Febrile reactions to platelet transfusion: the effect of increased interleukin 6 levels in concentrates prepared by the platelet-rich plasma method.

BACKGROUND: A relation between febrile reactions to platelet transfusion and high cytokine levels in platelet concentrates (PCs) was found previously. The levels of cytokines such as interleukin (IL)-6 are related to the while cell content of the PC during storage. Therefore, early removal of white cells should prevent reactions. STUDY DESIGN AND METHODS: This prospective study was set up to compare methods for the preparation of random PCs, the platelet-rich plasma method (PRP-PCs), which results in a high white cell content, and the buffy coat method (BC-PCs), which results in a low white cell content, with regard to the frequency and severity of reactions to platelet transfusion and the IL-6 level of the PC. IL-6 was chosen because it is the major mediator of the acute-phase response. White cells were reduced in all PCs before transfusion. RESULTS: Platelet transfusions (n = 584) in 64 patients were studied. An overall reaction frequency of 7.2 percent was observed. Transfusion reactions were seen predominantly in patients who received PRP-PCs (PRP-PCs: 9.3% vs. BC-PCs: 2.7%, p = 0.007). Allergic reactions were limited to transfusions of PRP-PCs. The following PRP-PC characteristics were significantly correlated with febrile transfusion reactions: IL-6 level (p < 0.0001), initial white cell count (p = 0.001), and storage time (p = 0.02). In this group, reactions were less frequent in patients receiving pretransfusion medication (p < 0.001). In the PRP-PC group, IL-6 content (p = 0.01) and initial white cell count (p = 0.04) were also significantly correlated with allergic reactions, which indicated that these or associated factors might have an effect on the outcome of this type of reaction. CONCLUSION: Febrile reactions are highly correlated with IL-6 levels in PCs. The low white cell content of BC-PCs is associated with undetectable IL-6 levels and a reduced frequency of febrile as well as allergic reactions in recipients. The BC method is the preferable one for the production of random-donor PCs.

Interleukin 4 and interferon gamma costimulate the expansion of early human myeloid colony-forming cells. Proposal of a model for the regulation of myelopoiesis by interleukin 4 and interferon gamma and its integration with the regulation of the immune response.

We have previously shown that interleukin 4 (IL-4) and interferon gamma (INF-gamma) reciprocally regulate the production of granulocytes and monocytes from mature monopotential hematopoietic progenitor cells, while at the level of the very primitive stem cells IFN-gamma is a selective inhibitor of proliferation and differentiation, and IL-4 has weak stimulatory effects. We investigated the effects of IL-4 and IFN-gamma on the expansion in suspension culture of myeloid colony-forming cells (CFCs) induced by either IL-3 or IL-1+IL-3, using on the one hand more differentiated CD34+HLA-DR strongly positive (HLA-DR++) and on the other hand more primitive Cd34+HLA-DR weakly positive (HLA-DR+/-) human bone marrow cells. It is shown that both IL-4 and IFN-gamma stimulate the IL-3- and IL-3+IL-1-induced expansion of the number of CFCs in the HLA-DR+/- population. In the presence, but not in the absence of IL-1, additive effects of IL-4 and IFN-gamma were seen. We could not demonstrate any IL-3-like effect by IL-4 on early human hematopoietic progenitors. No expansion of CFC number was seen in the HLA-DR++ population. Based on these data and on data which we have published previously, a model for the regulation of myelopoiesis by IL-4 and IFN-gamma is proposed. In this model, IL-4 and IFN-gamma, which are both immune recognition induced inflammatory cytokines, both stimulate the expansion and recruitment of early myeloid progenitors, whereas at the level of their terminal differentiation, the balance between both cytokines determines whether preferentially monocytes/macrophages (IFN-gamma) or granulocytes (IL-4) are being produced. At the level of the most primitive cells, the inhibitory action of IFN-gamma might prevent differentiative exhaustion of the stem cell compartment in situations of hematopoietic stress.

Interferon gamma selectively inhibits very primitive CD342+CD38- and not more mature CD34+CD38+ human hematopoietic progenitor cells.

To assess the effects of interferon gamma (IFN-gamma) on very primitive hematopoietic progenitor cells, CD34(2+)CD38- human bone marrow cells were isolated and cultured in a two-stage culture system, consisting of a primary liquid culture phase followed by a secondary semisolid colony assay. CD34(2+)CD38- cells needed at least the presence of interleukin 3 (IL-3) and kit ligand (KL) together with either IL-1, IL-6, or granulocyte-colony-stimulating factor (G-CSF) in the primary liquid phase in order to proliferate and differentiate into secondary colony-forming cells (CFC). Addition of IFN-gamma to the primary liquid cultures inhibited cell proliferation and generation of secondary CFC in a dose-dependent way. This was a direct effect since it was also seen in primary single cell cultures of CD34(2+)CD38- cells. The proliferation of more mature CD34+CD38+ cells, however, was not inhibited by IFN-gamma, demonstrating for the first time that IFN-gamma is a specific and direct hematopoietic stem cell inhibitor. IFN-gamma, moreover, preserves the viability of CD34(2+)CD38- cells in the absence of other cytokines. IFN-gamma could, therefore, play a role in the protection of the stem cell compartment from exhaustion in situations of hematopoietic stress and may be useful as stem cell protecting agent against chemotherapy for cancer.

Transforming growth factor-beta regulates the cell cycle status of interleukin-3 (IL-3) plus IL-1, stem cell factor, or IL-6 stimulated CD34+ human hematopoietic progenitor cells through different cell kinetic mechanisms depending on the applied stimulus.

The immediate cell kinetic response of highly purified human bone marrow progenitor cells (CD34+ sorted fraction) to the inhibitory effects of transforming growth factor-beta (TGF-beta) was studied using the BrdU-Hoechst flow-cytometric technique. The progenitor cells were stimulated with either interleukin-3 (IL-3) alone or with IL-3 in combination with IL-1, stem cell factor (SCF), or IL-6, and the inhibitory action of TGF-beta was evaluated in each phase of the first three consecutive cell cycles. Semisolid methylcellulose cultures were also performed to compare these initial events to the effects observed after 7, 14, and 21 days of incubation. Within the CD34+ compartment, the progenitor cells can be discriminated on a functional basis, i.e., in terms of TGF-beta sensitivity. Very primitive progenitors, recruited out of the G0 phase by IL-3 plus an early-acting factor (IL-1, SCF) are, upon addition of TGF-beta, arrested specifically in the G1 phase of the second cell cycle. In the clonogenic assays, the increased colony formation due to IL-1 or SCF was completely abolished by the counteracting effect of TGF-beta that diminished colony output back to the level of TGF-beta-plus-IL-3 supplemented colony growth. Addition of TGF-beta to CD34+ progenitors responding to IL-3 alone resulted in an overall retardation, but without an apparent specific accumulation of cells in any of the cell cycles. Finally, within the CD34+ compartment, there exists a subset of IL-3-responsive, but TGF-beta-insensitive, progenitor cells that were, upon addition of TGF-beta, not arrested at all. In conclusion, our results demonstrate that TGF-beta exerts different cell kinetic effects on CD34+ progenitor cell growth depending on the applied stimulus.

Effect of prestorage leukocyte removal on the cytokine levels in stored platelet concentrates.

The effect of the prestorage removal of leukocytes from platelet concentrates (PC) on the cytokine levels during its storage was studied. Two methods for leukocyte removal were examined: filtration and preparation of the PC by the buffy coat method. Cytokine levels were measured at various storage times. Highly increased levels of tumor necrosis factor-alpha (TNF-alpha; 120 +/- 131 ng/l) and interleukin 6 (IL-6; 988 +/- 494 ng/l) were found after a 5-day storage in the control group, whereas no increased levels were found in filtered PC (TNF-alpha 14 +/- 4 ng/l, IL-6 < 4 ng/l) or in buffy coat PC (TNF-alpha 8 +/- 2 ng/l, IL-6 < 4 ng/l). Furthermore an effect of the pooling of buffy coats or PC on the cytokine levels was not found. Transfusion of PC containing high levels of IL-6 and TNF-alpha has been associated with febrile transfusion reactions in the recipient and therefore the prestorage leukocyte removal might prevent these febrile transfusion reactions. The preparation of buffy coat PC, through its simplicity, seems to be the method of choice.

Differential regulation of the expression of CD38 and human leukocyte antigen-DR on CD34+ hematopoietic progenitor cells by interleukin-4 and interferon-gamma.

We studied the effects of interleukin-4 (IL-4) and interferon-gamma (IFN-gamma) on the expression of CD38 and human leukocyte antigen (HLA)-DR on purified CD34+ bone marrow progenitor cells. CD34+CD38- and CD34+HLA-DR- cells are largely nonoverlapping populations. After culture for 4 days in IFN-gamma, the expression of CD38 and HLA-DR is significantly increased and the disappearance of the CD38- and HLA-DR- populations is virtually complete. Moreover, IFN-gamma induces a population of CD34+ cells with a very high expression of CD38 (CD34+CD38++ cells), which were absent in the initial CD34+ population. IL-4 has no effect on the expression of CD38, but induces a limited but significant increase in the expression of HLA-Dr. After culture in IFN-gamma, CD34+ cells show a higher cloning efficiency of the colony-forming unit-macrophage (CFU-M) and burst-forming unit-erythroid (BFU-E) compared to cells cultured in medium alone. After culture in IL-4, a limited increase in CFU-granulocyte (CFU-G) and BFU-E is seen, whereas CFU-G, CFU-M, and BFU-E are increased after culture in IL-4 plus IFN-gamma. We further investigated the functional properties of the CD34+CD38++ cells generated in the presence of IFN-gamma. This cell population is highly enriched for BFU-E but partially depleted of CFU-M. Most of the CFU-M were found in the CD34+CD38+/-(CD34+CD38- and CD34+CD38+ cells) population.

Quantitative cell-cycle progression analysis of the first three successive cell cycles of granulocyte colony-stimulating factor and/or granulocyte-macrophage colony-stimulating factor-stimulated human CD34+ bone marrow cells in relation to their colony formation.

The bromodeoxyuridine (BrdU)-Hoechst flow cytometric technique was applied to study the immediate cell kinetic response of highly purified human (h) bone marrow progenitor cells (CD(34+)-sorted fraction) to h granulocyte colony-stimulating factor (G-CSF) and/or h granulocyte-macrophage colony-stimulating factor (GM-CSF). The technique permits us to differentiate cycling from noncycling cells and to make a quantitative assessment of cell cycles after stimulation. Semisolid agar and single-cell liquid cultures were also performed to compare these initial events to the effects observed after 14 days of culture. The combination of G-CSF plus GM-CSF, acting synergistically in day 14 cultures, was found to have a subadditive effect in the first cell cycles, thereby indicating partial overlap of the different target cells. However, this combination accelerated transit through the cell cycle, as could be seen from the higher number of cells in the third cell cycle after 72 hours of stimulation. We conclude that, apart from the unresponsive cells, the CD34+ compartment consists of cells responsive to both G-CSF and GM-CSF, and cells responsive to either one of the CSFs alone, and that the combination of the two CSFs speeds up the cell cycle traverse rate for a significant fraction of the target cells that are initially responsive for both G-CSF and GM-CSF. The latter supports the hypothesis of an overlapping signalling pathway of G-CSF and GM-CSF.

Interferon-gamma and interleukin-4 reciprocally regulate the production of monocytes/macrophages and neutrophils through a direct effect on committed monopotential bone marrow progenitor cells.

We studied the direct effects of interferon-gamma (IFN-gamma) in single cell colony assays of CD34+HLA-DR++ bone marrow progenitor cells stimulated by either granulocyte-colony-stimulating factor (G-CSF), interleukin(IL)-3, granulocyte/macrophage-colony-stimulating factor (GM-CSF), combinations of these CSF or medium conditioned by the 5637 human bladder carcinoma cell line. In this culture system IFN-gamma stimulated monocytic colonies (CFU-M) no matter which CSF or CSF combination was used to support them and inhibited granulocytic colonies (CFU-G) if they were generated in the presence of G-CSF. IL-4 antagonized the myelopoietic effects of IFN-gamma: the IFN-gamma-induced suppression of G-CSF-supported CFU-G, as well as the stimulation of CFU-M, were reversed by IL-4. In all cultures, IFN-gamma had a limited, but statistically non-significant, inhibitory effect on CFU-GM, which was not affected by the presence of IL-4. These data show that IFN-gamma and IL-4 reciprocally regulate the generation of myeloid cells involved in humoral (neutrophils) and cellular (macrophages) immune responses through a direct effect on monopotential myeloid progenitor cells.

Effects of interleukin-4 on myelopoiesis: localization of the action of IL-4 in the CD34+HLA-DR++ subset and distinction between direct and indirect effects of IL-4.

We compared the myelopoietic effects of interleukin-4 (IL-4) on CD34+HLA-DR+ and on CD34+HLA-DR++ bone marrow progenitor cells stimulated by either granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF) or interleukin-3 (IL-3). IL-4 stimulates G-CSF-induced colony-forming unit-granulocyte (CFU-G) and inhibits all colony types induced by GM-CSF and IL-3 in the HLA-DR++ population, but not in the HLA-DR+ population. In CD34+HLA-DR+ cells, however, a stimulation of G-CSF-supported CFU-G was also seen with sequential application of IL-4 in liquid cultures followed by G-CSF in agar cultures. In order to confirm that these are direct effects of IL-4, single-cell culture experiments were performed with CD34+HLA-DR++ cells. In these cultures IL-4 stimulates G-CSF-induced CFU-G and only inhibits colony-forming unit-macrophage (CFU-M) regardless of the CSF used to generate them.

Effects of interleukin-4 (IL4) on myelopoiesis: studies on highly purified CD34+ hematopoietic progenitor cells.

We studied the effects of interleukin 4 (IL4) on myelopoiesis supported by either granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage CSF (GM-CSF) or IL3 on purified CD34+ bone marrow progenitor cells. IL4 stimulates the colony-forming unit granulocyte (CFU-G) induced by G-CSF and inhibits all colony types supported by either IL3 and GM-CSF, although inhibition of CFU-M (macrophage) was significantly stronger than that of CFU-G and CFU-GM. When the cells were first incubated in liquid culture for 4 days in IL4, followed by agar culture in G-CSF, there was a significant increase in the number of CFU-G compared to cells which had been incubated in medium alone for 4 days before plating out in agar containing G-CSF. The inhibitory effects of IL4 on GM-CSF or IL3 supported colony formation, however, disappear with sequential incubation in IL4 in liquid culture followed by culture in agar with either GM-CSF or IL3.

Increased tumor necrosis factor alpha (TNF alpha), interleukin 1, and interleukin 6 (IL-6) levels in the plasma of stored platelet concentrates: relationship between TNF alpha and IL-6 levels and febrile transfusion reactions.

Increased interleukin 6 (IL-6) levels were found in 8 of 12 platelet concentrates (PCs) after 3 days of storage and in 10 of 12 PCs after 5 and 7 days of storage. Most of the PCs with an increased IL-6 level also showed increased tumor necrosis factor alpha (TNF alpha) and interleukin 1 beta (IL-1 beta) levels. Levels of IL-6 increased by 3 log10 over the base level during storage. Increased levels were found when the PC white cell count exceeded 3 x 10(9) per L. A linear correlation was found among the levels of TNF alpha, IL-1 beta, IL-1 alpha, and IL-6 in the PCs (r > 0.885). Comparison of the TNF alpha, IL-1 beta, and IL-6 levels in samples taken at various storage times indicates that the increased levels are the result of an active synthesis and release of interleukins during storage. In a second part of the study, 45 transfusions of white cell-reduced PCs were studied. Six transfusions were complicated by a febrile reaction. These reactions were related to high levels of IL-6 and TNF alpha in the PCs (p < 0.0001). These cytokines are known as endogenous pyrogens. These findings indicate that transfusion reactions might be due to the intravenous administration of plasma with high cytokine levels and might not always result from an antigen-antibody reaction.

Direct effects of 13-cis and all-trans retinoic acid on normal bone marrow (BM) progenitors: comparative study on BM mononuclear cells and on isolated CD34+ BM cells.

The effects of both 13-cis-and all-trans retinoic acid (RA) on colony formation of normal bone marrow (BM) progenitors were investigated in semi-solid (methylcellulose) assays, using either isolated CD34+ cells or BM mononuclear cells. Single cell liquid cultures were performed to further discriminate between direct and indirect effects. RA action results in significant decrease of colony forming units (CFUs). This effect is more pronounced starting from CD34+ progenitors than starting from total BM. This overall decrease in CFUs is due to selective inhibition of CFU-M (macrophage) and erythroid colonies (BFU-E). At the single cell level the CFU-M inhibition is confirmed with--in addition--a significant inhibition of CFU-GM (granulocyte-macrophage) and a marked stimulation of CFU-G (granulocyte)s. Both retinoids exert the above-mentioned effects. All-trans RA, however, is effective at a tenfold lower concentration (10(-7)M) than 13-cis RA (10(-6)M). Results on CD34+ BM fractions (substantially reduced in accessory cells) demonstrate that the described effects can probably be attributed to the direct action of RA on these progenitors; single progenitor (CD34+) cell liquid cultures further prove this point.

Reactions to platelet transfusion: the effect of the storage time of the concentrate.

Random platelet concentrates were pooled and depleted of leucocytes by centrifugation immediately prior to transfusion. The incidence and severity of reactions to 570 leucocyte-poor platelet transfusions in 74 patients were studied. An overall transfusion reaction rate of 13.7% was observed. The reaction rate to platelets stored for less than 3 days (8.7%) was significantly different from the reaction rate to platelets stored for 3 days or more (17.6%). Minor reactions as well as moderate and severe reactions were more frequent in the latter group. As most of the white blood cells were removed prior to transfusion, it is suggested that the reactions result from the transfusion of pyrogenic and/or vasoactive substances accumulated in the plasma of the concentrate during storage.

A randomized phase-I/II multicenter study of recombinant human granulocyte-macrophage colony-stimulating factor (GM-CSF) therapy for patients with myelodysplastic syndromes and a relatively low risk of acute leukemia. EORTC Leukemia Cooperative Group.

To assess the effects of GM-CSF in patients with myelodysplasia, a total of 101 patients with refractory anemia (RA), RA with ringed sideroblasts (RARS), and RA with an excess of blasts provided that the percentage of blasts in the bone marrow did not exceed 10% (RAEB) were enrolled in the EORTC Leukemia Cooperative Group study 06885. They were randomized to receive two daily subcutaneous injections of rhGM-CSF (mammalian, glycosylated, Sandoz/Schering-Plough) at a daily dose of either 108 micrograms glycoprotein (group I) or 216 micrograms glycoprotein (group II) for 8 weeks. Response was defined as an increase in Hb (greater than 2.5 g%), neutrophil count (more than 100%), or platelet count (more than 100%) without progression of the disease. After exclusion of 19 patients who did not meet the entry criteria, 82 were evaluated. Fifty-four patients (66%) responded (27 of 42 patients in group I and 27 of 40 in group II). Progressive disease was seen in two patients of group I and in four of group II. Two of the latter developed leukemia. All responses were reflected in the granulocytic series. In two patients platelet numbers also increased. Cytogenetic analysis, successfully performed in 43 cases, showed that 14 of 16 patients with normal karyotypes responded, compared with 14 of 27 patients with abnormal karyotypes (p = 0.008). In some cases GM-CSF was reduced in dose or discontinued prematurely due to side effects so that only 35% of all evaluable patients finished 8 weeks of treatment without a change of dose.

Stromal populations and fibrosis in human long-term bone marrow cultures.

An immunofluorescence study of the adherent layer of human long-term bone marrow cultures (HLTBMC) revealed the following surface markers on the different stromal cell populations: stromal fibroblastic cells CD10+, FIB86.3+, CD13+, CD71+; adipocytes CD10+, FIB86.3-, CD13+, CD71-/+; and macrophages CD10-/+, FIB86.3+, CD13+, CD71-/+, CD14+, CD33+, CD25+, HLA-DR+, CD4+, CD19+, CD45+. The markers of the stromal fibroblastic cells in HLTBMC were similar to those of twice-passaged fibroblasts not only from bone marrow and spleen, but also from a hemopoietic non-supportive organ such as the skin. Some of the cultured human umbilical vein endothelial cells used as controls were found to be CD25+, demonstrating for the first time the interleukin-2 receptor p55 chain on normal non-hemopoietic cells. The stromal fibroblastic cells are overrepresented compared to the small non-macrophage hemopoietic cell population in the adherent layer of HLTBMC. In addition, silver staining revealed an increased reticulin content in most of the HLTBMC. An excessive growth of stromal fibroblastic cells and an excessive deposition of their product, the reticulin fibers, are the hallmark of myelofibrosis. The finding of equivalent observations in HLTBMC suggests that the hitherto unexplained, premature quenching of hemopoiesis in HLTBMC might at least partly be due to mechanisms similar to those operating in myelofibrosis in vivo.

Aberrant Ki-67 expression in normal bone marrow revealed by multiparameter flow cytometric analysis.

Ki-67 is a commercially available monoclonal antibody that reacts with a nuclear antigen detectable in proliferating cells only. Since its first description, it has been widely used as a "universal" proliferation marker and few groups have questioned the validity of the initially described reactivity, although this was tested only on very restricted experimental models. We wanted to check its reactivity on normal bone marrow (BM) samples using a multiparameter flow cytometric analysis. Although we were able to reproduce the findings of Ki-67 positivity on cultured and stimulated cells, we could not detect any convincing Ki-67 positivity on nuclei of normal BM samples. These samples all had a noticeable proliferating compartment as evidenced by their DNA content. These data are in contrast with the data we obtained starting from stressed marrows and marrows cultured in the presence of hematopoietic growth factors, where we found a marked Ki-67 positivity. This discrepancy suggests that bone marrow cells, growing and proliferating under steady-state conditions and guided by natural control mechanisms, may lose their Ki-67 expression upon exiting the progenitor compartment and entering the differentiating compartment.

Flow-cytometric analysis of erythrocytic blood group A antigen density profile.

Blood group A antigen density on red blood cells (RBC) was studied using flow cytometry (FCM) and fluoresceinated polyclonal and monoclonal IgG anti-A antisera. Agglutination was a problem, which could only be solved by prefixation of the RBC with glutaraldehyde or formaldehyde. However, this treatment resulted in a significant reduction of the number of antigen sites as compared to the native (i.e. nonfixed) RBC. Two major new findings came out of this study: (1) A antigen density on native RBC seems to be higher than previously recognized, and (2) A antigen density distribution is probably non-Gaussian. The absolute number of A antigen sites was determined, using a human polyclonal IgG antiserum and commercially available absolute fluorescence standards. The site numbers on fixed RBC were comparable to those found by earlier radioimmunological studies (x 10(6)/RBC): A1, 1.07 +/- 0.28; A2, 0.21 +/- 0.09; A1B, 0.79 +/- 0.26 sites (mean +/- SD). The values found for native RBC were considerably higher (x 10(6)/RBC): A1, 2.86 +/- 0.95; A2, 0.47 +/- 0.29; A1B, 1.98 +/- 0.58 sites (mean +/- SD). With the 1 monoclonal and the 3 polyclonal antisera used in this study, and in contrast to Rh D, the erythrocytic A antigen density distribution of a given sample is highly asymmetrical. This non-Gaussian distribution profile does not seem to be affected by such factors as antibody heterogeneity, variability in antibody fluoresceination range, RBC density and reticulocyte content. This suggests that the asymmetrical A antigen distribution may be an intrinsic property of the RBC population.

Pentostatin in refractory chronic lymphocytic leukemia: a phase II trial of the European Organization for Research and Treatment of Cancer.

Pentostatin was used to treat 26 patients with advanced B-cell chronic lymphocytic leukemia resistant to conventional treatment. Twenty patients had progressive disease on previous regimens and six had had partial remission and then relapsed 3-34 months after previous chemotherapy. Eleven patients had previously been treated with three different regimens. 10 had been treated with two regimens, and five had been treated with one regimen. Pentostatin was administered at a dosage of 4 mg/m2 weekly for 3 weeks, then 4 mg/m2 every other week for 6 weeks and once a month for 6 months. Seven of 26 assessable patients (27%) achieved partial remission and five (19%) achieved clinical improvement. The median duration of partial remission until relapse or death was 210 days. Myelosuppression was minor and transient in responsive patients, indicating some degree of selective effect on lymphocytes. Except for one patient who died of cerebral hemorrhage during the first 6 weeks of treatment, no drug-related deaths were registered. Major toxic effects included nausea in 17 patients (mainly grade 1), infections in 15, and liver enzyme elevations in five. Thus, pentostatin is active, even in patients with advanced B-cell chronic lymphocytic leukemia that is refractory to multiple chemotherapy regimens. Response can be achieved with mild myelosuppression.