Intratumoral CD14⁺ Cells and Circulating CD14⁺HLA-DR^lo/neg Monocytes Correlate with Decreased Survival in Patients with Clear Cell Renal Cell Carcinoma

This article is featured in Highlights of This Issue, p. 4025

It is estimated that approximately 70,000 new cases and 13,000 deaths occur annually from cancer of the kidney and renal pelvis in the United States (1). Recent advances have focused on targeting receptor tyrosine kinases, angiogenesis, or mTOR pathways (2–4). Despite improvements in treatments, durable responses remain a challenge. Immunotherapy for renal cell carcinoma (RCC) has had mixed results. High-dose IL2 regimens have only been successful in a small percentage of patients. Other immune-modulating approaches like peptide and dendritic cell vaccines for RCC showed promising results, but additional clinical trials are needed to optimize treatment protocols to improve their efficacy (5–7). With the preliminary successes of checkpoint inhibitors to CTLA-4 and PD-1 (8, 9) in other cancers, there has been a renewed interest for adding immunotherapy to the treatment repertoire for RCC.

Tumor-mediated immune suppression remains a significant barrier to cancer treatment. RCCs can induce both local and systemic immune suppression via multiple mechanisms (reviewed by Bedke and Stenzl; ref. 10). However, it is unclear how immunosuppressive mechanisms counteract the components needed to generate an effective, multifactorial antitumor response. Thus, additional understanding of the immune status of cancer patients is critical for both improving the outcomes of a specific immunotherapy and identifying relevant biomarkers to guide the selection of patients most likely to benefit from specific immunotherapy strategies.

Recently, we used quantitative whole blood flow cytometry on 160 healthy volunteers and patients from four cancer types, including RCC, to identify immune profiles of cancer patients that were prognostic for overall survival (11). We took a comprehensive approach by measuring all of the major leukocyte populations, including several immunosuppressive cell types such as regulatory T cells and myeloid-derived suppressor cells (MDSC). We identified a dominant leukocyte population (CD14⁺ HLA-DR^lo/neg monocytes) within the immune profiles associated with decreased survival. In this study, we aimed to understand the source and consequence of CD14⁺HLA-DR^lo/neg monocytes in RCC patient cell populations.

Peripheral blood samples were collected from 25 patients with metastatic RCC before cytoreductive nephrectomy and 40 healthy volunteers. Three patients had papillary histology. The clinical characteristics of the RCC patients have been described elsewhere (11). All studies were performed under the approval of the Mayo Clinic Institutional Review Board. Informed consent was obtained from all patients. We also identified 375 patients treated with radical or partial nephrectomy for unilateral, sporadic, noncystic clear cell RCC between 2000 and 2002, from the Mayo Clinic Nephrectomy Registry. The clinical features studied included age, gender, and symptoms at presentation. Patients with a palpable flank or abdominal mass, discomfort, gross hematuria, acute onset varicocele, or the constitutional symptoms of rash, sweats, weight loss, fatigue, early satiety, or anorexia were considered symptomatic at presentation. The pathologic features studied included histologic subtype classified according to the Union Internationale Contre le Cancer, American Joint Committee on Cancer, and Heidelberg guidelines, the 2002 primary tumor classification, regional lymph node involvement, distant metastases, the 2002 TNM (tumor node metastasis) stage groupings, tumor size, nuclear grade, coagulative tumor necrosis, sarcomatoid differentiation, and SSIGN (Stage, SIze, Grade, Necrosis) score (12). To obtain these features, a pathologist (J.C. Cheville) reviewed the microscopic slides from all specimens without knowledge of patient outcome. Vital status for patients in the Nephrectomy Registry is updated each year. If a patient died in the previous year, a death certificate was ordered to determine the cause of death. A visit to our institution within 6 months of the date of death for metastatic RCC was good documentation that RCC was the cause of death. If the death certificate did not support this, the medical history was reviewed to determine the cause of death. If a death certificate could not be obtained, the cause of death was verified with the patient's family or local physician.

Quantitative whole blood flow cytometry and data analysis, including hierarchical clustering, for the patient and healthy volunteer samples were previously described (11). Additional principle component analysis of flow data was performed in Partek Genomics Suite 6.6 (Partek Inc.). Color dot plots of flow data were generated using the Kaluza flow analysis software (Beckman Coulter).

A formalin-fixed, paraffin-embedded block with representative tumor tissue was selected for each patient in the cohort. Staining for CD14 was conducted as follows. Tissue sections were deparaffinized in xylene and rehydrated in a graded series of ethanols. Antigen retrieval was performed by heating tissue sections in Target Retrieval solution pH 6 (Dako #S1699) to 121°C using a Digital Decloaking Chamber (Biocare Medical). Sections were cooled to 90°C and incubated for an additional 5 minutes before the Decloaking Chamber was opened. Sections were washed in running DH₂O for 5 minutes and incubated for 5 minutes in Wash Buffer (Dako #S3006) before being placed on the Autostainer Plus (Dako) for the following protocol. Sections were blocked for endogenous peroxidase for 5 minutes using Endogenous Blocking solution (Dako #S2001), washed twice in wash buffer, incubated 5 minutes in Background Reducer Sniper (Biocare Medical #BS966L), washed twice in wash buffer, and incubated for 60 minutes in purified rabbit anti-human CD14 antibody (Sigma PA #HPA001882) diluted 1:300 with DaVinci Green antibody diluent (Biocare Medical #PD900M). Sections were washed twice in wash buffer, incubated 15 minutes in the rabbit probe from the Mach 3 Rabbit HRP-Polymer Kit (Biocare Medical #M3R531L), washed twice in wash buffer, and incubated 15 minutes with the rabbit polymer from the Rabbit HRP-Polymer Kit. Sections were washed with wash buffer and visualized by incubating Betazoid DAB (Biocare Medical #BDB2004L) for 8 minutes. Sections were washed with DH₂O, counterstained with hematoxylin, dehydrated in ethanol, and cleared in xylene. Coverslips were mounted in permanent mounting media. All stained slides were reviewed and quantitated by a pathologist (J.C. Cheville) without knowledge of patient outcome. Intratumoral and (if possible) peritumoral CD14 expression was assessed as absent, focal, moderate, or marked.

CD14⁺ monocytes were purified from normal donor leukoreduction chambers by density gradient separation (13) and selection of CD14⁺ monocytes from peripheral blood mononuclear cells by immunomagnetic selection via the AutoMacs Cell Separator (Miltenyi Biotech). CD14⁺ cells (2 × 10⁶/mL) were plated into control medium (DMEM) alone, or incubated with 1 × 10⁶ cells/mL of RCC cell lines or normal renal cell epithelial cells. ACHN (CRL-1611, lot #57586991), 786-O (CRL-1932, lot #7643580), and Caki-1 (HTB-46, lot #5074655) were purchased from the American Type Culture Collection (ATCC) in 2009, and primary renal epithelial cells (PCS-400-012, lot #58488854) in 2010. RCC cell lines were cultured as per the ATCC instructions less than 6 months before the coculture experiments, and primary cells were cultured in Renal Cell Growth medium (Lonza) no more than 10 passages. ATCC uses short tandem repeats (SRT) profiling to characterize most cell lines. RCC cells were also cultured alone at 1 × 10⁶ cells/mL in DMEM/10% FBS. Cells and conditioned supernatants were harvested after 72 hours. For cocultures, monocytes were reselected with CD14 microbeads with the negative fraction containing RCC cells. Flow cytometry was performed to assess purity of cultures.

The Proteome Profiler Human Angiogenesis Array kits (R&D Systems; catalog # ARY007) were used to analyze supernatants and cell lysates according to the manufacturer's instructions. Equal volumes were used for supernatants and for lysates, and volumes were normalized to equal cell numbers. FGF2 ELISAs (R&D Systems) were performed via manufacturer's instructions.

RNA was isolated from cells via the Qiagen RNeasy Plus Kit (Qiagen) and made into cDNA with the Transcriptor First Strand cDNA Kit (Roche Diagnostics GmbH). Primers for FGF2, IL1b, GM-CSF, PTX3, and ARG-1 were designed with the Roche Universal Probe Library (UPL) system. PCR reactions were performed with the LightCycler Taqman Master Mix, ran and analyzed on the Roche Light Cycler System. Relative transcript levels were determined by assessing the crossing points (Cp) for each gene for each experiment. The relative transcript levels were first normalized to the reference gene GAPDH and calculated by the log₂ difference of the Cp (ΔCp) of the reference gene and the gene of interest. Gene expression changes between samples were calculated as the log₂ fold change by subtracting the average of the ΔCp from the control sample from the average ΔCp of the test sample (ΔΔCp) and their fold difference 2^−ΔΔCp.

Values between groups of data were tested for statistical significance using the Mann–Whitney nonparametric test for unpaired samples. Correlation analysis was performed using Pearson correlation using Prism, version 5.0 software (GraphPad Software). For the qPCR analysis, values were measure for statistical significance by using a t-test against a normalized value (1.0; representing 100%). Associations of CD14 expression with clinical and pathologic features were evaluated using χ² tests. Associations of CD14 expression with cancer-specific survival were illustrated using Kaplan–Meier curves. The magnitudes of the associations of CD14 expression with death from RCC were evaluated using Cox proportional hazards regression models and summarized with HRs and 95% confidence intervals (CI). The duration of follow-up was calculated from the date of nephrectomy to the date of death or last follow-up. Patients who died from causes other than RCC were censored at the date of death, whereas patients who were alive at the time of analysis were censored at the date of last follow-up. Statistical analyses were performed using the SAS software package (SAS Institute). All tests were two-sided, and P values <0.05 were considered statistically significant. The 3 × 3 Fisher exact test was used for the grouping of intratumoral CD14 staining with peripheral blood phenotypes. The significance level was set at probability of significance set at less than 0.05 with specific calculated P values provided when applicable.

By combining whole blood quantitative flow cytometry and bioinformatics analyses, we generated immune profiles from 160 healthy volunteers, cancer patients, and patients with or at risk of sepsis (11). The overall survival of patients with an abnormal immune profile was significantly shorter than those patients with an immune profile similar to healthy volunteers. Twenty-five patients with RCC were included in this analysis, 9 of which were clustered into abnormal immune profiles. Here, we further investigated the immunophenotypic changes in peripheral blood leukocytes of these patients. Because the immune profiles were generated from 10 immune phenotypes [granulocytes, monocytes, lymphocytes, T cells, B cells, natural killer (NK) cells, CD4⁺ T cells, CD4⁺CD25⁺CD127^lo regulatory T cells (Tregs), CD14⁺HLA-DR^lo/neg monocytes, and CD86⁺ monocytes], we wanted to identify the immune phenotypes in RCC patients that were most responsible for the variation and to determine potential relationships between seemingly unrelated cells. We performed a principal component analysis to identify the immune phenotypes with the largest variation in this dataset (Supplementary Fig. S1). Lymphocytes, CD14⁺HLA-DR^lo/neg monocytes, Tregs, and B cells were the four immune phenotypes that appeared to be responsible for most of the variability. In univariate analyses, 4 of the 10 immune phenotypes were different in RCC patients compared with healthy volunteers. B cells were lower (healthy volunteer: mean = 247 ± 131.4 (SD) cells/μL vs. RCC: 204 ± 209.6), whereas granulocytes (healthy volunteer: 4,784 ± 1,620 cells/μL vs. RCC: 7,383 ± 5,025), monocytes (healthy volunteer: 508.9 ± 145.9 cells/μL vs. RCC: 718.2 ± 229), and CD14⁺HLA-DR^lo/neg monocytes (healthy volunteer: 57.7 ± 36.6 cells/μL vs. RCC: 249.4 ± 178.6) were elevated in RCC patients (Fig. 1A). CD14⁺HLA-DR^lo/neg monocytes showed the greatest increase of any immune phenotype in RCC patients at nearly 5-fold above the mean of healthy volunteers. Correlative analyses identified expected and novel relationships (Fig. 1B, data not shown). For example, the levels of B cells correlated positively with CD4⁺ cells and granulocytes with CD14⁺HLA-DR^lo/neg monocytes in both healthy volunteers and RCC patients; however, Tregs were correlated to CD4⁺ T cells only in RCC patients. The magnitude of change in the levels of CD14⁺HLA-DR^lo/neg monocytes in RCC patients suggests that this immune phenotype is important in the pathology of RCC.

We (14–16) and others (17–20) have demonstrated that CD14⁺HLA-DR^lo/neg monocytes are major mediators of tumor-induced immunosuppression. Other myeloid cells/MDSCs have been implicated in RCC-induced immunosuppression, including CD11b⁺CD14⁻CD15⁺ cells (21), lineage-negative (LIN⁻)CD33⁺HLA-DR⁻CD11b⁺ (22), CD15⁺CD11b⁺CD14⁻CD66b⁺CD33⁺ (23), CD33⁺HLA-DR⁻, and CD15⁺CD14⁻ cells (24). Because these reports measured cell populations from density-gradient–purified mononuclear cells, we wanted to assess the levels of these cells in fresh unfractionated whole blood. Because there is considerable diversity in the use of phenotypic markers for human MDSCs (25), we narrowed the characterization of these cells to CD14⁺HLA-DR^lo/neg monocytes, immature LIN⁻CD33⁺HLA-DR⁻ cells, and CD33⁺HLA-DR⁻ cells (gating strategies outlined in Supplementary Fig. S2). Figure 2A shows the forward and side scatter properties of each of these cell types in an RCC patient to demonstrate the forward and side scatter positioning of these cells relative to each other. LIN⁻CD33⁺HLA-DR⁻ cells reside mainly in the lymphocyte compartment (low side scatter and forward scatter). LIN⁻CD33⁺HLA-DR⁻ cells and CD33⁺HLA-DR⁻ granulocytes are distinct populations (26). CD14⁺HLA-DR^lo/neg cells have typical forward/side scatter properties of monocytes and should not be considered MDSCs as per convention, but are really monocytes characterized by the single-marker CD14 that has lost HLA-DR expression. As monocytes can be further subgrouped into classical (CD14⁺CD16⁻), intermediate (CD14⁺CD16⁺), and nonclassical (CD14^loCD16⁺; ref. 27), we measured the distribution of these subsets in RCC patients and their corresponding HLA-DR expression. The distribution of the subsets was not different in RCC patients compared with healthy volunteers, and HLA-DR was significantly lower in each of the subsets in RCC patients (Supplementary Fig. S3). Although there was some overlap with monocytes, CD33⁺HLA-DR⁻ cells almost entirely resided in the high forward/side scatter compartment typified by granulocytes. Not unexpectedly, CD15 and CD66b also identified granulocytes. Interestingly however, as CD33⁺HLA-DR⁻CD15⁺CD11b⁺CD14⁻CD66b⁺ granulocytes have been shown to copurify with mononuclear cells in RCC patients (21, 22, 24, 28), we rarely observed CD15⁺CD66b⁺ cells in the mononuclear “fraction” in staining of whole blood samples. We did not observe an increase in the amount of LIN⁻CD33⁺HLA-DR⁻ cells or a difference of CD66b expression on granulocytes in RCC patients versus healthy volunteers (Fig. 2B).

The assessment of peripheral blood immune phenotypes in fresh unmanipulated cells allows us to build a complete picture of the differences between the typical peripheral blood composition (both in quantity and composition) of healthy volunteers and that of RCC patients. Figure 2C shows the distribution of immune markers from the mean of 40 healthy volunteers and 25 RCC patients. The leukocyte compartment of RCC patients is enlarged by about a third over healthy volunteers with granulocytes and CD14⁺HLA-DR^lo/neg monocytes significantly expanding in this population.

Monocytes are capable of migrating from blood to tissues. Therefore, we investigated if the increase in the periphery of CD14⁺HLA-DR^lo/neg cells was associated with the presence of monocytes in the tumor. We examined the associations of tumor CD14 expression with clinical and pathologic features and patient outcome using a large cohort (n = 375) of RCC patients. A summary of clinical and pathologic features for the 375 ccRCC patients studied is shown in Table 1. Mean age at surgery was 63 years (median, 63; range, 26–87), mean tumor size was 6.8 cm (median, 5.8; range, 1.0–22.0), and mean SSIGN score was 3.9 (median, 3; range, 0–15). Tumors from 375 RCC patients were stained by immunohistochemistry for CD14 expression. These sections were evaluated and divided into four groups classified as high (marked CD14 expression), moderate (intermediate CD14 expression), and low (focal) or none (absent CD14 expression). Intratumoral, peritumoral, and maximal CD14 expressions between the two are summarized in Supplementary Table S1. Associations of maximal CD14 expression with clinical and pathologic features are summarized in Table 2. The association of maximal intratumoral and peritumoral CD14 expression with cancer-specific survival is illustrated in Fig. 3A. The magnitudes of the associations of CD14 expression with death from RCC are summarized in Supplementary Table S2. At last follow-up 159 patients had died, including 100 who died from RCC at a mean of 2.3 years following surgery (median, 1.7; range, 0–9). Among the 216 patients who were still alive at last follow-up, the mean duration of follow-up was 7.1 years (median, 7.3; range, 0–10); only 5 (2.3%) patients had fewer than 2 years of follow-up. Estimated cancer-specific survival rates (95% CI, number still at risk) at 1, 3, 5, and 7 years following surgery were 91.0% (88.1–94.0; 328), 79.0% (74.9–83.4; 268), 75.3% (70.9–80.0; 232), and 71.9% (67.3–77.0; 136), respectively. By univariate analysis, patients whose tumors contained moderate CD14 expression were nearly 5 times more likely to die from RCC compared with patients whose tumors contained absent/focal CD14 expression (HR, 4.79; P < 0.001). Patients whose tumors contained marked CD14 expression were over 11 times more likely to die from RCC compared with patients whose tumors contained absent/focal CD14 expression (HR, 11.51; P < 0.001). These associations were also evaluated in a multivariable setting adjusting for the SSIGN score. After accounting for the association between SSIGN score and death from RCC, the HRs for the associations of moderate and marked CD14 expression with death from RCC were 2.87 (P < 0.001) and 3.03 (P < 0.001), respectively. After further adjustment for age at surgery, gender, symptoms at presentation, sarcomatoid differentiation, and the SSIGN score, the HRs were 2.53 (95% CI, 1.45–4.42; P = 0.001) and 2.69 (1.44–5.01; P = 0.002), respectively.

We confirmed that increased fraction of CD14⁺HLA-DR^lo/neg cells in blood was correlated to intratumoral CD14 expression by measuring both in 23 RCC patients and proceeded to classify them as high, moderate, or low/none for intratumoral expression. We performed CD14/CD15 dual labeling immunohistochemistry on additional RCC tumors (n = 6) and found that the mean fluorescent intensity of HLA-DR expression on peripheral blood CD14⁺ monocytes in RCC patients inversely correlated with the intensity of CD14 staining in their corresponding tumors (Fig. 3B). We also confirmed that the CD14⁺ cells identified in RCC tumor tissue lack expression of granulocyte cell markers (CD15), indicating that these are not granulocytes with aberrant CD14 expression (Fig. 3C). Taken together, these results confirmed positive correlation between increased intratumoral CD14 monocytes and increased CD14⁺HLA-DR^lo/neg monocytes in systemic circulation, suggesting that CD14⁺HLA-DR^lo/neg monocytes can be a surrogate marker for intratumoral CD14 and a predictor of decreased survival.

Knowing that patients with high intratumoral CD14 have decreased survival, we developed an in vitro model to examine monocyte/tumor cell interaction. We isolated CD14⁺ monocytes from healthy volunteers and coincubated them for 72 hours with the renal carcinoma cell line ACHN. HLA-DR expression was reduced by more than 75% upon coculture with ACHN cells versus monocytes in media alone (Fig. 4A, left graph; experimental replicates = 7; P = 0.026). This result was specific for tumor cells as HLA-DR expression was not downregulated during coculture with normal primary renal epithelial cells. We also observed loss of HLA-DR expression when monocytes were cocultured with another RCC cell line, 786-0 (Supplementary Fig. S4). Tumor cells promoted survival of monocytes in these coculture experiments as recovery of the CD14⁺ monocytes was 31% compared with 5% for monocytes alone (P = 0.004; experimental replicates = 5; Fig. 4A, right graph).

Because myeloid cells have been implicated in the process of angiogenesis (reviewed by Schmid and colleagues; ref. 29), we hypothesized that the interaction between the tumor and monocytes highlighted by the loss of DR may also represent the increase of an angiogenesis program by monocytes. We used a proteomics approach to test supernatants and cell lysates from CD14⁺ and ACHN cells following coculture for the simultaneous expression of 55 proteins involved in angiogenesis (Fig. 4B). We detected specific increases in a number of secreted proteins in the supernatant following monocyte/tumor cell coculture including IL1β, FGF2, and CSF2/GM-CSF. We also found other angiogenic factors, PTX3 and VEGF, that were secreted by monocytes and/or tumor cells but were not changed upon coculture (Fig. 4B). By examining the cell lysates, we could further determine the cell types (CD14⁺ or ACHN cells) that were producing the proteins detected in the supernatant. For example, IL1β and FGF2 were specifically induced in monocytes when cocultured with tumor cells, whereas GM-CSF and PTX3 were induced in tumor cells when cocultured with monocytes. Again, there were other angiogenic factors, like VEGF, that were detected in the cell lysates but did not change upon coculture. These proteins were not induced in cocultures of monocytes with primary renal epithelial cells (Supplementary Fig. S4). To further verify these changes, we repeated these experiments with up to 6 other normal donor monocyte preparations but assessed changes by quantitative RT-PCR (Fig. 4C). The RNA transcript levels of GM-CSF and PTX3 were significantly induced in tumor cells, whereas IL1β and FGF2 were induced in monocytes. Interestingly, FGF2 levels were not detectable in monocytes cultured in control media, but were greatly induced upon coculture with tumor cells. Since FGF2 expression in tumors has been implicated as a prognostic factor in RCC (30), we measured the plasma levels of FGF2 in our cohort of RCC patients. Although we failed to see a significant difference between healthy volunteer and RCC patient plasma FGF2 levels, we found that FGF2 levels inversely correlated with the expression of HLA-DR on CD14⁺ monocytes (Fig. 4D). Taken together, these data demonstrate that monocytes and tumor cells are undergoing reciprocity of induction where monocytes are changing the tumor microenvironment and the tumor environment is changing the monocytes leading to monocytic survival, increased immune suppression, and increased angiogenesis.

The relationship between the power of the immune system to eliminate tumors and the tumors' ability to stifle that power has been a source of intense investigation. Most models of tumor/immune system interaction consist of the tumor actively and unilaterally inhibiting an antitumor response through the release of tumor-suppressing growth factors, or by specific antigen escape. The data we have presented here provide an alternative and powerful model. Our data suggest that tumors interact with monocytes resulting in altered monocytes with both immunosuppressive and immune-independent functional characteristics. This model also provides a framework to describe systemic immune suppression as the monocytes have the capacity to leave the tumor environment and move to the periphery.

CD14⁺HLA-DR^lo/neg monocytes have known immunosuppressive properties via multiple mechanisms, including effector T-cell inhibition, decreased antigen presentation, and defective dendritic cell maturation (14, 15, 18, 19, 31). The degree by which these cells were increased in RCC patients suggests a dominant role in mediating tumor progression. Other myeloid cells, including CD33⁺HLA-DR⁻CD15⁺CD66b⁺ granulocytes, were also elevated in RCC patients. This observation is generally consistent with previous findings that granulocytic MDSCs are elevated in RCC patients in the mononuclear fraction of density gradient centrifugation–purified cells (21–24). However, these results are not directly comparable because we performed our flow assays on unfractionated whole blood. In our experience, flow cytometry performed on whole blood is more amenable to standardization and direct inter-laboratory comparisons than flow cytometry performed on fractionated samples (32). The correlation between the levels of CD14⁺HLA-DR^lo/neg monocytes and granulocytes is quite interesting. We do not have any data suggesting that these cells influence each other; however, it is possible that tumor-derived factors that promote the survival, mobilization, and/or proliferation of monocytes also influence granulocytes. Our in vitro monocyte/coculture system supports this observation with increased GM-CSF induced during cocultures. It is likely that this observation is an indication of increased program of survival for certain classes of leukocytes.

We found considerable variability in the amount of CD14 staining in RCC samples but found a strong adverse correlation to the amount of intratumoral CD14 staining and survival. Sections of the highest grade from the tumor blocks were selected for staining and were generally consistent. Multiple blocks were not assessed; therefore, little is known regarding whole tumor infiltration of monocytes. In addition, it is not known if the tumor heterogeneity or specific oncogene expression influences the recruitment of monocytes differently at primary sites as opposed to metastatic sites.

Our in vitro coculture model of CD14⁺ monocytes and RCC cell lines demonstrated that the interaction between monocytes and tumor cells leads to the upregulation of factors that favor tumor progression. Monocytes induced GM-CSF production from tumor cells, a known potent survival factor for monocytes. Tumor cells caused the downregulation of HLA-DR on monocytes and induced expression of the angiogenic factors FGF2 and IL1β (RNA transcript and protein levels). IL1β is a potent inducer of angiogenesis and has been reported to contribute to tumor invasiveness, metastasis, and immune suppression (33). In the RCC setting, Chittezhath and colleagues recently identified the IL1-IL1R pathway via transcriptomic and molecular profiling as an important mechanism for the tumor-promoting role of monocytes (34). They found that peripheral blood monocytes from RCC patients displayed elevated expression of genes involved in angiogenesis and invasion. Blockade of the IL-1IL1R pathway in tumor/monocyte cocultures inhibited the induction of these genes. Through their analysis of tumor-gene expression data from 34 RCC patients, they found that IL1β expression correlated with myelomonocytic markers, including CD14. We identified that IL1β protein levels were induced in monocytes cocultured and assessed CD14 accumulation in tumors by immunohistochemistry. Taken together, our results combined with those of Chittezhath and colleagues independently confirm the role of tumor-promoting effects of monocytes and their impact on patient survival. FGF2 has been shown to be a prognostic factor in RCC (35). The high levels of intratumoral CD14⁺ cells combined with the observation that FGF2 was predominantly induced in monocytes may provide a mechanistic explanation to the failure of anti-VEGF therapies in some RCC patients. In many respects, the actions of tumor-infiltrating monocytes mimic the natural phenomena of tissue repair. Circulating monocytes are very efficient at homing to sites of injury, whereby they initiate proinflammatory responses as part of the innate immune response. The monocytes/macrophages then switch to an antiinflammatory response that is required for efficient tissue regeneration. Angiogenesis and fibroblast recruitment, mediated in part by FGF2, are additional functions that these cells provide. This process certainly demonstrates the remarkable plasticity of the functions of monocytes (36, 37).

Our data additionally suggest that CD14⁺HLA-DR^lo/neg monocytes are both an important therapeutic target and a strong prognostic biomarker for outcome in RCC patients. Perhaps by targeting these cells through elimination or blocking their migration to tumors, the efficacy of conventional therapies will likely improve. For example, we have recently shown that in a similar in vitro model with non-Hodgkin lymphomas, the presence of monocytes protected the tumor from direct killing with doxorubicin by altering the expression patterns of proapoptotic and antiapoptotic proteins (38). This study identifies an additional mechanism, protection from chemotherapy-induced cytotoxicity, in which the interaction of monocytes and tumor cells reduces the efficacy of anticancer treatments.

Because CD14⁺HLA-DR^lo/neg monocytes have multiple immunosuppressive functions, they will likely negatively impact immunotherapeutic approaches. There is emerging evidence that this phenomenon is already happening in RCC patients. These cells have been shown to adversely affect survival of RCC patients treated with a multipeptide cancer vaccine (6) and impede the differentiation into fully mature dendritic cells (31). Circulating monocytes and granulocytes have previously been independently associated as prognostic factors for poor clinical responses to IL2 immunotherapy (39, 40). Therefore, the synergistic interaction between tumor and monocytes may explain the ineffectiveness of many immune therapies by shutting down both local and systemic antitumor immunity. Taken together, our results reveal why the presence of CD14⁺ cells may reflect the aggressiveness of RCC tumors and the failure of these tumors to respond to current treatments.

E.D. Kwon has ownership interest (including patents) in Bristol-Myers Squibb and MedImmune. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M.P. Gustafson, Y. Lin, D. Warad, M.K. Tollefson, P.L. Crispen, B.C. Leibovich, E.D. Kwon, A.B. Dietz

Development of methodology: M.P. Gustafson, Y. Lin, D. Warad, M.K. Tollefson, P.A. Bulur, E.D. Kwon

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.P. Gustafson, Y. Lin, J.S. Bleeker, D. Warad, M.K. Tollefson, P.L. Crispen, P.A. Bulur, S.M. Harrington, B.C. Leibovich, J.C. Cheville, E.D. Kwon, A.B. Dietz

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.P. Gustafson, Y. Lin, J.S. Bleeker, D. Warad, R.R. Laborde, J.C. Cheville, E.D. Kwon, A.B. Dietz

Writing, review, and/or revision of the manuscript: M.P. Gustafson, Y. Lin, J.S. Bleeker, D. Warad, M.K. Tollefson, P.L. Crispen, P.A. Bulur, R.R. Laborde, D.A. Gastineau, B.C. Leibovich, J.C. Cheville, E.D. Kwon, A.B. Dietz

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.S. Bleeker, J.C. Cheville

Study supervision: Y. Lin, A.B. Dietz

The authors thank Debra Head for the assistance with the collection of samples.

This study was funded in part by grants from the Department of Laboratory Medicine and Pathology, Mayo Clinic.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.