Prognostic impact of in vivo soluble cell adhesion molecules in metastatic renal cell carcinoma

The purpose of the study was to determine prognostic significance of pretreatment serum levels of different molecules involved in cell to cell interactions along with other clinical parameters in patients with metastatic renal cell carcinoma. sICAM-1, sVCAM-1 and sELAM-1 serum levels were determined by ELISA assays in sera from 99 patients with histologically confirmed progressive metastatic renal cell carcinoma prior to initiation of systemic therapy. Kaplan–Meier survival analysis, log-rank statistics and two-proportional Cox regression analyses were employed to identify risk factors and to demonstrate statistical independence. In univariate analyses, the following pretreatment risk factors could be identified: serum sICAM-1 level > 360 ng ml−1, erythrocyte sedimentation rate > 70 mm h−1, serum C-reactive protein level > 8 mg l−1, serum lactic dehydrogenase level > 240 U/ l and neutrophil count > 6000 μl−;1. Multivariate analyses demonstrated statistical independence for serum sICAM-1 level, erythrocyte sedimentation rate (ESR) and serum C-reactive protein (CRP) level as pretreatment predictors of overall patient survival. The prognostic significance of sICAM-1 might indicate a role of this molecule for tumour progression, potentially in association with the abrogation of anti-tumour immune responses. The possibility of defining a pretreatment risk model based on sICAM-1 level, ESR and CRP also warrants further investigation, with regard to a possible linkage between acute phase proteins and sICAM-1 levels. © 1999 Cancer Research Campaign

Renal cell carcinoma is an indolent disease diagnosed with distant metastases in a significant proportion of patients. It is usually resistant to chemotherapy; a significant improvement in treatment could be achieved by introduction of cytokines into treatment regimens (Atzpodien et al, 1990). The molecular mechanisms of metastases formation are largely unknown.
Cell adhesion molecules are cell surface receptors involved in leucocyte homing, adhesion and migration. Although these proteins are usually membrane-bound, soluble forms exist and are generated by shedding of the extracellular portions from the cell surface (Rothlein et al, 1991).
Intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule 1 (VCAM-1) belong to the immunoglobulin superfamily and are expressed on a wide variety of cells including endothelia, antigen-presenting cells and B-cells (Roit et al, 1996;van den Stolpe and van der Saag, 1996). The ICAM-1-ligand, LFA-1 (Marlinn and Springer, 1987;Roit et al, 1996), is expressed on most leukocytes, including T-cells and NK-cells; the VCAM-1 ligand, VLA-4, is primarily expressed on lymphocytes. These ligands belong to the integrin superfamily (Roit et al, 1996). Eselectin (ELAM-1) is expressed on endothelium and its N-terminal lectin domain binds to carbohydrate groups associated with a variety of proteins of leukocyte cell surfaces, like the CD15 associated Lewis-X carbohydrate group (Roit et al, 1996).
Prior studies have demonstrated significant differences in expression of VLA-integrins between normal kidney tissue and low/high grade renal cell carcinoma specimens (Anastassiou et al, 1995). Another example for a possible relevance of cell adhesion molecule expression in the course of malignant disease is the fact that endothelial and leucocyte expression of PECAM-1 has shown impact on leucocyte infiltration of the tumour tissue and on patient survival (Anastassiou et al, 1995). However, it remains unclear whether serum levels of the soluble isoforms are of prognostic value. To test this hypothesis in advanced renal cell carcinoma, we examined sera from 99 patients for pretreatment serum levels of sICAM-1, sVCAM-1 and sELAM-1.

Patient selection and treatment
Between 1989 and 1996, 99 patients were assigned to receive either immunotherapy or combined chemo-/immunotherapy according to the following schedule: Interferon-α subcutaneously, 5 MIU m -2 on day 1 of weeks 1 and 4, on days 1, 3, 5 of weeks 2 and 3; 10 MIU m -2 on days 1, 3, 5 of weeks 5-8. Interleukin-2 subcutaneously, 20 MIU m 2 , on days 3, 4, 5 of weeks 1 and 4, and 5 MIU m -2 on days 1, 3, 5 of weeks 2 and 3. Additionally, for patients in the combined chemo-/ immunotherapy group, 5-fluorouracil was administered on day 1 of weeks 5-8 as a 1000 mg m -2 intravenous infusion. Concomitant medication was given as needed to control adverse effects of immunotherapy.
All patients had histologically (by nephrectomy) confirmed metastatic renal cell carcinoma in progressive state and a Karnofsky performance status of ≥ 70%. Written informed consent was obtained from each patient prior to administration of any Prognostic impact of in vivo soluble cell adhesion molecules in metastatic renal cell carcinoma study medication; response and survival were assessed according to World Health Organization (WHO) criteria. Table 1 shows the patient characteristics.

Sample collection and soluble cell adhesion molecule serum level measurements
Serum samples were obtained from each patient prior to initiation of therapy and stored at -70°C until analysis. Soluble cell adhesion molecule serum levels were determined by enzyme-linked immunosorbent assay (ELISA) as follows: sVCAM-1, Medgenix, Ratingen, Germany; sICAM-1 and sELAM-1, Quantikine, R&D, Wiesbaden, Germany. All analyses were performed in duplicate, and sera were diluted to reach a concentration within the standard curve of the respective assay, when necessary.

Statistical analysis
The statistical end point was overall survival from treatment start. Survival analysis and univariate risk factor analysis were performed according to the method of Kaplan and Meier, and multvariate risk factor analysis according to the two-proportional Cox regression analysis with forward selection of variables. Impact of potential risk factors on survival was calculated with log-rank statistics. Clinical parameters and soluble cell adhesion molecules were tested as dichotomized variables, defining the cutoff value as the value that best discriminates between good and poor overall survival.

Overall survival and univariate risk factor analysis
Median survival of all patients entered into the study was 23 months, with 35 patients being alive at last follow-up. Survival ranged from 1 to 80+ months. As shown in Table 2, of all cell adhesion molecule serum levels assayed, only sICAM-1 showed a predictive significance, with a median survival of 27 months in patients with pretreatment sICAM-1 serum levels below 360 ng ml -1 as opposed to a median survival of 8 months in patients with elevated pretreatment sICAM-1 serum levels (P = 0.0011).

Multivariate risk factor analysis
Among all pretreatment risk factors identified by univariate analysis, a multivariate Cox regression analysis was performed to identify statistically independent predictors of survival. Elevated serum sICAM-1 level, ESR and serum CRP were found to be significant and statistically independent (Table 3). Of these predictors, sICAM-1 level showed a particular power to discriminate between good risk and poor risk patients (P = 0.0092).

DISCUSSION
The present study examined prognostic significance of pretreatment soluble cell adhesion molecule serum levels and other clinical laboratory parameters in unselected metastatic renal cell carcinoma patients. Serum sICAM-1 levels, ESR and serum CRP levels could be identified as highly significant and statistically independent predictors of overall survival. While ESR and CRP are rather unspecific indicators of inflammatory activity, sICAM-1 expression needs closer attention.
ICAM-1 has been reported to be aberrantly expressed on cells from a variety of human malignancies: myeloid leukaemias (Maio et al, 1990), B-cell lymphoproliferative disorders (Maio et al, 1990), neuroblastoma (Foreman et al, 1993), non-small-cell lung cancer (NSCLC) (Passlick et al, 1994), prostate carcinoma (Wolff et al, 1995) and head and neck squamous cell carcinoma (Vora et al, 1997). While its expression on tumour cells is generally linked to poor prognosis, in NSCLC it appears to be rather expressed by well-differentiated and localized tumours (Passlick et al, 1994). Particular attention has been directed to ICAM-1 expression in malignant melanoma, where it can be demonstrated in the majority of cases. Increased expression in metastases as compared to the primary tumour has been demonstrated, suggesting a role in the formation of metastases (Natali et al, 1990). Furthermore, ICAM-1-positive primary melanomas tend to recur after surgical excision (Natali et al, 1997). Surprisingly, these features seemed to be unique to melanoma and could not be reproduced in a variety of malignancies from different embryonal origins (Natali et al, 1990) and in renal cell carcinoma (Heicapell et al, 1994).
Not only the expression of ICAM-1 on tumour cells has demonstrated prognostic significance, but also the level of the soluble form as measured in patients' sera. This has been demonstrated for acute lymphoblastic leukaemia (Hatzistilianou et al, 1997), malignant melanoma (Schadendorf et al, 1996), non-Hodgkins lymphoma (Christiansen et al, 1996), Hodgkin's disease (Christiansen et al, 1995) and hepatocellular carcinoma (Shimizu et al, 1995), amongst others.
We demonstrated the prognostic significance and independence not only for pretreatment sICAM-1 level, but also for ESR and CRP. Notably, elevated serum sICAM-1 by itself was capable of identifying patients at highest risk, i.e. with a 5-year survival probability of 0%. Although we could demonstrate statistical independence of sICAM-1 levels and each of the above acute phase reactants, earlier studies demonstrated a possible correlation between acute phase reactants and sICAM-1 levels (Reinhardt et al, 1996) or between ESR and serum CRP level (Ljungberg et al, 1990).
Since earlier studies have shown soluble ICAM-1 to be active in influencing the cell-surface ICAM-1/LFA-1 interaction (Welder et al, 1993;Meyer et al, 1995), another issue is that of a potential role of sICAM-1 in formation of metastases and tumour progression, leading to impaired prognosis.
It might be hypothesized that ICAM-1 on circulating tumour cells results in attachment to the capillary endothelium via specific ligands, which in this case still need to be defined. This attachment of cells might ultimately lead to formation of metastases. Since the soluble form of ICAM-1 is produced by shedding of the extracellular portion of the membrane-bound form, it seems likely that high sICAM-1 levels can be found in patients whose tumours express large amounts of surface ICAM-1. However, in that scenario, high sICAM-1 serum levels would occupy the endothelial receptors, preventing tumour cells from adherence and metastases formation.
ICAM-1 is not only involved in cell homing and migration, but also in various cell-to-cell interactions with immunological Months from treatment start sICAM-1 ≥ 360 ng ml -1 sICAM-1< 360 ng ml -1 P =0.0011 Probability of survival Figure 1 Pretreatment sICAM-1 serum levels and survival in 99 patients with metastatic renal cell carcinoma. Survival curves (Kaplan-Meier estimate) of patients with either low (< 360 ng ml -1 ) or elevated (≥ 360 ng ml) pretreatment serum levels of sICAM-1. P-value was determined by log-rank test. Tick marks represent patients for whom data were censored (n = 34 and n = 1, for patients with low or elevated pretreatment serum sICAM-1 levels, respectively) significance, such as T-cell-mediated cytotoxicity (Makgoba et al, 1998), NK-cell-mediated cytotoxicity (Chong et al, 1994), B-Tcell interaction (Boyl et al, 1998;Amblard et al, 1994), T-cell adhesion (Sung et al, 1992), or B-cell activation (Moy and Brian, 1992). Hence, another scenario seems possible: by binding to its ligands on immune effector cells, like LFA-1 on T-cells, the soluble form could effectively inhibit different anti-tumour immune effects like effector mechanisms of cytotoxic T-cell or NK-cells or antigen-dependent respectively T-cell-dependent Bcell activation; this could be part of an immune escape mechanism of tumour cells, and could possibly explain the coincidence of high serum sICAM-1 levels and impaired prognosis. Some of these effects have already been demonstrated in vitro (Fecondo et al, 1991;Becker et al, 1993), underlining the possibility of a specific action of ICAM-1 in the immune escape of malignant cells. Although further investigation is needed, this could clearly be a target for future therapeutic trials.