1
|
Zhang P, Tu S. MGAE-DC: Predicting the synergistic effects of drug combinations through multi-channel graph autoencoders. PLoS Comput Biol 2023; 19:e1010951. [PMID: 36867661 PMCID: PMC10027223 DOI: 10.1371/journal.pcbi.1010951] [Citation(s) in RCA: 7] [Impact Index Per Article: 7.0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/03/2022] [Revised: 03/20/2023] [Accepted: 02/14/2023] [Indexed: 03/04/2023] Open
Abstract
Accurate prediction of synergistic effects of drug combinations can reduce the experimental costs for drug development and facilitate the discovery of novel efficacious combination therapies for clinical studies. The drug combinations with high synergy scores are regarded as synergistic ones, while those with moderate or low synergy scores are additive or antagonistic ones. The existing methods usually exploit the synergy data from the aspect of synergistic drug combinations, paying little attention to the additive or antagonistic ones. Also, they usually do not leverage the common patterns of drug combinations across different cell lines. In this paper, we propose a multi-channel graph autoencoder (MGAE)-based method for predicting the synergistic effects of drug combinations (DC), and shortly denote it as MGAE-DC. A MGAE model is built to learn the drug embeddings by considering not only synergistic combinations but also additive and antagonistic ones as three input channels. The later two channels guide the model to explicitly characterize the features of non-synergistic combinations through an encoder-decoder learning process, and thus the drug embeddings become more discriminative between synergistic and non-synergistic combinations. In addition, an attention mechanism is incorporated to fuse each cell-line's drug embeddings across various cell lines, and a common drug embedding is extracted to capture the invariant patterns by developing a set of cell-line shared decoders. The generalization performance of our model is further improved with the invariant patterns. With the cell-line specific and common drug embeddings, our method is extended to predict the synergy scores of drug combinations by a neural network module. Experiments on four benchmark datasets demonstrate that MGAE-DC consistently outperforms the state-of-the-art methods. In-depth literature survey is conducted to find that many drug combinations predicted by MGAE-DC are supported by previous experimental studies. The source code and data are available at https://github.com/yushenshashen/MGAE-DC.
Collapse
Affiliation(s)
- Peng Zhang
- Department of Computer Science and Engineering, Center for Cognitive Machines and Computational Health (CMaCH), Shanghai Jiao Tong University, Shanghai, China
| | - Shikui Tu
- Department of Computer Science and Engineering, Center for Cognitive Machines and Computational Health (CMaCH), Shanghai Jiao Tong University, Shanghai, China
| |
Collapse
|
2
|
Preuer K, Lewis RPI, Hochreiter S, Bender A, Bulusu KC, Klambauer G. DeepSynergy: predicting anti-cancer drug synergy with Deep Learning. Bioinformatics 2018; 34:1538-1546. [PMID: 29253077 PMCID: PMC5925774 DOI: 10.1093/bioinformatics/btx806] [Citation(s) in RCA: 232] [Impact Index Per Article: 38.7] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/09/2017] [Revised: 12/06/2017] [Accepted: 12/14/2017] [Indexed: 12/29/2022] Open
Abstract
Motivation While drug combination therapies are a well-established concept in cancer treatment, identifying novel synergistic combinations is challenging due to the size of combinatorial space. However, computational approaches have emerged as a time- and cost-efficient way to prioritize combinations to test, based on recently available large-scale combination screening data. Recently, Deep Learning has had an impact in many research areas by achieving new state-of-the-art model performance. However, Deep Learning has not yet been applied to drug synergy prediction, which is the approach we present here, termed DeepSynergy. DeepSynergy uses chemical and genomic information as input information, a normalization strategy to account for input data heterogeneity, and conical layers to model drug synergies. Results DeepSynergy was compared to other machine learning methods such as Gradient Boosting Machines, Random Forests, Support Vector Machines and Elastic Nets on the largest publicly available synergy dataset with respect to mean squared error. DeepSynergy significantly outperformed the other methods with an improvement of 7.2% over the second best method at the prediction of novel drug combinations within the space of explored drugs and cell lines. At this task, the mean Pearson correlation coefficient between the measured and the predicted values of DeepSynergy was 0.73. Applying DeepSynergy for classification of these novel drug combinations resulted in a high predictive performance of an AUC of 0.90. Furthermore, we found that all compared methods exhibit low predictive performance when extrapolating to unexplored drugs or cell lines, which we suggest is due to limitations in the size and diversity of the dataset. We envision that DeepSynergy could be a valuable tool for selecting novel synergistic drug combinations. Availability and implementation DeepSynergy is available via www.bioinf.jku.at/software/DeepSynergy. Contact klambauer@bioinf.jku.at. Supplementary information Supplementary data are available at Bioinformatics online.
Collapse
Affiliation(s)
- Kristina Preuer
- Institute of Bioinformatics, Johannes Kepler University, Linz, Austria
| | - Richard P I Lewis
- Department of Chemistry, Centre for Molecular Science Informatics, University of Cambridge, Cambridge, UK
| | - Sepp Hochreiter
- Institute of Bioinformatics, Johannes Kepler University, Linz, Austria
| | - Andreas Bender
- Department of Chemistry, Centre for Molecular Science Informatics, University of Cambridge, Cambridge, UK
| | - Krishna C Bulusu
- Department of Chemistry, Centre for Molecular Science Informatics, University of Cambridge, Cambridge, UK
- Oncology Innovative Medicines and Early Development, AstraZeneca, Hodgkin Building, Chesterford Research Campus, Saffron Walden, Cambs, UK
| | - Günter Klambauer
- Institute of Bioinformatics, Johannes Kepler University, Linz, Austria
| |
Collapse
|
3
|
Schedule-dependent interactions between perifosine and radiotherapy in prostate cancer. ACTA ACUST UNITED AC 2013. [DOI: 10.1007/s13566-013-0101-z] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.1] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/26/2022]
|
4
|
Grunberg SM, Dugan MC, Greenblatt MS, Ospina DJ, Valentine JW. Phase I/II Trial of Paclitaxel and Vinorelbine in Advanced Non-Small Cell Lung Cancer. Cancer Invest 2009; 23:392-8. [PMID: 16193638 DOI: 10.1081/cnv-67137] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.1] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/03/2022]
Abstract
Chemotherapeutic combination regimens for advanced non-small cell lung cancer traditionally have been based on platin compounds. However, a mechanistic rationale could lead to effective non-platin combinations. Paclitaxel and vinorelbine are antimicrotubule agents with different mechanisms of action, both of which have single agent activity against non-small cell lung cancer. A Phase I/II trial of paclitaxel Day 1 and vinorelbine Days 1-3 every 21 days was, therefore, performed for patients with Stage IIIB or Stage IV non-small cell lung cancer who had not previously received chemotherapy for metastatic disease. In the Phase I investigation, up to 4 patients were treated at each dose level. The maximum tolerated dose level was found to be paclitaxel 150 mg/m2 IV Day 1 and vinorelbine 16 mg/m2 IV Days 1-3, with dose-limiting toxicities of fatigue, myalgia, and mucositis at higher doses. This dose level was then expanded with an additional 15 patients. Of the 23 patients treated for up to 10 cycles at or near the maximum tolerated dose level (19 patients with paclitaxel 150 mg/m2 IV Day 1 and vinorelbine 16 mg/m2 Days 1-3, and 4 patients with paclitaxel 150 mg/m2 IV Day 1 and vinorelbine 13 mg/m2 Days 1-3), 7 patients achieved partial response and 5 patients achieved minor response. Fatigue, myalgia, peripheral neuropathy, and transient leukopenia were the most common cumulative toxicities seen. The non-platin chemotherapy doublet of paclitaxel and vinorelbine given on this convenient 3-day schedule is worthy of further investigation in the treatment of non-small cell lung cancer.
Collapse
Affiliation(s)
- Steven M Grunberg
- Division of Hematology/Oncology, University of Vermont, Burlington, Vermont, USA.
| | | | | | | | | |
Collapse
|
5
|
Stathopoulos GP, Veslemes M, Georgatou N, Antoniou D, Giamboudakis P, Katis K, Tsavdaridis D, Rigatos SK, Dimitroulis I, Bastani S, Loukides S, Vergos K, Marossis K, Grigoratou T, Kalatzi E, Charalambatou M, Paspalli A, Michalopoulou P, Stoka M, Gerogianni A. Front-line paclitaxel–vinorelbine versus paclitaxel–carboplatin in patients with advanced non-small-cell lung cancer: a randomized phase III trial. Ann Oncol 2004; 15:1048-55. [PMID: 15205198 DOI: 10.1093/annonc/mdh260] [Citation(s) in RCA: 47] [Impact Index Per Article: 2.4] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/12/2022] Open
Abstract
PURPOSE This randomized phase III trial of advanced or metastatic non-small-cell lung cancer (NSCLC) was designed to compare a standard treatment such as carboplatin (CRP)-paclitaxel (PCT) with a new combination, vinorelbine (VRL)-PCT-two agents acting in microtubules. PATIENTS AND METHODS Three hundred and sixty patients (stage IIIa, IIIb and IV) were included and evaluated for response rate, survival and toxicity. Arm A patients were treated with the control combination of CRP 6 AUC and PCT 175 mg/m(2) repeated every 3 weeks for six cycles, and arm B with the investigational combination of VRL 25 mg/m(2) and PCT 135 mg/m(2) repeated every 2 weeks for nine cycles. The patients were well balanced with respect to gender, disease stage and performance status. Arm A received 849 cycles (mean 4.59 per patient) and arm B 951 cycles (mean 5.39 per patient). RESULTS Complete and partial response rates were 45.95% and 42.86% for arms A and B, respectively. Median survival was 11 and 10 months, 1-year survival 42.7% and 37.85% and 2-year survival 10.12% and 19% for arms A and B, respectively. Toxicity was similar in all patients, except for neutropenia, which was significantly greater in arm B. CONCLUSIONS PCT combined with VRL produces similar (non-significant) response rates, survival and toxicity (except for neutropenia, as noted above) to standard CRP-PCT treatment in untreated advanced-stage NSCLC.
Collapse
Affiliation(s)
- G P Stathopoulos
- First Department of Medical Oncology, Errikos Dynan Hospital, Athens.
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
Collapse
|
6
|
Ballestrero A, Montemurro F, Gonella R, Capaldi A, Danova M, Friedman D, Puglisi M, Aglietta M, Patrone F. Dose-dense vinorelbine and paclitaxel with granulocyte colony-stimulating factor in metastatic breast cancer patients: anti-tumor activity and peripheral blood progenitor cell mobilization capability. Breast Cancer Res Treat 2004; 82:185-90. [PMID: 14703065 DOI: 10.1023/b:brea.0000004374.72658.17] [Citation(s) in RCA: 8] [Impact Index Per Article: 0.4] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/12/2022]
Abstract
We studied the safety, activity and peripheral blood progenitor cell mobilizing capability of a dose-dense combination of vinorelbine (VNB) and paclitaxel (PTX) as first-line chemotherapy for patients with metastatic breast cancer (MBC). Forty-three MBC patients were submitted to four cycles of VNB 30 mg/m2 and PTX 175 mg/m2 intravenously, every 2 weeks, as the first induction step of a tandem high-dose chemotherapy program. Granulocyte colony-stimulating factor (G-CSF) 5 microg/kg was administered daily from day +5 to +10 in order to accelerate hematopoietic recovery, or 48h after the last VNB-PTX when a leukapheresis was planned (after the third or fourth cycle). A total of 172 cycles were administered. The mean delivered dose-intensity of VNB and PTX was 14.7 and 86 mg/m2/week, respectively (98% of the planned dose-intensity). The main per-patient toxicities were: peripheral neurotoxicity (G1/2 60%, G3 5%), constipation (G1/2 10%), oral mucositis (G1/2 20%), and asthenia (G1/2 35%). Hematological toxicity was unremarkable, except for anemia with hemoglobin (Hb) values < 10 g/dl (28%), and lymphopenia with lymphocyte counts < 1000/mm3 (28%). Two complete (5.1%) and 24 partial (61.5%) responses were observed in 39 assessable patients, for an overall response rate of 66.6% (95% CI 51.6-80.9). A median of one apheretic procedure (range 1-3) was required to achieve the target number of 6 x 10(6)/kg CD34+ cells. The median number of CD34+ harvested per patient was 15 x 10(6)/kg (range 6.4-36.5). Four cycles of dose dense VNB and PTX showed a favorable toxicity profile, a relevant anti-tumor activity and a high peripheral blood progenitor cell mobilizing activity.
Collapse
|
7
|
Zelek L, Barthier S, Riofrio M, Fizazi K, Rixe O, Delord JP, Le Cesne A, Spielmann M. Weekly vinorelbine is an effective palliative regimen after failure with anthracyclines and taxanes in metastatic breast carcinoma. Cancer 2001; 92:2267-72. [PMID: 11745280 DOI: 10.1002/1097-0142(20011101)92:9<2267::aid-cncr1572>3.0.co;2-q] [Citation(s) in RCA: 120] [Impact Index Per Article: 5.2] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/06/2022]
Abstract
BACKGROUND Currently, there is no gold standard for the treatment of patients with metastatic breast carcinoma who have experienced failure with anthracyclines and taxanes. A biologic rationale suggests that the mechanism of taxane resistance could be because of an excess of depolymerized tubulin that could enhance sensitivity to vinorelbine. The objective of the study was to assess the tolerance and efficiency of weekly vinorelbine in metastatic breast carcinoma after failure with taxanes. METHODS Patients with measurable disease, a World Health Organization performance status of less than 3 and a life expectancy longer than 3 months were eligible. Persistent taxane-induced neuropathy higher than Grade 1 was an exclusion criterion. The initial planned dose was 30 mg/m(2)/week on an outpatient basis without granulocyte colony-stimulating factor (G-CSF). Neutrophil and platelet counts of 1.0 and 80 g/L, respectively, were required before each new injection; otherwise vinorelbine was delayed for 7 days with a dose reduction of 5 mg/m(2) at the second episode. The dose also was reduced if Grade 3 or 4 toxicity occurred. If the adverse event persisted or if the delay exceeded 14 days between 2 injections given at a dose of 20 mg/m(2), vinorelbine was definitively discontinued. RESULTS Between November 1997 and March 1999, 40 patients with a median age of 49 (range, 39-69) were enrolled. All of them had previously received anthracyclines and taxanes. Because of the delays in neutrophil recovery, the median dose intensity did not exceed 22.5 mg/m(2)/week (range, 11.25-30), and the initial planned dose of 30 mg/m(2)/week appeared unfeasible without G-CSF. The starting dose therefore was 25 mg/m(2)/week after the first 6 patients. Neutropenia led to fever in only three patients. Other severe toxicities were Grade 2-3 neuropathy (n = 5), Grade 2-3 ileus (n = 7), Grade 3 anemia (n = 4), and Grade 3 sepsis (n = 1). Objective responses were observed in 10 of 40 patients (25%), 7 of whom had visceral metastases and 4 who were refractory to taxanes (including 2 patients with liver involvement > 50%). The median time to failure was 6 months (range, 4-12) for responding patients. Disease stabilization was achieved in 9 patients (23%) for a median duration of 5 months (range, 4-6). The median survival duration for the whole population was 6 months (range, 2-18+). CONCLUSIONS Weekly vinorelbine is an active salvage therapy for metastatic breast carcinoma after failure with anthracyclines and taxanes, even in patients with taxane-refractory metastatic breast carcinoma. This confirms that vinorelbine and taxanes are not cross-resistant.
Collapse
Affiliation(s)
- L Zelek
- Department of Medicine, Institut Gustave-Roussy, Villejuif, France.
| | | | | | | | | | | | | | | |
Collapse
|
8
|
Abstract
As combinations and sequences of anthracyclines and taxanes increasingly become standard adjuvant treatment for early breast cancer, a major need for new treatment options for metastatic breast cancer will arise. Vinorelbine is highly active in the treatment of metastatic breast cancer, both as a single agent and in combination regimens. Furthermore, it is well tolerated, with a low incidence of subjective toxicities. It is anticipated, therefore, that vinorelbine will become increasingly utilized for treating metastatic breast cancer due to its favorable safety profile, good tolerability, and promising results in combination with other chemotherapy agents. Combinations with trastuzumab and newer molecular targeting agents are being explored. Doublets or triplets of vinorelbine with drugs other than anthracyclines and taxanes could be considered in the next generation of adjuvant and neoadjuvant trials, where it is anticipated that anthracycline/taxane combinations are likely to replace anthracycline/cyclophosphamide combinations as the mainstay of adjuvant treatment.
Collapse
Affiliation(s)
- G H Domenech
- Sylvester Cancer Center, University of Miami School of Medicine, Miami, FL, USA.
| | | |
Collapse
|
9
|
Konecny G, Untch M, Slamon D, Beryt M, Kahlert S, Felber M, Langer E, Lude S, Hepp H, Pegram M. Drug interactions and cytotoxic effects of paclitaxel in combination with carboplatin, epirubicin, gemcitabine or vinorelbine in breast cancer cell lines and tumor samples. Breast Cancer Res Treat 2001; 67:223-33. [PMID: 11561768 DOI: 10.1023/a:1017980411398] [Citation(s) in RCA: 26] [Impact Index Per Article: 1.1] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/12/2022]
Abstract
The purpose of this study was to analyze the drug interactions of paclitaxel (PTX) with epirubicin (EPI), carboplatin (CBDCA), gemcitabine (GEM) and vinorelbine (VIN) in human breast cancer cells and compare the cytotoxic activity of each drug combination in primary breast cancer samples. These experiments were intended to identify the most active agents in combination with PTX, and to provide a preclinical rational for future clinical investigations in breast cancer. Multiple drug effect/combination index (CI) isobologram analysis was applied to combinations of PTX with either CBDCA, EPI, GEM or VIN in MCF-7, MDA-MB-231 and SK-BR-3 human breast cancer cell lines. Drug concentrations were limited to the ranges achievable in humans in vivo, and the drugs were applied simultaneously at fixed molar ratios for each drug combination. Interactions were assessed at multiple effect levels (IC10-IC90). Additionally, the cytotoxic activity of these combinations was assessed in tumor samples of 50 primary breast cancer patients, utilizing the ATP-tumorchemosensitivity assay (ATP-TCA). Drug interactions were shown to be strongly dose-related in the human breast cancer cell lines investigated. At clinically relevant concentrations, CBDCA/PTX demonstrated synergistic (MCF-7) or additive (MDA-MB-231, SK-BR-3) interactions, and EPI/PTX showed additive (SK-BR-3, MCF-7) and antagonistic (MDA-MB-231) interactions. GEM/PTX and VIN/PTX, however, demonstrated antagonism over multiple dose effect levels at clinically relevant drug concentrations in all three cell lines tested. At plasma peak concentrations, EPI/PTX, CBDCA/PTX, GEM/PTX and VIN/PTX achieved > or = 90% tumor growth inhibition in 93, 86, 63 and 50%, respectively, of primary breast cancer samples investigated with the ATP-TCA. Cumulative dose-response plots of primary breast cancer tumor cells responding in vitro with > or = 90% growth inhibition showed a strong dose dependence for both EPI/PTX and CBDCA/PTX. In conclusion, the current data indicate favorable drug interactions for CBDCA/PTX at clinically relevant drug concentrations in breast cancer cells, and demonstrate superior in vitro cytotoxicity of EPI/PTX and CBDCA/PTX compared to GEM/PTX and VIN/PTX in primary breast cancer cultures.
Collapse
Affiliation(s)
- G Konecny
- Department of Medicine, University of California, Los Angeles School of Medicine, 90095-1678, USA.
| | | | | | | | | | | | | | | | | | | |
Collapse
|
10
|
Ibrahim NK, Buzdar AU, Valero V, Dhingra K, Willey J, Hortobagyi GN. Phase I study of vinorelbine and paclitaxel by 3-hour simultaneous infusion with and without granulocyte colony-stimulating factor support in metastatic breast carcinoma. Cancer 2001. [DOI: 10.1002/1097-0142(20010215)91:4<664::aid-cncr1049>3.0.co;2-p] [Citation(s) in RCA: 9] [Impact Index Per Article: 0.4] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/07/2022]
|
11
|
Zoli W, Ricotti L, Tesei A, Barzanti F, Amadori D. In vitro preclinical models for a rational design of chemotherapy combinations in human tumors. Crit Rev Oncol Hematol 2001; 37:69-82. [PMID: 11164721 DOI: 10.1016/s1040-8428(00)00110-4] [Citation(s) in RCA: 55] [Impact Index Per Article: 2.4] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 01/09/2023] Open
Abstract
Today, drug combinations are frequently used in the treatment of cancer to increase therapeutic efficacy. Currently used clinical protocols for cancer combination therapies are mainly obtained empirically or on the basis of results from previous clinical trials. Information obtained from clinical protocols is invaluable, but it is time-consuming, expensive and does not provide data on the biochemical and molecular mechanisms of interaction of the drugs used in combination treatments at cellular level. Therefore, in vitro drug combination studies on established cell lines or primary cell cultures play an important role in designing and optimising combination protocols. A variety of in vitro assays and different mathematics models have been developed to investigate cytotoxic effects and to analyse the type of drug interactions. Increased knowledge of the cellular targets of traditional and new drugs and the development of new technologies have resulted in a new role for the in vitro tests which are no longer used only to evaluate the cytotoxic effects of drugs, but also to investigate the interference on cell cycle, induction of apoptosis and molecular or biochemical interactions. A review on in vitro preclinical tests used to evaluate the effects of drug combinations and to design the rationale of combined chemotherapy protocols is presented.
Collapse
Affiliation(s)
- W Zoli
- Divisione di Oncologia Medica, Ospedale G.B. Morgagni-L. Pierantoni, viale Forlanini 34, 47100 Forlì, Italy.
| | | | | | | | | |
Collapse
|
12
|
Shah MA, Schwartz GK. The relevance of drug sequence in combination chemotherapy. Drug Resist Updat 2000; 3:335-356. [PMID: 11498402 DOI: 10.1054/drup.2000.0165] [Citation(s) in RCA: 37] [Impact Index Per Article: 1.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/18/2022]
Abstract
The concept of combining chemotherapeutic agents to increase the cytotoxic efficacy has evolved greatly over the past several years. In the past, the rationale for combination chemotherapy centered on attacking different biochemical targets, overcoming drug resistance in heterogenous tumors, and increasing the dose-density of combination chemotherapy to take advantage of tumor growth kinetics. The overall goal was to improve clinical efficacy with acceptable clinical toxicity. It is now apparent that the sequence of drug administration can significantly enhance the therapeutic effect of chemotherapy. These sequence-dependent effects can be explained by chemotherapy-induced cell cycle perturbations, or by pharmacodynamic interactions between the agents in combination. In this review, we focus on drug combinations with taxanes and camptothecins, which we believe best illustrate the importance of the cell cycle and pharmacologic interactions in the sequential administration of chemotherapy. As our understanding of the cell cycle grows, our ability to appropriately sequence chemotherapy can have a great impact on the treatment of human cancers. Copyright 2000 Harcourt Publishers Ltd.
Collapse
Affiliation(s)
- Manish A. Shah
- Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York, USA
| | | |
Collapse
|