decreases serum albumin levels.41 Therefore, it is not surprising that serum IL-6 levels were correlated inversely with serum albumin levels in our patient population. The correlation between high IL-6 levels and poor performance status may be due to a combination of many of the properties discussed above.

The Role of IL-10

To our knowledge, IL-10 has not been examined previously in patients with pancreatic carcinoma. The IL-10 molecule is a multifunctional cytokine produced by Type 2 helper cells (Th2)43 as well as monocytes, macrophages, and B lymphocytes.44–46 It is highly homologous to an open reading frame of Epstein–Barr virus called BCRF1, and Epstein–Barr virus infection of B cells up-regulates IL-10 production. IL-10 production has strong immunosuppressive effects through the inhibition of Th1-type cytokines, including interferon γ and IL-2.43 In cell lines derived from B-cell lymphomas, it has been found that IL-10 serves as an autocrine growth factor.47–49 The serum IL-10 level is an important prognostic factor in lymphoproliferative disorders3, 14, 15, 50 and in certain solid tumors, including lung and colorectal carcinomas and melanoma.9, 51, 52 In our patients with pancreatic carcinoma, high levels of IL-10 were correlated with poor survival (Fig. 3), poor performance status, low albumin levels, and advanced disease stage (T classification) (Table 4).

The Role of IL-8

IL-8 is a chemokine that may stimulate tumor growth directly or through its proangiogenic effects.11 High IL-8 expression has been reported in various human malignancies, including acute myelogenous leukemia, B-cell chronic lymphocytic leukemia, brain tumors, Hodgkin disease, melanoma, mesothelioma, and ovarian, prostate, renal, breast, colon, cervical, and gastric carcinomas (reviewed in Xie, 200111). In our patients, IL-8 levels were high and were correlated with weight loss, but not with survival.

The Role of IL-1RA

IL-1 is an important mediator of inflammation and tissue damage in multiple organs, both in experimental animal models of disease and in human illness. The IL-1 family consists of two agonists (IL-1α and IL-1β), two receptors (biologically active IL-1RI and inert IL-1RII), and a specific receptor antagonist (IL-1RA). The balance between IL-1 and IL-1RA in local tissues plays an important role in the susceptibility to and severity of many diseases.53

The role of the IL-1/IL-1RA system has been invoked in a variety of human malignancies. For instance, in acute and chronic myeloid leukemia, excess IL-1 drives cell growth.16, 54 IL-1RA reduces hepatic metastases of melanoma cells but enhances the growth of hepatoma cells in animal models (reviewed in Arend, 200253). These differing activities may be due to the variable ability of IL-1 to stimulate or suppress growth, depending on the type of malignant cells.

In our patients, IL-1RA levels were high compared with the levels in the control group. However, low IL-1RA levels were associated with a poorer outcome (Fig. 1). It is noteworthy in this regard that we have found that pancreatic carcinoma cell lines (BxPC-3, Capan-1, and HS766-T) expressed high levels of IL-1RA (as well as intracellular, but not secreted, IL-1α and IL-1β) and that inhibition of IL-1RA with neutralizing antibody enhanced their growth (R.K., unpublished data). These results suggest that IL-1 serves as a growth factor for pancreatic carcinoma, but its effects are balanced by the presence of IL-1RA. If this is true, then these in vitro observations may explain why individuals with low levels of IL-1RA do worse, suggesting that the administration of IL-1RA to these patients warrants pursuit in a clinical trial.

Summary

Our data indicate that patients with pancreatic carcinoma express high levels of several cytokines—IL-6, IL-8, IL-10, and IL1RA—compared with healthy individuals. Furthermore, high levels of IL-6 and IL-10 and low levels of IL-1RA correlate with a poor prognosis. In a multivariate analysis of clinical factors and cytokines, T4 tumors and high IL-6 levels were selected as independent predictors for shorter survival. Furthermore, poor performance status and/or weight loss correlated with the presence of high levels of IL-6, IL-8, and/or IL-10. Studies currently are underway in a larger group of patients to delineate independent correlates better in multivariate analysis. Therapeutic strategies for antagonizing some of these cytokines (i.e., IL-6, IL-8, and/or IL-10) with the use of monoclonal antibodies or other inhibitors also merit investigation. With regard to the IL-1 system, clinical data demonstrating poorer survival for patients with low IL-1RA levels, together with preclinical (in vitro) work (R.K., unpublished data) showing that IL-1RA antagonizes the growth-promoting effects of IL-1 in pancreatic carcinoma, suggests that the administration of IL-1RA to patients with low levels warrants study. Such cytokine modulatory approaches may alleviate the profound cachexia and malaise associated with pancreatic carcinoma and also potentially may induce antitumor effects.

Ancillary