Granulocyte-macrophage colony- stimulating factor (GMCSF, sargramostim(Drug information on sargramostim) [Leukine]) has been widely used to restore bone marrow function after dose-intensive chemotherapy. In various clinical scenarios, both granulocyte colony-stimulating factor (G-CSF, filgrastim(Drug information on filgrastim) [Neupogen]) and GM-CSF have been found to effectively prevent chemotherapy- induced neutropenia. Compared to G-CSF, however, the clinical use of GM-CSF as an adjunct to antineoplastic chemotherapy has been compromised by the activation of eosinophils, which may well have been the main reason for a considerable incidence of treatment-related type I hypersensitivity-like reactions. More recently, however, GM-CSF has been recognized as a powerful cytokine playing a major role in the generation of dendritic cells from mononuclear precursors.[1,2] Dendritic cells are the most important antigen-presenting cells able to establish an intrinsic cellular immunologic response against autologous tumor cells. GM-CSF has therefore been used as part of cytokine cocktails for ex vivo expansion of dendritic cells from peripheral blood mononuclear cells. Local or systemic administration of GM-CSF alone or in combination with interferons and interleukins also increases the dendritic cell fraction in vivo.[3,4] In metastatic melanoma, GM-CSFbased cytokine combinations have now become a routinely used immunotherapy.[ 3,5] In contrast, singleagent GM-CSF treatment is far less well established at present. However, adjuvant low-dose prolonged administration GM-CSF has significantly improved the median survival of patients with stage III and IV melanoma in a single institution trial. The local and systemic toxicity of this regimen was remarkably low. Moreover, peritumoral GM-CSF injection has prolonged the survival of individuals with metastatic melanoma, an effect that cannot be explained by local effects alone. At present, data on systemic GMCSF treatment in epithelial tumors other than prostate cancer are still lacking.[7,8] Nonetheless, this approach may be particularly interesting in gynecologic tumors since GM-CSF modulated activation of transforming growth factor-beta (TGF-beta) receptors has been shown to be a negative autocrine growth regulation mechanism in derivates of normal and neoplastic Müllerian epithelium.[ 9,10] Based on observations we have made in some individuals with breast or ovarian carcinoma treated with low-dose subcutaneous molgramostim(Drug information on molgramostim), chronic rather than intervallic administration of GM-CSF may be crucial for establishing a sustained antitumoral immune response (Kurbacher, unpublished data, 2002). This open-label pilot trial was therefore initiated to investigate the activity and toxicity of continuous low-dose GM-CSF monotherapy in heavily pretreated patients suffering from either metastatic breast or female genital tract carcinomas. Patients and Methods A total of 19 patients with intensively pretreated metastatic breast cancer (n = 7), recurrent ovarian carcinoma (n = 10), recurrent endometrial carcinoma (n = 1), or recurrent squamous cell cancer of the cervix uteri (n = 1) were recruited into this pilot trial. Prior to study entry, patients had failed a median of 4 prior systemic treatments (range: 2-7). Metastatic breast cancer patients had undergone 3 to 5 (median: 5) prior therapies including anthracyclines in 6 cases and taxanes in 5. Patients with gynecologic tumors had failed a median of 3 (range: 2-7) prior cytotoxic treatments, all comprising both platinum compounds and taxanes. Patients had either bidimensionally measurable lesions or evaluable disease (elevated or rising CA 125 serum levels). Inclusion criteria were as follows: (a) Karnofsky performance status ≥ 60%, (b) absence of severe hematologic and serologic abnormalities, (c) no history of colony-stimulating factor (CSF)-related homogeneously staining regions or any other serious adverse events related to the use of colony-stimulating factors, (d) absence of serious medical disorders other than those related to advanced tumor burden, (e) estimated life expectancy ≥ 8 weeks, and (f) written informed consent. This trial was performed in accordance with the institutional ethical guidelines and the German law (Arzneimittelgesetz). GM-CSF was administered as daily subcutaneous injections. Hematologic functions were controlled every second week by whole blood cell counts with leukocyte differentiation. Therapy was monitored by biweekly tumor marker determinations and an adequate tumor imaging performed every 8 weeks (in patients with measurable disease). The starting dose of GM-CSF was 125 μg/d, which was increased at increments of 25 μg/d after every second week of treatment up to a maximum of 250 μg/d until an objective tumor response was achieved. Dose intensification was discontinued in the case of any local or systemic toxicity, a total leukocyte count > 20 g/L, or eosinophils > 15%. GM-CSF therapy was continued until there was clear evidence of primary or secondary treatment failure or the occurrence of side effects exceeding National Cancer Institute Common Toxicity Criteria (NCICTC) grade 2. Responses were classified using common Eastern Cooperative Oncology Group (ECOG) criteria (in patients with measurable disease) or Rustin criteria (in patients with evaluable disease). Overall survival was calculated from the start of therapy until death from any cause or loss to follow-up. Results All patients treated were evaluable for both toxicity and response. The maintenance dose was 125 μg/d in 4 patients, 150 μg/d in 10, 175 μg/d in 2, 200 μg/d in 2, and 250 μg/d in 1 patient, respectively (median: 150 μg/d; mean: 158 μg/d) (see Table 1). Treatmentrelated toxicities are shown in Table 2. Ten patients developed mild to moderate leukocytosis (10-20 g/L). Generally, GM-CSF treatment was well tolerated. No patient suffered from hypersensitivity-like reactions, fluid retention, edema, or impairment of cardiac function. Mild to moderate flu-like symptoms (arthralgia, myalgia, bone pain, mild to moderate febrile reactions) not exceeding NCI-CTC grade 2 were the most frequent side effects. Grade 1 injection site reactions occurred in three patients. In only one patient, GM-CSF therapy was terminated because of unresolved grade 3 fatigue requiring hospitalization. Any other severe or life-threatening complications were not observed. Detailed information concerning therapeutic efficacy can be obtained from Table 3. One complete and six partial responses accounted for an objective response rate of 37% (confidence interval [CI]: 16%-62%) with a median response duration of 6 months (range: 4-13 months). Four additional patients achieved disease stabilization of 3 to10 months. Only eight patients progressed on therapy. In summary, 11 patients (58%, CI: 34%-80%) benefited from GM-CSF treatment. It should be noted that 5 of 7 patients responding to treatment and 8 of 11 patients who benefited had recurrent ovarian carcinoma. Recently, only one patient was free from progression and seven patients are still alive, resulting in a median progression-free survival of 4 months and a median overall survival of 6 months (see Figure 1). Patients benefiting from therapy had a progression- free survival of 6 months. Six of these patients are still alive, so that the median survival time in this subgroup has not yet been reached. Interestingly, 6 of 7 patients responding to GM-CSF and 9 of 11 patients benefiting from GM-CSF but only 1 of 8 patients with progression developed mild to moderate leukocytosis. Discussion GM-CSF is a powerful cytokine that plays a crucial role in the generation of antigen-presenting cells from blood and tissue precursors both ex vivo and in vivo.[2,5] Accompanying interleukins and interferons, it is therefore an integral component of different combination regimes used for both immunotherapy of clinical malignancies and ex vivo activation of dendritic cells. Recently, an increasing body of evidence exists suggesting that GM-CSF treatment is able to expand the dendritic cell population in both the peripheral blood and tumorassociated lymph nodes.[3,4] Compared to immunotherapy using ex vivo generated dendritic cells, GM-CSF given directly to patients may be a reasonable and easier to handle alternative. In advanced malignant melanoma, single-agent subcutaneous or perilesional GM-CSF has already been used successfully as systemic treatment after complete or incomplete surgical resection. However, experience with this therapy in other solid tumors including breast and gynecologic malignancies is still limited. In prostate cancer, some serologic responses and disease stabilizations have been reported.[7,8] The incidence of both severe toxicities and elevation of white blood cell counts was neglected in all these trials, which did not utilize GM-CSF continuously. However, our personal experience with two patients previously treated with molgramostim has led us to conclude that continuous exposure to GMCSF may be crucial for a maintained immunologic response (Kurbacher, unpublished data, 2002). Since the balance of pro- and antitumoral effects induced by GM-CSF appears to be dose-dependent, administration of very low doses may well be another important determinant of its clinical efficacy. This trial, which to our knowledge is the first of its kind in breast or female genital tract cancer, was therefore initiated to evaluate both the safety and antineoplastic efficacy of chronic low-dose subcutaneous GM-CSF. Starting with a very low dose of 125 μg per day, most patients were exposed to chronic GM-CSF at 150 to 175 μg daily. In general, the treatment was well tolerated. Only one patient experienced grade 3 toxicity (fatigue) and was thus withdrawn from the study. As could be expected from other immunotherapy trials, flu-like symptoms were frequent but never dose-limiting. Regarding the intensive pretreatment of most patients, the objective response rate of 37% with one complete response and six partial responses was exceptionally high considering that another four patients experienced durable disease stabilization. The median response duration (6 months) was long enough to indicate significant clinical utility. Interestingly, a mild to moderate increase of the whole blood count was noticed in 6 of 7 responders but in only 1 of 8 patients progressing while receiving GM-CSF. Leukocytosis may thus be a valuable surrogate marker for further clinical response to GM-CSF. At present, there is no evidence that chronic bone marrow stimulation was harmful to any of the patients. In particular, there was no evidence of impaired cardiac function in the treated population, which may be associated with higher doses of GM-CSF. Although the number of patients is still too small to draw any definite conclusion, a trend suggests that patients with tumors arising from the Müllerian epithelium (specifically, ovarian carcinoma) may benefit the most from chronic GM-CSF therapy. This seems to be of particular interest since several authors have shown that GM-CSF acts as a negative autocrine growth factor in the normal and neoplastic endometrium.[9,10] It may therefore well be that GM-CSF in these tumors has a pleitropic effect by both activating the dendritic cell-mediated antitumor response and directly inducing a growth arrest by stimulating intratumoral GM-CSF receptors. Conclusion In conclusion, chronic low-dose GM-CSF was safe and highly active in this heavily pretreated population of patients. The results of our study offer a new and practical approach to immunotherapy in patients with solid tumors others than melanoma or renal cell cancer and, therefore, warrant confirmation by larger-scaled clinical trials.