MRzero - Automated discovery of MRI sequences using supervised learning

PURPOSE

A supervised learning framework is proposed to automatically generate MR sequences and corresponding reconstruction based on the target contrast of interest. Combined with a flexible, task-driven cost function this allows for an efficient exploration of novel MR sequence strategies.

METHODS

The scanning and reconstruction process is simulated end-to-end in terms of RF events, gradient moment events in x and y, and delay times, acting on the input model spin system given in terms of proton density, T 1 and T 2 , and Δ B 0 . As a proof of concept, we use both conventional MR images and T 1 maps as targets and optimize from scratch using the loss defined by data fidelity, SAR penalty, and scan time.

RESULTS

In a first attempt, MRzero learns gradient and RF events from zero, and is able to generate a target image produced by a conventional gradient echo sequence. Using a neural network within the reconstruction module allows arbitrary targets to be learned successfully. Experiments could be translated to image acquisition at the real system (3T Siemens, PRISMA) and could be verified in the measurements of phantoms and a human brain in vivo.

CONCLUSIONS

Automated MR sequence generation is possible based on differentiable Bloch equation simulations and a supervised learning approach.

Efficacy and safety of lipegfilgrastim versus pegfilgrastim in elderly patients with aggressive B cell non-Hodgkin lymphoma (B-NHL): results of the randomized, open-label, non-inferiority AVOID neutropenia study

Background

Lipegfilgrastim has been shown to be non-inferior to pegfilgrastim for reduction of the duration of severe neutropenia (DSN) in breast cancer patients. This open-label, non-inferiority study assessed the efficacy and safety of lipegfilgrastim versus pegfilgrastim in elderly patients with aggressive B cell non-Hodgkin lymphoma (NHL) at high risk for chemotherapy-induced neutropenia.

Patient and methods

One hundred and one patients (median age, 75 years) were randomized to lipegfilgrastim or pegfilgrastim (6 mg/cycle) during six cycles of R-CHOP21.

Results

Lipegfilgrastim was non-inferior to pegfilgrastim for the primary efficacy endpoint, reduction of DSN in cycle 1. In the per-protocol population, mean (standard deviation) DSN was 0.8 (0.92) and 0.9 (1.11) days in the two groups, respectively; the adjusted mean difference between groups was − 0.3 days (95% confidence interval, − 0.70 to 0.19). Non-inferiority was also demonstrated in the intent-to-treat population. The incidence of severe neutropenia in cycle 1 was 51% (21/41) in the lipegfilgrastim group and 52% (23/44) in the pegfilgrastim group. Very severe neutropenia (ANC < 0.1 × 10⁹/L) in cycle 1 was reported by 5 (12%) patients in the lipegfilgrastim group and 8 (18%) patients in the pegfilgrastim group. However, over all cycles, febrile neutropenia (strict definition) was reported by only 1 (2%) patient in each treatment group (during cycle 1 in the lipegfilgrastim group and cycle 6 in the pegfilgrastim group). The mean time to absolute neutrophil count recovery (defined as ≥ 2.0 × 10⁹/L) was 8.3 and 9.4 days in the two groups, respectively. Serious adverse events occurred in 46% of patients in each group; none were considered treatment-related. Eight patients died during the study (2 in the lipegfilgrastim group, 5 in the pegfilgrastim group, and 1 who died before starting study treatment). No deaths occurred during the treatment period, and all were considered to be related to the underlying disease.

Conclusions

This study shows lipegfilgrastim to be non-inferior to pegfilgrastim for the reduction of DSN in elderly patients with aggressive B cell NHL receiving myelosuppressive chemotherapy, with a comparable safety profile.

Trial registration number

ClinicalTrials.gov identifier NCT02044276; EudraCT number 2013-001284-23

Electronic supplementary material

The online version of this article (10.1007/s00520-020-05711-7) contains supplementary material, which is available to authorized users.

OP0239 WHY DOES ALCOHOL INHIBIT ARTHRITIS? - AN EXPLANATION OF THE MECHANISM OF ARTHRITIS INHIBITION BY ETHANOL

Background:

Alcohol consumption has emerged as consistent protective factor for the development of autoimmune diseases such as rheumatoid arthritis (RA). The underlying mechanism for this tolerance-inducing effect of alcohol, however, is unknown.

Objectives:

To understand the anti-arthritogenic effect of alcohol

Methods:

The immune-regulatory properties of alcohol consumption in vivo were tested in the collagen-induced arthritis (CIA) and serum-induced arthritis (SIA) model as well as after immunization with T cell- dependent (NP-CGG) and independent (TNP-FICOLL) antigens. Additional experiments in vivo experiments in these models were done with acetate- the metabolite of ethanol. The models were analysed for T- cell lineage and plasma cell differentiation, germinal centre formation and IgG levels and sialylation. Molecular expression of T follicular helper cell (TFH) activation such as IL-21, Bcl-6 and PD-1, as well as TFH: B cell conjugates were also assessed. Furthermore, TFH cells were generated in vitro, exposed to ethanol or acetate and tested for IL-21 production, PD1 expression and conjugate formation with B cells.

Results:

Ethanol exposure significantly inhibited arthritis in the active adaptive immunity-driven model of arthritis (CIA) but not in the passive innate immunity-driven model (STA) suggesting that the immune suppressive effect of alcohol is based on interference of T- and B- cell activation. In line ethanol and even more its metabolite acetate, suppressed T cell dependent antibody formation after NP-CGG immunization, while T cell independent antibody formation after TNP-FICOLL immunization was not suppressed. Ethanol, as well as its metabolite acetate, specifically altered the functional state of T follicular helper (TFH) cells in vitro and in vivo, thereby exerting immune regulatory and tolerance-inducing properties. Alcohol-exposed mice showed reduced Bcl6 and PD-1 expression as well as interleukin (IL)-21 production by TFH cells, preventing proper spatial organization of TFH cells to form TFH: B cell conjugates in the germinal centre. This effect of alcohol on TFHcells was associated with impaired autoantibody formation, higher sialylation of autoantibodies and less arthritis. In accordance, overexpression of IL-21 in vivo completely reversed the immune regulatory effects of alcohol.

Conclusion:

In summary, these data provide a new mechanistic explanation for the immune regulatory and tolerance-inducing effect of alcohol consumption in arthritis.

Acknowledgments:

Funden by DFG-FOR2886, DFG–CRC1181, Staedtler foundation, Johannes und Frieda Marohn-Stiftung, Else Kröner-Fresenius foundation, Interdisciplinary Centre for Clinical Research, Erlangen, BMBF-MASCARA, IMI funded project RTCure.

Disclosure of Interests:

Vugar Azizov: None declared, Maria V Sokolova: None declared, Kerstin Sarter: None declared, Vladimir Temchura: None declared, Ulrike Steffen (née Harre): None declared, Martin Herrmann: None declared, Georg Schett Speakers bureau: AbbVie, BMS, Celgene, Janssen, Eli Lilly, Novartis, Roche and UCB, Mario Zaiss: None declared

OP0245 MICROBIOTA-INDUCED INTESTINAL BARRIER DYSFUNCTION PRECEDES THE ONSET OF ARTHRITIS AND ALLOWS THE SHUTTLING OF IMMUNE CELLS FROM THE GUT THE JOINTS

Background:

While it is known that microbial dysbiosis is associated with the onset of rheumatoid arthritis, mechanistic insights how it facilitates the development of arthritis remained largely elusive to date. It is especially interesting how microbial dysbiosis affects the transition from asymptomatic autoimmunity to arthritis. We speculated that a breakdown of intestinal barrier function caused by microbial dysbiosis allows immune cells to shuttle from the gut to the joints.

Objectives:

To test whether intestinal barrier function is impaired before the onset of human RA and experimental arthritis and to seek for evidence that immune cells from the gut migrate to the joints.

Methods:

In a longitudinal cohort of RA-at risk individuals markers of disturbed intestinal barrier function, such as zonulin, were analysed and linked to RA onset. Furthermore, new-onset RA patients were assessed for gut leakiness and their intestinal biopsies for the expression of tight junction proteins and immune cell infiltration. In the murine model of collagen-induced arthritis, sequential analysis of intestinal dysbiosis, intestinal barrier function and arthritis onset was carried out. Additionally, barrier function was assessed on intestinal organoids exposed to faecal supernatants from eu- and dysbiotic mice with and without inhibition of zonulin. Furthermore, three types of interventions restoring intestinal barrier function were carried out for testing their effects on the inhibition of arthritis onset. Finally, photo- converted cells from the gut were traced in the joints to test for cellular trafficking from one to the other compartment.

Results:

Zonulin, a potent regulator for intestinal tight junctions, was elevated in autoimmune mice and men before the onset of arthritis and predicted the onset of human RA. Intestinal barrier functions as well as epithelial tight junctions were decreased before the onset of experimental arthritis and at onset of human RA. In mice, induction of autoimmunity was followed by rapid intestinal dysbiosis followed by gut leakiness before arthritis started. Faecal supernatants of arthritic mice induce epithelial barrier dysfunction in intestinal organoids in zonulin dependent manner. Restoration of the intestinal barrier in the pre-phase of arthritis using butyrate, CB1R agonist or zonulin antagonist larazotide inhibited the development of arthritis. Finally, using photoconvertible mice, gut-borne immune cells were identified that homed to the joints when barrier function was impaired.

Conclusion:

In summary, these data show the intestinal barrier dysfunction precedes the onset of RA and allows the trafficking of immune cells from the gut to the joints. Targeting of intestinal tight junction function may therefore allow preventing the onset of RA.

Acknowledgments:

Funded by the DFG-FOR2886 PANDORA, DFG–CRC118, Staedtler foundation, Johannes und Frieda Marohn-Stiftung, Else Kröner-Fresenius foundation, Interdisciplinary Centre for Clinical Research, Erlangen (IZKF), BMBF-MASCARA and the IMI funded projectRTCure.

Disclosure of Interests:

Mario Zaiss: None declared, Narges Taijc: None declared, Kerstin Sarter: None declared, Vugar Azizov: None declared, laura Bucci: None declared, Yubin Luo: None declared, Juan de Dios Cañete: None declared, francesco ciccia Grant/research support from: pfizer, novartis, roche, Consultant of: pfizer, novartis, lilly, abbvie, Speakers bureau: pfizer, novartis, lilly, abbvie, Georg Schett Speakers bureau: AbbVie, BMS, Celgene, Janssen, Eli Lilly, Novartis, Roche and UCB

Combined PET/MRI: Global Warming-Summary Report of the 6th International Workshop on PET/MRI, March 27-29, 2017, Tübingen, Germany

The 6th annual meeting to address key issues in positron emission tomography (PET)/magnetic resonance imaging (MRI) was held again in Tübingen, Germany, from March 27 to 29, 2017. Over three days of invited plenary lectures, round table discussions and dialogue board deliberations, participants critically assessed the current state of PET/MRI, both clinically and as a research tool, and attempted to chart future directions. The meeting addressed the use of PET/MRI and workflows in oncology, neurosciences, infection, inflammation and chronic pain syndromes, as well as deeper discussions about how best to characterise the tumour microenvironment, optimise the complementary information available from PET and MRI, and how advanced data mining and bioinformatics, as well as information from liquid biomarkers (circulating tumour cells and nucleic acids) and pathology, can be integrated to give a more complete characterisation of disease phenotype. Some issues that have dominated previous meetings, such as the accuracy of MR-based attenuation correction (AC) of the PET scan, were finally put to rest as having been adequately addressed for the majority of clinical situations. Likewise, the ability to standardise PET systems for use in multicentre trials was confirmed, thus removing a perceived barrier to larger clinical imaging trials. The meeting openly questioned whether PET/MRI should, in all cases, be used as a whole-body imaging modality or whether in many circumstances it would best be employed to give an in-depth study of previously identified disease in a single organ or region. The meeting concluded that there is still much work to be done in the integration of data from different fields and in developing a common language for all stakeholders involved. In addition, the participants advocated joint training and education for individuals who engage in routine PET/MRI. It was agreed that PET/MRI can enhance our understanding of normal and disrupted biology, and we are in a position to describe the in vivo nature of disease processes, metabolism, evolution of cancer and the monitoring of response to pharmacological interventions and therapies. As such, PET/MRI is a key to advancing medicine and patient care.

An overview of CEST MRI for non-MR physicists

The search for novel image contrasts has been a major driving force in the magnetic resonance (MR) research community, in order to gain further information on the body’s physiological and pathological conditions.

Chemical exchange saturation transfer (CEST) is a novel MR technique that enables imaging certain compounds at concentrations that are too low to impact the contrast of standard MR imaging and too low to directly be detected in MRS at typical water imaging resolution. For this to be possible, the target compound must be capable of exchanging protons with the surrounding water molecules. This property can be exploited to cause a continuous buildup of magnetic saturation of water, leading to greatly enhanced sensitivity.

The goal of the present review is to introduce the basic principles of CEST imaging to the general molecular imaging community. Special focus has been given to the comparison of state-of-the-art CEST methods reported in the literature with their positron emission tomography (PET) counterparts.

A new contrast in MR mammography by means of chemical exchange saturation transfer (CEST) imaging at 3 Tesla: preliminary results

PURPOSE

To evaluate the feasibility to detect and delineate malignant breast lesions in human patients by chemical exchange saturation transfer (CEST) as an MR imaging technique without the need for contrast agent administration.

MATERIALS AND METHODS

Six female patients referred for pre-surgical staging due to histologically confirmed breast cancer were examined with MR at 3 T. The routine breast protocol included T (2w), STIR, T (1w)-DCE and contrast-enhanced T (1w) imaging with SPAIR fat suppression. For CEST imaging, a 3D RF-spoiled gradient echo (GRE) sequence with an optimized saturation pulse train was applied. To assess the diagnostic value of the technique, CEST effects observed between frequency offsets of 1.2 to 1.8 ppm from the bulk water resonance were compared to pharmacokinetic parameter maps (k (ep)) obtained by DCE-MRI.

RESULTS

In 3 of 6 patients, regions with high CEST signal intensity correlated well with tumor areas as determined by DCE-MRI. Analysis of signal intensities from ROIs in tumor, fibroglandular, adipose, and muscle tissue revealed significantly higher CEST values in tumor tissue compared to fibroglandular tissue. The detection of lesions was equally well possible with DCE-MRI and CEST-MRI. In the three other patients, the tumor regions could not be delineated based on the CEST image due to artifacts, which were most likely caused by a high content of fat tissue within the ROIs.

CONCLUSION

The results of this initial feasibility study indicate a significant potential of CEST-MRI to discriminate cancer from fibroglandular tissue in the human breast by a CEST contrast generated by endogenous solute molecules.

A multimodal treatment approach including high-dose chemotherapy in very advanced gastric cancer: evidence for control of metastatic disease

The present multimodal treatment approach was designed to achieve prolonged tumor control in advanced gastric cancer. A total of 26 patients with stage IV gastric cancer (metastatic disease n=25), ECOG performance status 0-3 and laparoscopically evaluated peritoneal status received a modified EAP schedule to prove chemosensitivity and to mobilize autologous peripheral blood stem cells (aPBSC). Patients without progressive disease proceeded to tandem high-dose chemotherapy (HD-CT) and aPBSCT. Patients with >50% reduction of the target lesion received a second cycle of HD-CT. Responders were selected for local R0 resections (D2 resection) according clinical criteria. Of 26 patients, 20(77%) achieved partial remission after dose-intensive chemotherapy: local R0 resection was achieved in 12 out of 14 patients selected for surgery (46% of all patients). Eight of these R0-resected patients initially had peritoneal carcinomatosis. With a median follow-up of 3.2 years, four patients are still alive. The median overall survival was 8.4 months (CI 2.5-14.4 months), for histologic regression grade 3 (seven out of 25 patients, 28%) 29 months (CI 12-46 months). The combined treatment approach is tolerable and feasible in advanced disease and opens a therapeutic window for a significant proportion of patients, even in cases with histologically proven peritoneal carcinomatosis.

Autologous tandem transplantation: almost complete reduction of neutropenic fever following the second transplantation by ex vivo expanded autologous myeloid postprogenitor cells

Reduced post-transplant performance status because of infectious complications is still a problem following autologous peripheral blood stem cell transplantation (aPBSCT). In this study, a tandem transplantation scheme for 15 patients with breast cancer including etoposide (1500 mg/m(2)), ifosfamide (12 g/m(2)) and carboplatin (1500 mg/m(2)) as conditioning regimens, followed by aPBSCT, was used to evaluate the potential clinical benefit of the additional retransfusion of low numbers of ex vivo expanded committed myeloid postprogenitor cells (PPCs) (median 408 x 103 CFU-c/kg BW, range 0.93-1995) following the second transplantation. Following a 7+2 days expansion (using recombinant human SCF, IL-1beta, IL-3, IL-6 + G-CSF), CFU-c generated from CD34-positive cells from leukapheresis products could be expanded by a median factor of 153 (range 5-434). Flow cytometric analysis and morphology of CFUs have shown a nearly exclusive expansion and differentiation of committed myeloid progenitor cells and a significant reduction of CD34-positive cells. In an intra- and interindividual comparison it could be shown that the retransfusion of committed myeloid postprogenitor cells significantly accelerates myeloid recovery. Although retransfusion of PPCs fails to abrogate severe neutropenia following aPBSCT, it significantly ameliorates infectious complications and shortens the duration of hospital stay.

Activation of lymphocytes inhibits human monocyte to macrophage differentiation

Tissue macrophages (MAC) differentiate from circulating blood monocytes (MO) during a maturation step that is of crucial importance for their functional competence. In vitro a similar process of maturation can be observed, if MO are cultured in the presence of serum. In the work presented here, we show that activated lymphocytes can interfere with MAC differentiation. Resting lymphocytes have only marginal influence upon MO to MAC transition in vitro. However, if cells are activated by the lectins PWM or ConA or by double-stranded RNA (polyinosinic-polycytidylic acid, pI:C), normal MAC maturation is suppressed: MO stay small and do not acquire MAC maturation-associated surface molecules like carboxypeptidase M (CPM, determined by antibody MAX.1) or CD84 (determined by antibody MAX.3). This phenomenon can be induced by small numbers of lymphocytes and can be transmitted by soluble factors in cultures stimulated with ConA or PWM. IFN-gamma is present in these conditioned media and partially suppresses MAC maturation but cannot fully substitute for the conditioned media. On the contrary, in pI:C stimulated cultures, suppression of MAC differentiation is dependent on cell-cell contact. In conclusion, activated lymphocytes are able to suppress the terminal differentiation of MAC by several pathways depending on the mode of lymphocyte stimulation.

The septal mucosa decongests with naphazoline: a study of mucosal dynamics with sonography

A new method using B-mode and power-Doppler-mode (pD) sonography for the investigation of changes in nasal mucosa swelling and perfusion was developed. The effect of naphazoline (0.25 mg/mL) on the nasal mucosa was visualized and recorded in 1-minute intervals in 40 patients. The effect of normal saline solution was studied in 27 healthy volunteers. The decongestant and normal saline were applied by flooding the anterior nasal cavity. A computer program automatically quantified pD color information. Normal saline solution induced a 4.8 +/- 2.4% increase in perfusion (+/- SEM, n.s.) after 5 minutes. In the naphazoline group, changes in stereometry were measured on B-mode-sequences in 24 (60%) and perfusion changes in 24 participants (60%). In 16 of 40 patients (40%), both stereometry and perfusion were analyzed. After 10 minutes, the septum and inferior turbinate mucosa thickness were reduced by 17 +/- 2.8% (p < 0.001) and 25 +/- 2.6% (p < 0.001). Perfusion of the septum and inferior turbinate mucosa as visualized with pD-sonography decreased by 33 +/- 3.3% (p < 0.001). The reduction of bloodflow induced by naphazoline as visualized with pD-sonography is within the range of perfusion changes found in LDF and Xenon clearance studies. Decongestion of the septum mucosa demonstrates erectile properties of the septum, which may contribute to the increase of nasal patency after nasal decongestion.

Transplant characteristics: minimal residual disease and impaired megakaryocytic colony growth as sensitive parameters for predicting relapse in acute myeloid leukemia

Dose escalation during consolidation therapy of de novo AML, including myeloablative chemotherapy supported with autologous peripheral blood stem cell transplantation (aPBSCT), continuously improved outcome. Therefore, quality control of transplants is getting increasing interest. We studied leukapheresis products (LPs), consecutively collected during postremission treatment of 20 patients with de novo AML for minimal residual disease (MRD) by 5-parametric flow cytometry and for myelodysplasia (MDS)-associated alterations by paired lineage-selected colony assays for colony-forming units-megakaryocytes (CFU-mega) and burst-granulocytes-monocytes colony-forming units (CFU) to evaluate the predictive value of these transplant-associated parameters on outcome. We defined the leukemia-associated immunophenotype at diagnosis and studied the impact of MRD detection in LPs collected after double induction with TAD (thioguanine, daunorubicin, cytarabine) and HAM (mitoxantrone, high-dose cytarabine, n=18 patients) and TAD consolidation treatment (n=20 patients) on outcome after aPBSCT. The level of MRD in the transplants correlated with the relapse-free survival (RFS) using a cut-off level of 1 x 10(-3) residual leukemic cells. The median RFS was 6 months for the group with > or = 1 x 10(-3) residual leukemic cells and has not been reached in the group with low MRD levels (< 1 x 10(-3)). By using the same cut-off level a weak correlation could also be demonstrated between MRD in the pregraft bone marrow and RFS (P = 0.04). Quantitatively abnormal megakaryocytic colony growth in the back-up LPs collected after double induction and in the transplant LPs was characterized by the ratio CFU-mega/CFU. In the group of relapsing patients the ratio CFU-mega/CFU was significantly lower than in the group of patients with CCR (P = 0.004), both in the back-ups and in the transplants. All patients with CFU-mega/CFU ratios < 0.12 relapsed, five of seven patients developed MDS before progressing to full leukemic relapse. Using the optimized cut-off level for the ratio CFU-mega/CFU (< vs > or = 0.12), seven of 10 relapsing patients (70%) could be identified to be at risk of relapse, whereas MRD in the transplants identified only 50% of the relapses and MRD in the pregraft bone marrow 25%. In conclusion, the study could identify two pretransplant risk factors predicting relapse in patients with AML receiving aPBSCT in first CR: MRD in transplants as well as MDS-like alterations within the transplants. These results may have multifold implications on the design of risk-adapted chemotherapy as well as on purging techniques and may contribute to a better understanding of leukemogenesis.

Whole blood analysis of reticulated platelets: improvements of detection and assay stability

The flow cytometric analysis of reticulated platelets based on the fluorescent derivatization of their RNA content is increasingly used for the diagnostic classification of patients with thrombocytopenia as well as the monitoring of thrombopoiesis recovering under therapy. Many different modifications of the analytical protocol have been published following the first description in 1990 but consensus on the method has not yet been established. We have now reevaluated the assay's methodology in order to optimize sensitivity and specificity and reduce the time length of incubation and washing procedures. In the modified experimental approach native whole blood is incubated for 15 min with an increased amount of thiazole orange (1 microg/ml) in the presence of phycoerythrin labeled antibodies directed against the constitutively surface expressed antigen GPlb. Data acquisition on the flow cytometer can be started immediately after stopping and stabilization of the reaction by paraformaldehyde fixation. Thiazole orange fluorescence was not significantly changed in thrombin-activated, degranulated platelets compared to resting platelets indicating no significant non-specific staining of platelet granules under the selected test conditions. In addition, experiments employing RNAse digestion demonstrated specificity of thiazole orange staining for platelet RNA.

G-CSF-mobilized peripheral blood progenitor cells: in vitro growth pattern and hematopoietic growth factor receptor profile

The kinetics of colony formation by granulocyte colony-stimulating factor (G-CSF)-mobilized peripheral blood progenitor cells (PBPCs) were monitored using clone-mapping experiments. Compared with normal resting bone marrow (BM), where the ratio of Day 7:Day 14 granulocyte-macrophage colony-forming cells (GM-CFCs) was 1:0.76-1.9, PB was found to be relatively deficient in progenitor cells with the capacity to form colonies by Day 7 (median ratio Day 7:Day 14 1:21). The most mature Day 7 GM-CFCs, those dispersing or extinguishing before Day 14, were almost absent in PB (< 1% of all GM-CFCs) but comprised 77% of Day 7 GM-CFCs and 32% of all GM-CFCs in BM. The expression patterns of high affinity receptors for G-CSF, GM-CSF, stem cell factor (SCF), and the ligand for flk-2 on CD38hi and CD38-/dim PB CD34+ cells were determined by binding of 125I-labeled ligand and autoradiography. G-CSF receptor (G-CSFR) expression was detected on approximately 25% of CD38-/dim cells (estimated mean 105 receptors per positive cell) and was higher in CD38hi cells (approximately 50% positive, with a mean of 227 receptors per cell). GM-CSFR expression was low (approximately 25% of cells positive, mean of 120 receptor per cell) and did not vary with CD38 expression. c-kit (SCFR) and flk-2 were expressed by > or = 90% and > or = 80% of CD34+ cells, respectively. SCF binding per cell was greater in the CD38hi population, while flk-2 expression did not vary with CD38 expression. These results confirm the heterogeneity of receptor expression by progenitor cells and imply differential regulation of receptor expression during maturation.