In this study, we investigated the role of interleukin-lp (IL-1s) in the malignant evolution of chronic myelogenous leukemia (CML) and the functional activity of IL-1 inhibitors. Bone marrow (BM) and peripheral blood (PB) low-density cells from 38 CML patients were studied in the colonyforming unit-granulocyte, erythrocyte, monocyte, megakaryocyte colony culture assay. Samples from patients with early stage, interferon-a (IFN)-sensitive disease formed hematopoietic colonies in the presence of fetal calf serum (FCS), erythropoietin (Epo), and one of the following: granulocyte-macrophage colony-stimulating factor (10 ng/mL), IL-3 (15 ng/mL), both, or phytohemagglutinin-conditioned medium. The addition of IL-1p augmented IFN-sensitive CML colony growth in a dose-dependent manner at concentrations of 10 to 100 U/mL. In sharp contrast, addition of the above growth factors did not augment the colony growthpromoting effect of FCS and Epo in samples from IFN-resistant patients; further, adherent cell fractionation or

p202 is an IFN-inducible, primarily nuclear, phosphoprotein (52-kDa) whose constitutive overexpression in transfected cells inhibits colony formation. To investigate the molecular mechanism(s) by which expression of p202 protein impairs colony formation, we established stable cell lines that inducibly express p202. Using this cell model, we demonstrate that the induced expression of p202 in asynchronous cultures of these cells was accompanied by: (a) an increase in steady-state levels of p21(WAF1/CIP1/SDI1) (p21) mRNA and protein; (b) a decrease in Cdk2 protein kinase activity; and (c) an increase in the functional form of retinoblastoma protein (pRb). Transient transfection of a p202-encoding plasmid in Saos-2 cells, which do not harbor a wild-type p53 protein, resulted in an increase in p21 protein, which indicated that p202 could regulate expression of p21 protein independent of p53 protein. Moreover, we demonstrate that expression of p202 in these cells increased cell doubling time without accumulation of cells in a particular phase of the cell cycle. Taken together, these results are consistent with the possibility that p202 protein contributes to the cell growth retardation activity of the IFNs, at least in part, by modulating p21 protein levels.

p202 is an interferon (IFN)-inducible, primarily nuclear, phosphoprotein (52-kDa) whose overexpression in transfected cells inhibits colony formation. p202 binds to the retinoblastoma tumor suppressor protein and two other members of the pocket family proteins (p107 and p130). Moreover, overexpression of p202 in transfected cells inhibits the transcriptional activity of E2Fs (E2F-1/DP-1 and E2F-4/DP-1), p53, AP-1 c-Fos and c-Jun, NF-kappaB p50 and p65. Here we demonstrate that inhibition of endogenous p202 production in murine AKR-2B fibroblasts did not result in an increase in cell proliferation. Instead, these cells exhibited increased susceptibility to apoptosis in response to decrease in serum concentrations in the growth medium. These observations are consistent with the notion that normal levels of p202 may be needed for the regulation of cell proliferation.

Murine p202 is an interferon‐inducible primarily nuclear phosphoprotein (52 kDa) whose expression in transfected cells inhibits colony formation. p202‐binding proteins include the pocket proteins (pRb, p107 and p130), a p53‐binding protein (sm53BP1), and transcription factors (e.g. NF‐κB (p50 and p65), AP‐1 (c‐Fos and c‐Jun), E2F‐1, E2F‐4, MyoD, and myogenin). p202 modulates the transcriptional activity of these factors in transfected cells. Here we demonstrate that p202 self‐associates directly and a sequence in p202, which is conserved among the members of the 200‐family proteins, was sufficient for self‐association in vitro. Our observations reported herein raise the possibility that self‐association of p202 may provide a mechanism for the regulation of its activity.

A Phase I study of the novel angiogenesis inhibitor TNP-470 was performed. Patients with inoperable recurring or metastatic squamous cell cancer of the cervix with evaluable disease, no coagulopathy, and adequate renal, hepatic, and hematological function were eligible. One course of treatment consisted of an i.v. infusion of TNP-470 over 60 min every other day for 28 days, followed by a 14-day rest period. The starting dose was 9.3 mg/m2. Eighteen evaluable patients were treated, with a median age of 48 years (range 27-55) and performance status Zubrod 1 (range 0-2). Grade 3 neurotoxicities consisting of weakness, nystagmus, diplopia, and ataxia were encountered in two patients receiving the 71.2 mg/m2 dose. An intermediate dose level of 60 mg/m2 was evaluated and found to be well tolerated by three patients. Only one patient experienced grade 3 nausea on the 60 mg/m2 dose level. No myelosuppression, retinal hemorrhage, weight loss, or significant alopecia were observed. One patient had a complete response, which continues for 26 months, and three patients with initially progressive disease stage had stable disease for 5, 7.7, and 19+ months. Other Phase I studies, including over 200 patients, were performed concurrently with this study. Based on this experience, the dose of TNP-470 recommended for further studies is 60 mg/m2 as a 60-min i.v. infusion every Monday, Wednesday, and Friday. Neurotoxicity was dose limiting, but appears to be reversible. Otherwise, the treatment was well tolerated. The drug may be active in squamous cell cancer of the cervix. Further studies of TNP-470 in squamous cell cancer of the cervix are warranted.

p202 is an interferon-inducible protein whose expression in transfected cells inhibits proliferation. p202 binds to the retinoblastoma tumor suppressor protein in vitro and in vivo and the transcription factors AP-1 c-Fos and c-Jun, NF-kappaB p50 and p65, and inhibits the transcriptional activity of these factors in vivo. Here we report that p202 nonspecifically binds to double-stranded DNA and to single-stranded DNA in vitro. Analysis with recombinant p202 revealed that DNA binding activity is intrinsic to p202. A C-terminal deletion mutant of p202 exhibited DNA-binding properties, indicating that the C-terminus is dispensable for DNA binding. We also found that underphosphorylated p202 efficiently binds to DNA. Our data suggest that DNA binding activity of p202 may contribute to its functions.

Previous work showed that IFN-alpha induced the autoimmune-associated lupus inclusions (LI) in all 16 umbilical cord mononuclear cell samples from healthy mothers, In contrast, IFN-alpha induced LI and the LI-associated protein, p36, in only 2 of 16 human B lymphoblastoid cell lines, Resistance of these 14 cell lines to form LI and p36 may be due to their stage of development or differentiation or their transformed state, We sought to determine whether aging, neoplastic transformation, and HIV infection affected the observed IFN-alpha induction of LI in cord blood mononuclear cells, Expression of LI and p36 was investigated in PBMC on IFN-alpha chemotherapy and on culturing IFN-alpha with PBMC samples prepared from healthy adults and AIDS patients, The IFN-alpha induction of LI (detected by electron microscopy) or p36 (detected by two-dimensional gels) in all of the PBMC samples from these individuals was indistinguishable from the cord blood mononuclear cell response, Furthermore, induction of p36 and LI was not a good indicator of effective IFN-alpha chemotherapy, It may be consequential for autoimmunity induced by IFN-alpha in cancer, AIDS, and systemic lupus erythematosus (SLE), An essential biologic role for p36 and LI is suggested by a highly homologous p36 gene in the invertebrate Caenorhabditis elegans.

Many of the antimicrobial, immunomodulatory and cell growth inhibitory activities of the interferons are mediated by interferon‐inducible proteins. Earlier we characterized an interferon‐inducible murine protein, p202, whose expression in transfected cells inhibits cell proliferation and which can form a complex with retinoblastoma protein (pRb). Here we report that in transfected cells expression of p202 inhibits E2F‐stimulated transcription of a reporter gene and of endogenous genes. Inhibition of the transcriptional activity of E2F by p202 does not depend on fully functional pRb and is correlated with inhibition of the sequence‐specific DNA binding of E2F. p202 interacts with the transcription factor E2F (E2F‐1/DP‐1) in vitro and in vivo. Inhibition of E2F activity by p202 may contribute to growth inhibition by the interferons.

PURPOSE The curative resection rate in patients with potentially resectable carcinoma of the esophagus is approximately 55% and their median survival time is 11 months. Preoperative chemotherapy with high doses of chemotherapeutic agents was used to evaluate clinical and pathologic responses, curative resection rate, toxicity, and survival. Colony-stimulating factor (CSF) was added to reduce the severity of myelosuppression. PATIENTS AND METHODS Twenty-six consecutive assessable patients with potentially resectable adenocarcinoma of the esophagus or gastroesophageal junction were treated with two preoperative courses of intensive chemotherapy (etoposide, doxorubicin, and cisplatin [EAP]) with granulocyte-macrophage CSF (GM-CSF). Additional three conventional-dose postoperative chemotherapy courses without GM-CSF were given to patients who responded to preoperative chemotherapy. RESULTS A median of three courses (range, one to six), were administered. Of 27 patients, 26 were assessable for response to preoperative EAP; 13 (50%) achieved a major response. Among 23 patients who underwent surgery, 15 (65%) had a curative resection (58% of 26 assessable patients); none of the patients had a pathologic complete response, but two patients had only microscopic carcinoma in the resected specimen. Six patients had carcinoma present at the resection margins and received postoperative radiotherapy. Two patients were found to have liver metastases at exploration. At a median follow-up of 22 months, the median survival of 26 patients was 12.5 months (range, 2 to 32 +). Fourteen patients died of their carcinoma; two patients died of treatment-related causes; one died of an unrelated CNS arterial malformation; and the causes of death in two patients remain unknown. Seven patients are alive with no evidence of relapse. Major toxicities of this regimen included severe myelosuppression, nausea and vomiting, infections, and severe constitutional symptoms related to GM-CSF. However, subcutaneous injection of GM-CSF was well tolerated. CONCLUSION High-dose EAP is active against locoregional adenocarcinoma of the esophagus and gastroesophageal junction but can be associated with significant toxicity. Although this strategy remains attractive and needs to be developed further, less toxic and more effective regimens need to be identified.

Abstract. Increasing evidence suggests that paraneoplastic syndromes may be mediated by tumour‐related cytokine release, although the specific factor(s) involved remain poorly defined. Colony‐stimulating factors (CSF) and interleukins (IL) promote colony growth in semi‐solid media and, when administered in recombinant form, increase blood counts in patients. However, normal serum CSF levels in individuals with physiologic blood counts and the relationship between specific serum CSF levels and paraneoplastic leukaemoid reaction are not well established. In this study, we found that normal serum levels of granulocyte‐macrophage CSF (GM‐CSF), as measured by ELISA, were generally < 55 pg ml−1; IL‐3, < 30 pg ml−1; and granulocyte CSF (G‐CSF), < 50 pg ml−1. In contrast, high levels of GM‐CSF (132 pg ml−1), but not G‐CSF or IL‐3, were found in a patient with a transitional cell carcinoma of the renal pelvis and increased leukocytosis correlating with the tumour burden. The GM‐CSF was biologically active, as demonstrated by its ability to stimulate colony growth in vitro. Based on these results it appears that autonomous production of GM‐CSF is one possible pathophysiologic mechanism underlying leukaemoid reaction in cancer patients.

Interleukin-4 (IL-4) is a cytokine with pleiotropic activities. In normal bone marrow cultures grown in the presence of either granulocyte-macrophage colony-stimulating factor (GM-CSF) or interleukin-3 (IL-3), IL-4 suppresses granulocyte-macrophage colony-forming unit (CFU-GM) proliferation but it enhances the colony-stimulatory effect of granulocyte colony-stimulating factor (G-CSF). We studied the effect of IL-4 on chronic myelogenous leukemia (CML) bone marrow or peripheral blood cells from 30 patients using the CFU-granulocyte-erythrocyte-monocyte-megakaryocyte colony culture assay. In several repetitive experiments, IL-4 inhibited CFU-GM colony replication by 24 to 65% in a dose-dependent fashion at concentrations ranging from 0.01 to 10 micrograms/ml when patients' cells were cultured in the presence of erythropoietin alone or with phytohemagglutinin-conditioned medium, GM-CSF, or IL-3. The addition of 100 U/ml of IL-1 beta to the CML cultures partially reversed the inhibitory effect of IL-4. Incubation of CML low-density peripheral blood cells with IL-4 resulted in down-regulation of IL-1 beta and IL-6 production in three of four samples, suggesting that the suppressive effect of IL-4 is mediated by inhibition of IL-1 and by other mechanisms including inhibition of IL-6 production. In contrast to the stimulatory effect exerted by IL-4 on G-CSF-dependent CFU-GM progenitor proliferation in normal marrow, the addition of IL-4 to CML cultures grown in the presence of G-CSF resulted in a divergent effect: suppression of CML CFU-GM in two, stimulation in three, and no significant effect in two CML patients' samples. It is therefore possible that IL-4 may have an in vivo antiproliferative effect in a subpopulation of CML patients.

BACKGROUNDAlpha interferon (IFN) is an effective single agent for patients with low-grade lymphoma, but until 1982 had not been integrated with standard chemotherapy for these patients. Since relapse from complete remission (CR) is the rule for patients with advanced stage low-grade lymphoma, a maintenance IFN schedule was explored for patients in CR.PATIENTS AND METHODSFrom 1982-1987, 127 patients with stage IV low-grade lymphoma were treated with cyclophosphamide, doxorubicin, vincristine, prednisone, and bleomycin (CHOP-Bleo) for 9-18 mo. (median 13 mo.), followed by interferon alfa-n-1 (Wellferon) maintenance therapy for 24 mo. for CRs.RESULTSThe overall response rate for the entire treatment program was 73% CR and 23% partial remission. The median follow up was 59 months. At 5 years, survival was 74%, failure-free survival (FFS) 47%, and FFS of CRs 60%. Compared to a group of 96 patients with similar pretreatment clinical features treated with CHOP-Bleo from 1972-1982, this represents a significant improvement for both overall FFS (p = 0.01) and FFS of CRs (P < 0.01). Toxicity from the IFN maintenance was generally acceptable, but even at the modest dose employed in this trial (3 x 10(6) U/m2 three times weekly), dose modification was required in more than 30% of patients, usually because of fatigue.CONCLUSIONSThe integration of IFN and conventional chemotherapy is feasible and effective. Maintenance IFN prolongs remission duration for patients with stage IV low-grade lymphoma.

Twenty-five patients with systemic malignancies or cutaneous disorders were treated with low dosages (1-16 IU/kg/day) of oral interferon-alpha in a phase-I study. There were no grade-3 or -4 toxicities. No patients achieved a partial or complete response. One patient with renal cell carcinoma had prolonged disease stabilization lasting for 22 months. No significant changes in lymphocyte subsets including T4/T8 ratio, or serum immunoglobulins were observed. Subjective improvement in general well-being was reported by three patients. Low dosages of interferon-alpha administered orally do not appear to have any significant immunomodulatory or antitumor activity.

We studied the in vitro effects of granulocyte-macrophage colony-stimulating factor (GM-CSF) in 13 patients with acute myeloid leukemia (AML) and one patient with refractory anemia with excess of blasts in transformation using the AML blast (AML colony-forming units, AML-CFU) and mixed (granulocyte erythrocyte macrophage megakarocyte colony-forming units, CFU-GEMM) colony culture assays. In parallel, these patients received GM-CSF s.c. at 125-mu-g/m2/day, or in escalated doses starting with 10-mu-g/m2/day for a week or until circulating blast counts reached 50 x 10(9)/liter, in an effort to sensitize leukemic blasts to cell-cycle-specific agents. Results of in vivo GM-CSF treatment were correlated with those of in vitro assays. In 9 of 12 patients (75%), GM-CSF treatment increased peripheral blood blast counts (in vivo effect). GM-CSF also stimulated in vitro AML blast colony proliferation in these nine patients and increased the S + G2M phases of the cell cycle in five out of five of these patients' samples. Two of three patients in whom an in vivo response could not be demonstrated also failed to have a detectable in vitro response. These observations suggest that the AML blast colony culture assay may be useful in predicting the response of AML to cytokine therapy. Finally, GM-CSF stimulated granulocyte-macrophage (granulocyte-macrophage colony-forming units, CFU-GM) and erythroid (erythroid burst-forming units, BFU-E) colony proliferation in 14 and 11 patients, respectively, including the 3 individuals who demonstrated no clinical effect on blast counts. It is, therefore, possible that GM-CSF may be used to stimulate proliferation of progenitors that differentiate into mature granulocyte, monocyte-macrophage, and erythroid cells.

Interleukin-1 (IL-1) has recently been reported to play an important role in acute myelogenous leukemia (AML) blast proliferation. We therefore investigated the effect of soluble IL-1 receptors (sIL-1R) and IL-1 receptor antagonist (IL-1RA) on the growth of AML bone marrow blast progenitors from 25 patients. In the AML blast colony culture assay, sIL-1R and IL-1RA inhibited blast colony-forming cell replication in a dose-dependent fashion, at concentrations ranging from 10 to 500 ng/mL (sIL-1R) and 10 to 1,000 ng/mL (IL-1RA), and their inhibitory effect was partially reversed by IL-1 beta. A similar inhibitory effect was also noted with the use of anti-IL-1 beta neutralizing antibodies. When AML blast progenitors were grown either in the presence of fetal calf serum (FCS) alone or with one of the following: phytohemagglutinin leukocyte-conditioned medium (PHA-LCM), granulocyte-macrophage colony-stimulating factor (GM-CSF), G-CSF, interleukin-3 (IL-3), or stem cell factor (SCF), addition of 100 ng/mL sIL-1R or IL-1RA inhibited blast colony formation by 3% to 96% and 2% to 97%, respectively. In sharp contrast, neither of these IL-1- inhibitory molecules significantly inhibited proliferation of normal marrow hematopoietic progenitors. Lysates of 2 x 10(7) low-density AML marrow cells were tested for intrinsic IL-1 beta content using an enzyme-linked immunoadsorbant assay (ELISA). Samples from five of six patients showed high concentrations (ranging from 501 to 2,041 pg), whereas 2 x 10(7) cells from two normal marrow aspirates yielded 54.6 pg of IL-1 beta. AML blast colony-forming cells from all six patients were inhibited by sIL-1R, IL-1RA, or both. Incubation of nine samples of AML low-density cells with either sIL-1R or IL-1RA reduced GM-CSF concentrations in cell lysates, and supernatants from nine (P less than .01) and six samples (P less than .037), respectively, and G-CSF concentration in lysates from six of nine samples (P less than .03), and in supernatants from five of six samples (P less than .06) when studied by ELISAs. Our data implicate IL-1 in AML blast proliferation and suggest the potential benefits of using IL-1-inhibitory molecules in future therapies for AML.

PURPOSEA prospective study in breast cancer patients was undertaken to determine whether escalating doses of doxorubicin and cyclophosphamide would result in a higher fraction of patients free of disease, and to evaluate the role of leukocyte alpha-interferon.PATIENTS AND METHODSBetween 1982 and 1986, 319 consecutive patients with stage II or III breast cancer with one or more positive nodes were assigned randomly to receive adjuvant chemotherapy that consisted of escalating doses of doxorubicin and cyclophosphamide in combination with vincristine and prednisone or the same chemotherapy regimen followed by 1 year of leukocyte alpha-interferon. Doxorubicin was administered by 72-hour continuous infusion through a central venous catheter (maximum total cumulative dose, 430 mg/m2). All patients with positive or unknown estrogen receptor status were also given tamoxifen for 1 year.RESULTSThe median follow-up was 71 months (range, 35 to 99 months). Correlation of disease-free survival (DFS) with dose-intensity of cyclophosphamide and doxorubicin showed no improvement in DFS for patients who were able to receive escalated drug doses compared with those who were not. Doxorubicin administered by continuous infusion was associated with a negligible risk of cardiotoxicity in this study despite the administration of higher accumulative doses than in our previous adjuvant therapy studies. The DFS rates of patients who did and those who did not receive leukocyte alpha-interferon were similar.CONCLUSIONSIn this study, there was no real evidence that higher drug dose intensity was associated with longer DFS. Leukocyte alpha-interferon as it was used in this study had no therapeutic value. Doxorubicin administered by infusion was associated with a reduced risk of cardiotoxicity.

BACKGROUND Retinoids (vitamin A derivatives) and interferon-alpha (IFN-alpha) are potent regulators of malignant cell differentiation and proliferation, and both have immunomodulatory and antiangiogenesis activity. A large body of preclinical and clinical data supports the use of combination therapy with 13-cis-retinoic acid (13-cRA) and IFN-alpha in patients with squamous cell carcinoma of the skin. This carcinoma is an extremely common and frequently severely disfiguring cancer, for which about 10% of patients remain uncured following standard local therapy. PURPOSE Our purpose was to test whether a 20% or greater complete response rate could be achieved using a combination of these two agents in patients with advanced squamous cell carcinoma of the skin in whom local therapy had failed or was unfeasible or who had regional and/or distant metastases. METHODS Thirty-two patients with heavily pretreated, advanced inoperable cutaneous squamous cell carcinoma of the skin were given a combination of oral 13-cRA (1 mg/kg per day) and subcutaneous recombinant human IFN alpha-2a (3 million units per day) for at least 2 months, unless disease progressed earlier, in a phase II trial. RESULTS Nineteen (68%) of the 28 assessable patients responded, seven (25%) of whom had complete responses. Response rates were 93% (13 of 14) in patients with advanced local disease (six complete responses), 67% (four of six) in patients with regional disease (no complete responses), and 25% (two of eight) in patients with distant metastases (one complete response). The relationship between decreased response rate and increased extent of disease was highly statistically significant (P less than .005, chi-square test). The median response duration was greater than 5 months. No life-threatening toxic effects occurred in assessable patients treated with this combination, although dose reductions were required in 18 patients. The major limiting side effect in this elderly patient population (median age, 67 years) was cumulative fatigue. CONCLUSION These results indicate that combined systemic therapy with 13-cRA and IFN alpha-2a is highly effective in patients with advanced squamous cell carcinoma of the skin.

alpha-Interferon induces hematological and cytogenetic remissions in some individuals with newly diagnosed Philadelphia-positive chronic myelogenous leukemia. However, interferon-resistant disease occurs in a consistent patient subset (primary resistance) and develops during therapy in additional patients (secondary resistance). Several alpha-interferon-inducible genes have been characterized. In interferon-resistant cell line variants, defects in these genes have been implicated in the mechanisms mediating resistance. We have, therefore, evaluated mRNA expression of four interferon-stimulated genes (ISGs) following alpha-interferon therapy. Twenty-seven chronic myelogenous leukemia patients (ten interferon-sensitive patients, 17 interferon-resistant patients) were studied. Peripheral blood samples were collected prior to and 1 to 7 days after starting interferon therapy and analyzed for the expression of 2'-5' oligoadenylate synthetase, ISG-15, ISG-54, and 6-16 transcripts. Following therapy with alpha-interferon, 2'-5' oligoadenylate synthetase, ISG-54, and 6-16 transcripts were discerned in all patients regardless of their response to interferon. The ISG-15 message was detected in eight of nine interferon-sensitive and in 15 of 16 interferon-resistant patients, as well. Overall, no consistent defect in the ISG system could be identified. Therefore, lack of induction of these genes cannot explain resistance to alpha-interferon in chronic myelogenous leukemia patients. Other mechanisms such as posttranslational modification, leading to defects in the ISG corresponding proteins, may play a role in the development of resistance.

PURPOSE AND RATIONALE: There has been no previously published experience with granulocyte-macrophage colony-stimulating factor (GM-CSF) at doses less than 15-mu-g/m2/d in patients with aplastic anemia, and most observations have been made at doses of 100 to 500-mu-g/m2/d (2.5 to 12.5-mu-g/kg/d). The benefits of using considerably lower doses, if effective, should include a decrease in cost and in side effects. We have therefore used very low doses of GM-CSF to treat a group of patients with aplastic Additionally, since severe anemia is often a problem in these patients, we recently started administering erythropoietin along with the GM-CSF. Herein we report the results of very-low-dose GM-CSF therapy in patients with aplastic anemia and our preliminary findings in those individuals who received combination therapy.PATIENTS AND METHODS: We administered recombinant human GM-CSF subcutaneously at doses of 5 to 20-mu-g/m2/d ("very-low-dose GM-CSF") to 12 patients with aplastic anemia. In addition, a 13th patient received erythropoietin together with the GM-CSF regimen, and three of the 12 individuals who initially received 1 or more months of GM-CSF alone were later also given erythropoietin (4,000 U/d subcutaneously).RESULTS: In five of 12 patients (42%) treated with very-low-dose GM-CSF, an increase in neutrophil counts (2.0- to 6.7-fold) was noted, and one of these subjects attained a bilineage response (neutrophil counts, 0.3 to 1.75 x 10(9)/L; platelet counts, 8 to 169 X 10(9)/L). Moreover, a sixth patient showed a rise in platelet counts (19 to 80 x 10(9)/L) without a concomitant increase in neutrophils. Constitutional side effects were minimal. Combining erythropoietin and very-low-dose GM-CSF produced a bilineage response (neutrophils, 1.0 to 3.0 X 10(9)/L; hemoglobin, 7.4 to 9-4 g/dL) in the one patient who received erythropoietin together with the GM-CSF from the time that GM-CSF was initiated. In one of the other patients who were given combination therapy, the addition of erythropoietin appeared to enhance the response; this patient demonstrated a neutrophil response to GM-CSF alone and a trilineage response (neutrophils, 0.8 to 3.75 X 10(9)/L; hemoglobin, 7.0 to 13.1 g/dL; and platelets, 10 to 34 X 10(9)/L) to the combination. No toxicity was associated with the addition of erythropoietin. CONCLUSIONS. Our observations Suggest that (1) very low doses of GM-CSF (5 to 20-mu-g/m2/d subcutaneously) may be used initially in neutropenic patients with aplastic anemia, and the dose subsequently increased only in patients who do not respond; and (2) the administration of erythropoietin together with GM-CSF is well tolerated, can augment responsiveness in some patients, and deserves further study.