The immunomodulatory agents lenalidomide and thalidomide for treatment of the myelodysplastic syndromes: A clinical practice guideline

Critical Reviews in Oncology/Hematology 85 (2013) 162 192 The immunomodulatory agents lenalidomide and thalidomide for treatment of the myelodysplastic syndromes: A clinical practice guideline Heather A. Leitch a,, Rena Buckstein b, April Shamy c, John M. Storring d a Hematology, St. Paul s Hospital and the University of British Columbia, Vancouver, Canada b MDS Program, The Odette Cancer Centre, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, Canada c Hematology, Jewish General Hospital and McGill University, Montreal, Canada d Hematology, Montreal General Hospital and the McGill University Health Centre, Montreal, Canada Accepted 10 July 2012 Contents 1. Target population... 163 1.1. Questions... 163 1.2. Introduction... 163 2. Methods... 164 2.1. Literature search strategy... 164 2.2. Study selection criteria... 165 2.2.1. Inclusion criteria... 165 2.2.2. Exclusion criteria... 165 2.3. Article selection... 165 2.4. Recommendations... 165 3. Results... 165 3.1. Literature search results... 165 3.1.1. Studies evaluating MDS with del5q... 166 3.1.2. Studies evaluating MDS of all or primarily non-del5q karyotype... 174 3.2. Practical aspects of management... 184 3.2.1. Do all lower risk del5q MDS require lenalidomide?... 185 4. Discussion... 185 5. Conclusions... 188 Conflict of interest statement... 188 Reviewer... 189 Appendix A. Members of the Canadian Consortium on Evidence-Based Care in MDS... 189 References... 189 Biographies... 191 Corresponding author at: 440-1144 Burrard Street, Vancouver, BC, Canada V6Z 2A5. Tel.: +1 604 684 5794; fax: +1 604 684 5704. E-mail address: hleitch@providencehematology.com (H.A. Leitch). 1040-8428/$ see front matter 2012 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.critrevonc.2012.07.003

H.A. Leitch et al. / Critical Reviews in Oncology/Hematology 85 (2013) 162 192 163 Abstract Background: Myelodysplastic syndromes (MDS) are clonal disorders that result in cytopenias and risk of acute myeloid leukemia. Incidence increases with age and more diagnoses are expected with the aging population. Treatment includes red blood cell transfusion for anemia. The immunomodulatory agents (imids) thalidomide and lenalidomide may induce transfusion independence. This guideline systematically reviews evidence on imids to treat MDS and makes evidence-based recommendations. Methods: The literature and meeting abstracts were searched for phase 2 3 clinical trials. Data on efficacy, toxicity, and which patients benefit were extracted. Results: 7019 citations on MDS management were identified. Thirteen publications and 9 meeting abstracts met eligibility criteria. Conclusions: Lenalidomide is recommended as first line therapy in lower risk del5q MDS. There is insufficient evidence to recommend lenalidomide for treatment of higher risk del5q MDS or AML, or for any risk non-del5q MDS or AML. Combining lenalidomide with other agents is not recommended. Thalidomide is not recommended. 2012 Elsevier Ireland Ltd. All rights reserved. Keywords: Myelodysplastic syndrome (MDS); Systematic review; Clinical trials; Immunomodulatory agents; Lenalidomide; Thalidomide 1. Target population The Canadian Consortium on Evidence-Based Care in MDS (CCMDS) is preparing a series of evidence-based guidelines addressing the treatment of adult patients with myelodysplastic syndromes (MDS) classified according to the World Health Organization (WHO) classification of tumors of hematopoietic and lymphoid tissue [1] and the French American British (FAB) classification of MDS [2]. These guidelines address patients with MDS of all subtypes, risk categories, comorbidities and performance status. The current guideline considers the use of the immunomodulatory agents (imids) thalidomide and lenalidomide in the treatment of MDS. Del5q MDS and MDS of either all or primarily nondel5q karyotype are addressed in separate sections of this guideline. 1.1. Questions For both thalidomide and lenalidomide 1. In patients with MDS, what is the efficacy of thalidomide or lenalidomide alone or in combination with other agents as measured by response rate (transfusion independence (TI), improvement in transfusion requirements, improvement in hemoglobin level, platelet count or neutrophil count, complete remission [CR] and partial remission [PR]), response duration, time to progression (TTP), overall survival (OS), and quality of life (QOL)? 2. What toxicities/risks are associated with the use of thalidomide and lenalidomide? 3. Which patients are more likely to benefit from treatment with thalidomide or lenalidomide? 1.2. Introduction The myelodysplastic syndromes are a heterogeneous group of clonal bone marrow malignancies characterized by ineffective hematopoiesis resulting in peripheral blood cytopenias and a variable risk of evolving to acute myeloid leukemia (AML). Features of lower risk MDS include an increase in programmed cell death of dysplastic hematopoietic cells, whereas higher risk MDS are characterized by an increased propensity toward clonal evolution in the malignant clone [3]. Resulting clinical manifestations include anemia requiring red blood cell (RBC) transfusion support in many patients. MDS is an acquired disorder primarily manifesting in older adults with a median age at diagnosis of 74. The only potentially curative therapy for MDS is allogeneic hematopoietic stem cell transplantation (SCT) but this procedure is generally limited to younger patients due to prohibitive toxicity. Treatments for the majority of MDS patients have until recent years been limited to supportive measures that aim to improve and optimize quality of life (QOL) and are generally not thought to impact on survival times [4]. The most widely used prognostic scheme in North America remains the International Prognostic Scoring System (IPSS), which risk-stratifies patients based on number of cytopenias, blast count in the marrow and karyotype analysis, allowing patients to be grouped into four categories: low, intermediate-1 (int-1), intermediate-2 (int-2) and high risk, with progressively decreasing predicted survival times and increasing risk of progression to AML [5]. For purposes of management, patients are often grouped into lower risk (low and int-1) in which non-leukemic deaths predominate, and higher risk (int-2 and high) in which leukemic deaths predominate. Prognostic systems for MDS are currently evolving and incorporate new information as it is shown to impact on patient outcome. The recognition of the importance of red blood cell (RBC) transfusion dependence as an adverse prognostic factor for AML evolution and overall survival [6] led to the development of the WHO-based Prognostic Scoring System (WPSS), which incorporates transfusion dependence into calculation of patient risk [7]. Both the IPSS and the WPSS use a limited number of cytogenetic abnormalities to denote karyotypic risk group. However, many additional cytogenetic abnormalities occur in MDS, and their impact on patient risk is addressed in a recently described Comprehensive Cytogenetic Scoring System [8]. The majority of the clinical studies done using imids to treat MDS predated these advances and

164 H.A. Leitch et al. / Critical Reviews in Oncology/Hematology 85 (2013) 162 192 Table 1 Goals of therapy in lower and higher risk myelodysplastic syndromes. IPSS classification Assessment of disease Therapeutic goals Low Lower risk disease Primary goal is improving quality of life Intermediate-1 Median survival measured in years Decrease burden of supportive care (e.g. reduction of transfusion requirements) Lower risk of AML transformation Minimize side effects of therapy Intermediate-2 Higher risk disease Goal is prolongation of survival Median survival measured in months Decrease risk of AML transformation High High risk of AML transformation Alleviate transfusion burden and impact of other cytopenias Increased impact of multiple cytopenias Increased acceptance for adverse effects of therapy Adapted from NCCN guidelines [19]. Abbreviations: AML, acute myelogenous leukemia; NCCN, National Comprehensive Cancer Care Network. used the IPSS to calculate patient risk. Table 1 provides an overview of treatment goals in MDS. MDS with del5q as a sole cytogenetic abnormality occurs predominantly in women, has characteristic morphologic features including a low (<5%) blast count in the marrow, is associated with a favorable course, and is recognized in the WHO 2008 classification as a specific MDS type [1]. The del5q can occur in association with other cytogenetic abnormalities, which confer additional risk. It may also occur in higher-risk MDS, including those with excess blasts and in frank AML. A proportion of lower risk patients experience mitigation of cytopenias, particularly anemia, with the use of growth factors [9] and a minority may respond to immunosuppressive medications such as ATG and cyclosporine A [10]. The majority will eventually develop RBC transfusion dependence, which has been shown in multiple studies to be associated with inferior survival [6,11 13] and QOL [14]. Transfusion dependence may also lead to other complications such as organ dysfunction due to iron overload [6,15,16]. Interventions that can lead to transfusion independence, then, may have a substantial impact on quality of life and possibly survival. While treatment with immunosuppressive therapies has been examined in MDS in addition and shows activity in some patients [17,18], the use of these interventions is the topic of a separate clinical practice guideline currently in preparation, and is not addressed in further detail here. In MDS, the immunomodulatory agents (imids) thalidomide and lenalidomide have been examined as therapeutic interventions. This guideline summarizes data reporting clinical outcomes with the use of lenalidomide and thalidomide in MDS. While recommendations as to their use are made by the National Comprehensive Cancer Network (NCCN) [19] and in other recent reviews [20], this guideline was prepared from a Canadian perspective to guide Canadian decision making. Thalidomide is an immunomodulatory agent with antiangiogenic properties and activity against tumor necrosis factor alpha (TNF- ) and other cytokines. It is a derivative of glutamic acid and its potential activity in MDS may be related to inhibition of TNF- [21], transforming growth factor beta (TGF- ), microvessel density via inhibition of VEGF/VEGF-R [22], or to its activity on both T- and NKcell function [23]. Because of its history of teratogenicity [24] its distribution is subject to significant regulation, which in Canada is administered through the RevAid program [25]. Lenalidomide is a 4-glutamyl analog of thalidomide which is more potent than its parent compound, has fewer side effects and its use is similarly restricted. Lenalidomide appears to have a direct antiproliferative effect in del5q by inducing cell cycle arrest and apoptosis, and possibly through other mechanisms (reviewed in [20]). Both agents have been investigated for the treatment of MDS and the impact of these agents in MDS, singly or in combination with other agents, is the subject of critical appraisal in this review. Pomalidomide, another imid, has not yet been extensively tested for the treatment of MDS. In Canada, lenalidomide is indicated for the treatment of patients with transfusion dependent lower IPSS risk del5q MDS with or without additional cytogenetic abnormalities. Health Canada approval was issued with conditions, pending the results of clinical trials to verify clinical benefit. Thalidomide is not approved by Health Canada for treatment of any MDS. The US Food and Drug Administration approval status for the two drugs is similar. 2. Methods This systematic review was developed by the CCMDS (see Appendix A for membership). Evidence for the current guideline was selected and reviewed by three members of the consortium (HAL, AS, JMS). This review is a summary of the best available evidence on the use of imids in patients with MDS. The evidence in this review is primarily comprised of prospective data from phase 2 clinical trials, though some data from randomized controlled trials (RCT) is available. The systematic review and companion practice guidelines are intended to promote evidence-based practice in Canada. This guideline is editorially independent of federal or provincial health ministries and agencies. 2.1. Literature search strategy The search strategy has been previously reported [26] and is described here briefly. The following databases were searched: Medline (1985 to week 24 [June 17] 2009);

H.A. Leitch et al. / Critical Reviews in Oncology/Hematology 85 (2013) 162 192 165 EMBASE (1985 to week 24 [June 17] 2009); Cochrane Database of Systematic Reviews (2009, Issue 2); and Cochrane Central Register of Controlled Trials (2009, Issue 2). The search was for studies related to the treatment of MDS including chronic myelomonocytic leukemia (CMML) using subject headings and various text words (e.g. myelodysplastic, MDS, chronic myelomonocytic leukemia, CMML, etc.). A filter for systematic reviews, meta-analyses, clinical trials, comparative studies and cohort studies was applied to refine the search results. The search was updated using PubMed in August week 2 [August 9] 2010 using the above search terms and the terms imids, thalidomide and lenalidomide. Conference proceedings from the annual meetings of the American Society of Clinical Oncology (ASCO), American Society of Hematology (ASH), and European Hematology Association (EHA, all 2006 2010) were searched as were proceedings from the International Symposium on MDS (ISMDS, 2007 and 2009), which is held every two years. Abstracts from these meetings reporting trials related to the treatment of MDS were included. 2.2. Study selection criteria 2.2.1. Inclusion criteria Clinical articles reporting prospective phase 2 3 studies or meta-analyses were selected for inclusion in the systematic review provided they were published full reports in the English language of: 1. Studies including adult patients with MDS by FAB or WHO classification and of any IPSS or WPSS risk score. 2. Studies evaluating thalidomide or lenalidomide as either a single agent or in combination with other agents. 3. Comparative trials in which thalidomide or lenalidomide were compared with any agent or combination of agents, placebo or standard of care. Results reporting one of more of the following outcomes: response rate and duration, disease control (time to leukemia, time to progression), overall and progression-free survival, QOL and adverse effects. Meeting abstracts from the ASCO, ASH, EHA and ISMDS were included if they reported unique studies or provided additional informative clinical data for the included clinical trials before or after their publication. In the case of clinical studies that were updated at multiple meetings, the most recent update was generally used unless additional details were provided in earlier updates. 2.2.2. Exclusion criteria Studies were excluded if they were: 1. Comments, books, notes, consensus guidelines, registry or editorial publications. 2. Reporting on fewer than 20 MDS patients in the group of interest, for example with del5q for the section addressing del5q MDS [27], with the exception of studies of particular interest, as detailed in point 3. 3. Studies in which the results of higher-risk MDS patients could not be distinguished from those with acute myelogenous leukemia (AML) by FAB classification or from those with myeloproliferative neoplasms (MPN) by FAB or WHO classification. Exceptions included those reporting unique treatment combinations or treatment for unique groups of patients [28 31]. 2.3. Article selection The initial search was much broader in scope than the current guideline as it was designed to identify all trials and studies investigating the management of MDS. The CCMDS subdivided topics according to treatment agents. An initial screening of search results was conducted by one reviewer who identified potentially relevant studies for inclusion and grouped the results by treatment agent. The citations for thalidomide and lenalidomide were subsequently screened by three reviewers for inclusion. Citations were not blinded during the selection process and each was scored as yes (inclusion criteria were met, no exclusion criteria were met), no (one or more exclusion criteria were met) or maybe (unclear from the citation if the article meets the criteria). Citations scored as yes or maybe were reviewed. Any discrepancies were resolved by consensus. 2.4. Recommendations Evidence-based recommendations were made using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) system [32]. 3. Results 3.1. Literature search results A total of 7019 citations investigating the management of MDS were identified. Following evaluation, 19 database citations were scored as yes or maybe. Of those, 3 were identified as meeting the eligibility criteria for the section on del5q MDS and were included, as were 10 for non-del5q MDS. Of 3 publications satisfying the inclusion/exclusion criteria addressing del5q MDS, all were phase 2 trial design. One was a trial of lenalidomide in lower risk MDS with del5q [33], one was thalidomide in all or unspecified risk MDS with del5q [34], and one examined lenalidomide in higher risk MDS with del5q [29]. Sixteen studies were excluded for the following reasons: could not determine whether patients with del5q were included [28,35 38], no patients with del5q [39 41], retrospective in nature [42], combined previously reported results with additional patients [43], included fewer than 20 patients [44], included fewer than 20 patients with del5q [45 47], review article without original clinical data [48] and reported patients not on clinical trial [49].

166 H.A. Leitch et al. / Critical Reviews in Oncology/Hematology 85 (2013) 162 192 Six meeting abstracts addressing del5q MDS were included; ASH 2009, n =3 [50 52]; EHA 2010, n =1 [53]; ASH 2010, n =2[31,54]. One was phase 3 trial design [54] and was subsequently updated in a full publication [55] and all others were phase 2. Fifty-one meeting abstracts were excluded for the following reasons: subsequently updated in more recent meeting abstracts, n = 8; retrospective data, n = 7; in vitro or ex vivo data, and phase 1 2 data, n =5 each; updated in subsequent peer-review publication, insufficient details provided, fewer than 20 MDS patients, n =4 each; duplicate abstracts, and out of the time window (5 years) for meeting abstracts, n = 2 each; no imid, no response data, could not distinguish outcomes from those of patients with MPN, WHO AML, meta-analysis that could not be critically appraised, not a clinical trial, pharmacoeconomic study, combined patients from previously reported trials, and MDS/MPN overlap syndromes, n = 1 each. All studies included used the IPSS. With this approach, all included trials were those in which the treatment of patients specifically with the del5q was examined. Of 10 publications satisfying the inclusion criteria for the section on non-del5q MDS, all were phase 2 trial design. Two were trials of lenalidomide in lower risk MDS of all karyotype, one primarily non-del5q [46], one entirely nondel5q [40]. Two were trials of thalidomide in primarily lower risk MDS of primarily non-del5q karyotype [36,47], and one was a trial of thalidomide in combination with erythropoietin (EPO) in primarily lower risk MDS of normal karyotype [39]. Three studies were trials of thalidomide in all risk MDS [28,37,38]; del5q was reported in 3 responders of 83 patients in one trial and was not otherwise specified [37]; del5q was reported in 4 of 34 patients in a second trial [38]; and del5q was reported in none of six responders in a third trial, and was not otherwise reported [28]. Two reports were trials of thalidomide in combination with other agents for higher risk MDS. One trial combined thalidomide with arsenic trioxide and retinoic acid in higher risk non-del5q MDS [41], and a second was a trial of topotecan followed by thalidomide; in this trial del5q was present in 3 of 38 (8%) evaluable patients [56]. Three studies were excluded from this section for the following reasons: included more than 20 patients with del5q; [33]; primary focus was del5q and some patients were previously reported in other studies in this guideline [34,47]; fewer than 20 evaluable patients, n = 1 each [35]. Three meeting abstracts were included for non-del5q MDS; ASH 2010, n =1[57]; ASCO 2008, n =1[45]; ASCO 2006, n =1 [58]. All were phase 2 trial design. Fifty-three meeting abstracts were excluded for the following reasons: the focus was del5q MDS, n = 14; subsequently updated in more recent meeting abstracts or peer-review publication, n = 12; fewer than 20 evaluable patients, n = 10; not phase 2 or superior in design, n = 9; pharmacoeconomic analysis, n = 3; retrospective in nature, n = 3; insufficient data for critical appraisal, n = 2; included MDS/MPN and MPN, could not distinguish outcomes from those of patients with MPN, n = 1. All but 2 studies included used the IPSS (not specified in 1 [57], FAB used in 1 [56]). No data appropriate for pooling or meta-analysis were identified. All papers and abstracts detailed below were suitable for critical appraisal. 3.1.1. Studies evaluating MDS with del5q 3.1.1.1. Low and int-1 IPSS risk MDS MDS with del5q. 3.1.1.1.1. Critical appraisal; study design. Of the three reports included in this subcategory, all were multi-center trials, one was phase 3 [54] and pharmaceutical sponsorship was noted in one [33]. Study design and key outcomes are detailed in Table 2. All three trials examined the use of lenalidomide. The MDS-003 multi-center phase 2 trial included 148 patients requiring transfusion of at least 2 units of RBC in the 8 weeks prior to enrollment [33]. Patients were excluded if they had severe neutropenia or thrombocytopenia (neutrophils < 0.5 10 9 /L or platelet count < 50 10 9 /L), CMML with a leukocyte count greater than 12 10 9 /L, treatment-related MDS, hypersensitivity to thalidomide or clinically significant co-morbid illnesses. MDS classification was by FAB criteria and IPSS score was available in 86%. AML was present by FAB criteria in 1 patient and by WHO criteria in an additional 30. Included were patients with the del5q with or without additional cytogenetic abnormalities. Myeloid growth factors for neutropenia were the only cytokines permitted. Predetermined criteria for RBC transfusion were used and the primary endpoint was transfusion independence by International Working Group (IWG) 2000 criteria [59]. Patients received lenalidomide 10 mg either 21 out of 28 days (n = 44), or daily. Response was assessed at predetermined intervals. All 148 patients were included in response and safety analyses by the data and safety monitoring committees, respectively. One pre-planned interim analysis showed favorable response and safety. Median follow-up was 104 weeks and details of lenalidomide discontinuation (in 20% of patients) were provided. The study was sponsored by Celgene Corporation, the manufacturer of lenalidomide, and analysis conducted by the investigators with unrestricted access in consultation with Celgene. A multi-center phase 2 study by Oliva et al. examined QOL, efficacy and safety with lenalidomide 10 mg daily in 49 patients with low or int-1 risk MDS with isolated del5q and a hemoglobin level < 10 g/dl [52]. Outcomes were assessed at predetermined intervals and response evaluated in 20 patients by modified IWG criteria [60]. Withdrawals were reported. QOL was assessed using the QOL-E v.2 questionnaire. Median follow-up was not reported, nor was funding source. The MDS-004 was a multi-center phase 3 randomized, double blind placebo-controlled trial of lenalidomide for lower IPSS risk transfusion dependent MDS with del5q with or without additional cytogenetic abnormalities. Exclusion criteria were: neutrophil count < 0.5 10 9 /L, platelet count < 25 10 9 /L, CMML with a WBC count > 12 10 9 /L, creatinine > 2.0 mg/dl, grade > 2 peripheral neuropathy [61],

Table 2 Clinical trials investigating the effect of immunomodulatory agents for the treatment of lower IPSS risk and all IPSS risk MDS associated with the del5q. Karyotype Study/reference population Lower risk del5q MDS-003 b List 2006 [33] Oliva ASH 2009 [52] MDS-004 Fenaux ASH 2010 ASH 2009 Brandenburg EHA 2010 [50,53 55] Major inclusion criteria FAB 1 MDS del5q c 2 U RBC/8 weeks pre-enrollment Low/int-1 IPSS del5q (isolated) HGB < 10 g/dl Low/int-1 IPSS del5q Lenalidomide naïve RBC transfusiondependent IPSS Phase Agent and starting dose(s) 37% low 44% int-1 5% int-2 or high 14% unclassified 2 Lenalidomide 10 mg 21/28 days or10 mg/day Not stated 2 Lenalidomide 10 mg/day 48% low 52% int-1 66% del5q 28% 1 additional abnormality 3 Crossover allowed after 16 weeks Lenalidomide 10 mg/day vs. 5 mg 21/28 days vs. placebo N ORR a Median response duration 148 At 24 weeks: 112 (76%) 67% RBC TI 9 50% reduction in RBC transfusion requirements 45% CCyR, 28% PCyR 36% mcr BM blasts < 5%, 74% 49 HI-E 85% at both 12 and 24 weeks ITT = 205 mitt = 138 HI-E d M: 41% 5 mg/day 56% 10 mg/day 6% placebo Median time to HI-EM 3.3 weeks, 5 mg/day 4.3 weeks, 10 mg/day CyR: 17% 5 mg/day 41% 10 mg/day P < 0.001 vs. PBO for all (HI-E) NR @ 2y 1 y 62% Median TTF Progression Median OS Quality of life and tolerability Time to AML: 9.3 mo. 5 mg/day 5.9mo 10 mg/day 3.1 mo placebo To more advanced MDS in 8 patients To AMLin 8 patients 34.8% AML at 3y (cumulative risk) Risk of AML reduced by 45% for patients achieving HI-EM Isolated del5q = 7.6 y 1 additional Cy abnormality = 5.6 y 56%@3y Risk of death reduced by 51% for patients achieving HI-E M QOL-E version 2 Physical +34% (P = 0.086) Correlated with improved HGB (P = 0.001) Social +54% (P = 0.021) FACT-An +5.2 on lenalidomide 10 mg/day vs. 3.3, placebo (P = 0.03) at 12 weeks remained significant at weeks 24, 36, 48 Comments Erythroid response influenced by: PLTS < 100 10 9 /L (P = 0.003) and RBC transfusion 4 U/8 weeks (P = 0.01) Median time to TI 4.6 weeks; median HGB rise from baseline 5.4 g/dl H.A. Leitch et al. / Critical Reviews in Oncology/Hematology 85 (2013) 162 192 167

168 H.A. Leitch et al. / Critical Reviews in Oncology/Hematology 85 (2013) 162 192 Comments Median TTF Progression Median OS Quality of life and tolerability N ORR a Median response duration IPSS Phase Agent and starting dose(s) Table 2 (Continued ) Karyotype Study/reference Major population inclusion criteria del5q:12 months if HI-EM; 5.6 months if HI-Em Non-del5q: 12 months if HI-EM; 5.7 months if HI-Em (Mean values reported) del5q: ORR a,e = 37% HI-EM 20% HI-Em 17% Non-del5q: ORR = 32% HI-EM, 18%; HI-Em, 14% 120 del(5q) in 24 (20%) Thalidomide 50 800 mg/day 93% low/int-1 2 (2 trials reported concurrently) <10% blasts RBC transfusion dependent All risk del5q Kelaidi 2008 [34] Abbreviations: abn, abnormality; AML, acute myeloid leukemia; ASH, American Society of Hematology; del, deletion; BMR, bone marrow resolution (of dysplasia); Cy, cytogenetic; DVT, deep venous thrombosis; E, erythroid; EHA, European Hematology Association; EPO, erythropoietin; ESA erythropoiesis stimulating agent; FAB, French-American-British; HGB, hemoglobin; HI, hematologic improvement; int, intermediate; IWG, International Working Group; M, major; m, minor; MDS, myelodysplastic syndrome; mo, months; N, number; NR, not reached; ORR, overall response rate; OS, overall survival; PBO, placebo; TD, transfusion dependent; TI, transfusion independent; TTF, time to treatment failure; QOL, quality of life; R, response; RBC, red blood cell; U, units; y, years. a IWG 2000 criteria unless otherwise stated. b Exclusion: ANC < 0.5 10 9 /L; PLTS < 50 10 9 /L; CMML with WBC > 12 10 9 /L); t-mds; hypersensitivity to thalidomide; significant co-morbidity. Myeloid growth factors permitted for neutropenia. c Isolated or with additional cytogenetic abnormalities. d By protocol criteria (transfusion independence 26 weeks) and IWG criteria. e IWG 2006 criteria. prior treatment with lenalidomide, receiving erythropoiesis stimulating agent (ESA), chemotherapy or investigational therapy within 28 days of randomization, or long-acting ESA within 8 weeks [55]. MDS was categorized by FAB subtype and WPSS criteria were provided in addition to IPSS scores. The trial randomized 205 patients to receive placebo, lenalidomide 5 mg or 10 mg daily with responders continuing treatment past 16 weeks. After 52 weeks, a cross-over design allowed non-responders randomized to placebo to receive lenalidomide in an open-label phase and sub-optimal responders were allowed to dose-escalate at specific time points in the trial [50], thus any differences between treatment arms and placebo would be attenuated in an intention to treat (ITT) analysis. One hundred and thirty eight patients were available for a modified ITT (mitt) analysis. The most common reason for exclusion from the mitt population was inadequate bone marrow specimen for central pathology review. For analysis of prognostic factors [54] the lenalidomide 5 mg and 10 mg groups were combined as they had similar baseline characteristics. QOL was measured by the Functional Assessment of Cancer Therapy-Anemia (FACT-An) questionnaire, administered at predefined intervals [53]. Responses were analyzed by both ITT and mitt [50]. Median follow-up at most recent analysis was 36 months for patients randomized to lenalidomide [62]. Study discontinuation from adverse events was reported. The primary endpoint was RBC-TI reported as study-defined criteria i.e. RBC-TI for 26 weeks or longer and by IWG 2000 criteria (RBC-TI lasting 8 weeks or longer) [59]. Editorial and printing support were provided indirectly by Celgene. 3.1.1.1.2. Outcomes. Question 1. In lower risk patients with MDS and del5q receiving lenalidomide, what is the efficacy of lenalidomide as measured by response rate (HI, CR, PR, TI) and response duration, disease control (TTP or LFS), OS and QOL? Response rate. Response rates by IWG 2000 criteria were reported in all three trials. In the MDS-003 study [33] the overall response rate (ORR) at 24 weeks was 76% and 67% of patients achieved RBC-TI, with an additional 9% having a 50% or more reduction in RBC transfusion requirements. The time to RBC-TI was a median of 4.6 (range 1 49) weeks, median peak hemoglobin in RBC-TI patients 13.4 g/dl and median hemoglobin rise from baseline 5.4 g/dl. Complete cytogenetic response (CCyR) was seen in 38% with partial cytogenetic response (PCyR) in an additional 28%. A complete bone marrow response (resolution of dysplasia) was seen in 36% with bone marrow blasts reduced to less than 5% in 74% of subjects with excess blasts at baseline, and ring sideroblasts reduced to less than 5% from a median of 40% in 64% of evaluable patients with RARS. The Oliva study reported hematologic improvement in the erythroid series (HI-E) in 85% of patients and mean hemoglobin increased from 8.6 to 11.1 g/dl at 12 weeks. A PCyR was seen in 5 of 13 patients at 24 weeks [52]. In the phase 3 MDS-004 trial, HI-E major (M) was seen in significantly more patients receiving lenalidomide than

H.A. Leitch et al. / Critical Reviews in Oncology/Hematology 85 (2013) 162 192 169 placebo; 41%, 52% and 6% in lenalidomide 5 mg, 10 mg and placebo groups, respectively, at 52 weeks (P < 0.001). The median time to HI-EM was 3.3 weeks in patients receiving lenalidomide 5 mg and 4.3 weeks in those receiving 10 mg. CyR was seen in 17% of the 5 mg group and 41% of 10 mg (P < 0.001 vs. placebo) with CCyR in 11% and 24% of the 5 mg and 10 mg groups, respectively [50]. Results were reported to be similar between the ITT and mitt populations. Response duration. Response duration was reported in the MDS-003 study; the median duration of HI-E was not reached at 2 years (range 8.6 to 89 weeks) with 62% of patients continuing to respond at 1 year or more [33]. Disease control (TTP, LFS). Parameters of disease control were reported mainly in the MDS-003 and -004 studies. In MDS-003 (n = 148), new cytogenetic abnormalities occurred in 24 patients, 13 of whom had no initial CyR. More advanced MDS occurred in 8 patients and AML in an additional 8. In MDS-004 (n = 205), 31% evolved to AML at 3 years and the projected median time to AML progression was 4.1 years. Rates of AML progression were similar across all three arms. In the Oliva study, of 20 evaluable patients, additional cytogenetic abnormalities were observed in 4 of 17 responders. Survival. Survival was evaluated in MDS-003 and -004. In MDS-003, the median survival for patients with an isolated del5q was 7.6 years, and for 1 or more additional cytogenetic abnormalities, median survival was 5.6 years. In MDS-004, 3 year OS was 56.5% [55]. Quality of life. QOL was reported in the Italian phase 2 trial [52] and MDS-004 [53]. In the Oliva study, patients responding to lenalidomide experienced improvement in QOL on the QOL-E v.2 questionnaire (a change of 10 points is considered clinically significant) with a trend seen in physical scores, which increased from a mean of 35 at baseline to 69 at week 24 (P = 0.086), and this improvement correlated with improvement in the hemoglobin (r = 0.768, P = 0.001). Social scores increased significantly from a mean of 29 at baseline to 83 at week 12 (P = 0.021). In MDS-004, QOL was reported separately and FACT- An scores were available at baseline and week 12 in 71% of patients. At week 12, there was a significant improvement in QOL with lenalidomide 10 mg vs. placebo with a mean change in score of 5.2 vs. 3.3, respectively (P = 0.03) and a trend toward improvement with lenalidomide 5 mg (P = 0.09). In patients randomized to receive lenalidomide, the change in scores remained significant at 24, 36 and 48 weeks. Question 2. In low and int-1 risk patients with MDS and del5q receiving lenalidomide, what is the tolerability/toxicity associated with the use of lenalidomide (cytopenias, dose adjustments, rate of infections, rate of bleeding, study discontinuation, treatment related mortality)? Grade 3 4 adverse events were reported in all three trials (reported as significant in the Oliva trial), and most were related to neutropenia (absolute neutrophil count [ANC] < 1.0 10 9 /L; 55 80%) and thrombocytopenia (platelet count < 50 10 9 /L; 15 44%). The percentage of change in neutrophil and platelet counts were not reported, however, in MDS-003, patients with an ANC < 0.5 10 9 /L or platelet count < 50 10 9 /L were excluded. Neutropenia was accompanied by fever and/or infection in 4%, 10% and 3% in MDS-003, Oliva and MDS-004, respectively. In MDS- 003, 62% of grade 3 4 myelosuppression occurred within the first 8 weeks of treatment. Dose reductions for adverse events were required in 84% of patients in MDS-003, including 91% in the 10 mg treatment group, and the median time to dose reduction was 22 days. Similarly, 80% of patients in the Oliva study required dose reduction within the first 8 weeks of treatment due to significant neutropenia. In MDS-004, dose reduction was required in 52% and 58% of patients in the lenalidomide 5 mg and 10 mg groups, respectively, due to neutropenia without, and with fever, respectively, and thrombocytopenia in 28%, 3% and 12% in the 5 mg arm, and 38%, 23% and 0% in the 10 mg arm. Drug discontinuation was required in 30 patients (20%) in MDS-003 due to adverse events, with 10 due to neutropenia or thrombocytopenia, 1 with anemia, and 1 with pneumonia. One patient withdrew from the Oliva study due to progressive anemia. In MDS-004, discontinuation due to adverse events in the first year of treatment occurred in 16%, 9% and 5% in the 5 mg, 10 mg and placebo arms, respectively. In MDS-003, there were 11 deaths either on treatment or within 30 days of last treatment, and 3 were judged to be possibly treatment-related; all were infection associated with neutropenia. Other adverse events were reported to be generally mild and in MDS-003 included pruritis, rash, diarrhea and fatigue. In MDS-004, deep venous thrombosis (DVT) was reported in 1%, 6% and 2% of patients in the 5 mg, 10 mg and placebo arms. Question 3. In low and int-1 del5q MDS, which patients are most likely to benefit from treatment with lenalidomide (baseline characteristics, remission status, age, AML, IPSS/FAB/WHO)? Baseline characteristics. In MDS-003, hematologic response rate was not influenced by age, gender, FAB type, IPSS or cytogenetic pattern, however, lower cytogenetic response rates occurred in patients with thrombocytopenia (P = 0.02) and age 60 years (P = 0.07) [33]. RBC-TI could be predicted by the development of new thrombocytopenia or neutropenia, however. An analysis of patients with del5q in the MDS-003 trial showed that in the first 8 weeks of treatment, of patients whose platelet count declined by 50%, RBC-TI occurred in 70% compared with 42% of patients in whom the platelet count decreased by <50% (P = 0.01). Similarly, for patients without baseline neutropenia, of patients experiencing an ANC decline of 75%, TI occurred in 82% compared with 51% in patients with an ANC decline of <75% (P = 0.02) [63]. In MDS-004, RBC-TI was not influenced by age, gender, FAB classification, IPSS, time from diagnosis, cytogenetic complexity, baseline platelet count or number of

170 H.A. Leitch et al. / Critical Reviews in Oncology/Hematology 85 (2013) 162 192 cytopenias. However, increased age and ferritin level were reported to impact negatively on overall and AML-free survival (P = 0.004 and 0.014 for age, respectively and P = 0.004 and 0.004 for ferritin level) [54]. Remission status. In MDS-003, disease progression was associated with a lower frequency of CyR (P = 0.05). In MDS-004, achieving RBC-TI 26 weeks or longer impacted on OS and AML-free survival (P = 0.008 and 0.022, respectively). To summarize, predictors of RBC-TI may include age, the occurrence of neutropenia or thrombocytopenia in the first 8 weeks of treatment and RBC transfusion burden. Predictors of OS and LFS include younger age, lower ferritin level, achievement of RBC-TI of 26 weeks or longer, and possibly cytogenetic response and RBC transfusion burden. RBC-TI was associated with improved QOL. 3.1.1.2. Low and int-1 IPSS risk MDS all karyotypes. 3.1.1.2.1. Critical appraisal; study design. Of the eight trials in this subcategory, all were excluded. Three were excluded for reporting on fewer than 20 patients with del5q [45 47]. Of these, one examined the use of lenalidomide as a single agent [46], one lenalidomide in combination with EPO [45] and one thalidomide [47]. Three trials were excluded for not reporting on patients with del5q [39,40,57]. Of these, one examined the use of lenalidomide as a single agent [40], one lenalidomide in combination with ESA in ESA-refractory patients [57] and one thalidomide combined with EPO for thalidomide- or EPO-refractory patients [39]. Two trials were excluded for not specifying whether del5q patients were included [35,36]. Of these, one was a trial of thalidomide [35] and one of thalidomide in ESA refractory patients [36]. 3.1.1.3. All IPSS risk MDS Del5q. 3.1.1.3.1. Critical appraisal; study design. There is one report included in this subcategory [34], a multi-center study of the Groupe Francophone des Myelodysplasies (GFM) including five prospective phase 2 trials as well as patients treated according to guidelines. Study design and key outcomes are detailed in Table 2. This study examined the use of thalidomide and ESA and covers the period from 1998 to 2006. Four hundred and three patients with lower risk MDS received ESA in 3 prospective clinical trials or according to guidelines; these are discussed further in Section 3.2.1. In addition, 120 patients, 24 of whom had del5q, received thalidomide in two consecutive prospective clinical trials. This report is included in the del5q category as these patients were analyzed separately. Inclusion criteria for the thalidomide trials were: <10% blasts in the marrow and RBC transfusion dependence. Patients received 200 800 mg/day thalidomide or 50 400 mg/day in the two trials. Responses were determined by IWG 2000 and 2006 criteria and analyzed by ITT. Thirty percent of patients had intermediate-2 IPSS risk MDS and all others were lower risk. This report examined response rates and drug discontinuations were accounted for. Median follow-up was not reported. 3.1.1.3.2. Outcomes. Question 1. In patients with MDS of all risk and del5q receiving thalidomide, what is the efficacy of thalidomide as measured by response rate and duration, disease control, survival and quality of life? Response rate. Of 24 patients with MDS and del5q, 8 had the 5q-syndrome. Eighteen received at least 12 weeks of thalidomide, while 6 stopped before week 8 because of side effects. According to IWG 2000 criteria, there were 9 responses, 5 major and 4 minor for a response rate of 37% by ITT. By IWG 2006 criteria, there were 7 responses for an ORR of 30%. Of the 8 patients with the 5q-syndrome, there were 4 responses, 1 major and 3 minor by IWG 2000 criteria, and 2 responses by IWG 2006. Responses occurred (IWG 2000) at: 100 mg/day, n = 4; 150 mg/day, n = 1; and 200 mg/day, n =4. Of 2 patients who had cytogenetic analysis performed following treatment with thalidomide, neither had a cytogenetic response. Response duration. By IWG 2000 and 2006 criteria, mean response duration was 9.5 and 11 months, respectively. Time to progression, disease-free survival, overall survival, quality of life. These endpoints were not reported. Question 2. In patients with MDS of all risk and del5q receiving thalidomide, what is the tolerability/toxicity associated with the use of thalidomide? Adverse events were not reported separately in these patients, however, 6 of 24 patients stopped treatment prior to week 8 because of side effects. In other trials of thalidomide, there were significant rates of discontinuation due to adverse events. In one trial, there was a 78% discontinuation rate due to side effects. These included: sedation (72%), constipation (40%), fatigue (25%), dizziness (25%), muscle cramps (12%) and peripheral neuropathy (8%) [35]. In a second report, 33 of 87 patients stopped treatment prior to week 12 because of: constipation, n =7; constipation and somnolence, n = 6; dizziness, fatigue or somnolence, n = 4; thrombosis, n = 4; grade 3 4 neutropenia, n = 3; edema, n = 2; grade 3 thrombocytopenia, n =1 [47]. Question 3. In patients with MDS of all risk and del5q, which patients are most likely to benefit from treatment with thalidomide? Responses have been reported in these patients, with a hematologic response seen in around 30%, and a mean response duration of less than one year. However, unlike with lenalidomide, response rates are not higher than those seen in MDS without del5q, and no cytogenetic responses were seen. 3.1.1.4. All or unspecified IPSS risk MDS All karyotypes. 3.1.1.4.1. Critical appraisal; study design. Of five trials in this subcategory [28,37,38,58,64], all were excluded as the presence of del5q was not specified. Three were trials of thalidomide as a single agent [28,37,38], one of thalidomide in combination with valproic acid [64] and one of thalidomide in combination with 5-azacytidine [58].

H.A. Leitch et al. / Critical Reviews in Oncology/Hematology 85 (2013) 162 192 171 3.1.1.5. Higher IPSS risk MDS and AML del5q. 3.1.1.5.1. Critical appraisal; study design. Four studies [29 31,51] address the treatment of patients with higher risk MDS and AML with del5q using imids, in this case lenalidomide alone or in combination with other agents. One was reported as a full paper and three are available in abstract form. Study design and key outcomes are detailed in Table 3. Ades et al. reported a phase 2 multi-institution trial of the GFM [29]. In this trial, 47 patients with higher risk MDS and del5q received 10 mg lenalidomide 21/28 days for 8 weeks. Inclusion criteria were adult patients with MDS by FAB or WHO criteria, including refractory anemia with excess blasts in transformation (RAEB-T) and CMML with a WBC count < 13 10 9 /L, intermediate-2 or high IPSS risk, and del5q by cytogenetic analysis with or without additional chromosomal abnormalities. Patients were required to have a platelet count > 25 10 9 /L and an ANC > 0.5 10 9 /L. Treatment could be continued for non-responding patients at the same dose or at 15 mg 21/28 days according to prespecified criteria. Responders continued treatment until relapse. Dose delays were prespecified for toxicity and G-CSF use was recommended for febrile neutropenia. RBC transfusion thresholds were prespecified. The primary endpoint was response by IWG 2006 criteria and RAEB-T patient response was also assessed by AML criteria [60]. Secondary endpoints were response duration, progression to AML, overall survival and safety. Analysis was by modified ITT, excluding diagnostic errors and withdrawal of consent. Patients were int-2 or higher risk IPSS in 100% of cases. Drug discontinuations and dose reductions were accounted for and median follow-up was reported. Study drug and a scientific grant were provided by Celgene. Mollgard et al. reported a multi-institution phase 2 study of lenalidomide in escalating doses over a period of 16 weeks [51]. Twenty five patients with higher risk MDS or AML and del5q or monosomy 5 with or without additional chromosomal abnormalities, not eligible for induction chemotherapy were enrolled, including primary cases and patients relapsed or refractory to intensive treatment. Patients received lenalidomide from 10 to 30 mg daily, with 8 weeks on the highest dose and with dose reductions prespecified for toxicity. The primary endpoint was cytogenetic response by fluorescence in situ hybridization (FISH) analysis of the bone marrow at weeks 8 and 16. Response criteria, classification scheme and risk score used were not reported. Fourteen patients had MDS and 11 AML. Drug discontinuations were accounted for but median follow up was not reported. Sekeres et al. reported a phase 2 trial of lenalidomide for patients age 60 years or more not eligible for intensive treatment, with previously untreated del5q AML with or without additional cytogenetic abnormalities [30]. AML was classified by WHO criteria and patients with t(15;17) were excluded. Patients received lenalidomide 50 mg daily for up to 28 days as induction, and those who achieved stable disease or better received lenalidomide 10 mg 21/28 days until progression or prohibitive toxicity. Response was assessed by IWG criteria (2000 vs. 2006 not reported) and alpha and beta values reported. Thirty-seven patients were evaluated. Drug discontinuations were accounted for and median follow-up was not specified. Ades et al. reported a phase 1/2 multi-institution trial from the GFM which evaluated the combination of lenalidomide and intensive treatment in intermediate-2 or high IPSS risk MDS and AML with del5q with or without additional cytogenetic abnormalities [31]. Daunorubicin 45 (n = 31) to 60 mg/m 2 /day days 1 3 (n = 17) combined with cytarabine 200 mg/m 2 /day days 1 7 were combined with lenalidomide 10 mg 21/28 days and G-CSF from day 8 to count recovery. Response was assessed by AML criteria [60] and responders received 6 consolidation cycles of daunorubicin 45 mg/m 2 on day 1 with cytarabine 120 mg/m 2 /day for 5 days and lenalidomide 10 mg days 1 15, followed by maintenance lenalidomide 14 days per month until progression. All patients had either AML (n = 36) or RAEB-2 (n = 12) by the WHO classification. Drug discontinuations and median follow-up were not specified. 3.1.1.5.2. Outcomes. Question 1. In patients with higher risk del5q MDS or AML receiving lenalidomide as a single agent or in combination with other agents, what is the efficacy of lenalidomide with or without other agents, as measured by response rate and duration, disease control, overall survival and quality of life? Response rate. In 47 patients with higher risk MDS and del5q reported by Ades et al. [29], 60% were int-2 and 40% high IPSS risk. Twenty-five patients had received prior MDS treatment, but only 3 had received intensive treatment. Eighteen patients had RAEB-T/AML with 20 30% blasts. Del5q was present as a single abnormality in 19%, with one additional abnormality in 23%, and more than one additional abnormality (median 5, range 2 16) in 58%. Forty-three patients were RBC transfusion dependent. At least 1 complete cycle was administered in 43 patients, with 13 achieving response: 7 CR, 2 marrow CR and 4 HI-E. Twelve patients became RBC transfusion independent, the exception being a patient with RAEB-1 and complex karyotype, who had a response duration of only 3.5 months. Of the 7 CR, 4 had CCyR and 3 PCyR, with an additional CCyR and PCyR in patients with a marrow CR and one with HI-E, respectively, and for an overall cytogenetic response rate of 9 (19%) patients. Only one patient with RAEB-T/AML had a response (CR). Twelve of 13 patients who responded did so at the first response assessment, after the second cycle of lenalidomide, and the 13th response was noted after 4 cycles. In the Mollgard study of higher risk MDS (n = 14) and AML (n = 11) [51], 22 patients had del5q, 5 with 1 additional abnormality and 9 with complex karyotype. Three patients had monosomy 5, all complex. Twenty started treatment, 7 completed 16 weeks, 6 had a cytogenetic and/or marrow response, and 4 had a hematologic response. Two MDS patients had a major CyR, one with del5q and 1 additional abnormality and 1 with two additional abnormalities. Two additional MDS patients with complex karyotype had

Table 3 Clinical trials evaluating the immunomodulatory agents for the treatment of higher risk MDS and AML with the del5q. Karyotype population Study/reference Higher risk/aml del5q Ades 2009 [29] Mollgard ASH 2009 [51] Sekeres ASH 2010 [30] Ades ASH 2010 [31] Major inclusion criteria Higher IPSS risk del5q RBC-TD n =43 Higher risk MDS del5q or 5 AML del5q Previously untreated Higher risk MDS and AML del5q IPSS Phase Treatment N ORR a Median response duration 40% int-2 60% high 19% del5q 23%+1 abnormality 58% + complex 35% int-2 12% high 54% AML (23% 2, 42% relapsed/refractory) 77% RBC-TD 51% prior MDS Isolated del5q; n =5 (+2 patients del5q by FISH only) +1 abn, n =5 75% AML 25% RAEB-2 Isolated del5q; n =5 +1 abn; n =5 Complex; n =38 2 Lenalidomide 10 mg/day 49 13 (27%) 7 (15%) CHR 2 marrow CR 4 HI-E 12 RBC-TI CCyR + PCyR, n =4+3 2 Lenalidomide 10 30 mg/day 2 Induction: Len 50 mg 21/28 If SD: len 10 mg 21/28 to progression c 2-Jan Induction: DNR 45 60 mg/m 2 d1-3 +ara-c 200 mg/m 2 d1-7 +len 10 mg d1-21 +GCSF from d8 d Consolidation given to those in CR 26 8 (31%) b MCyR, n =4 mcyr, n =2 BM blasts <5%, n =6 HR, n =5 41 17 (46%) b CR/CRi, n =4 (11%) SD, n =13 48 29 (60%) d CR, n =24 CRi, n =1 PR, n =1 e CHR without BMR, n =3 CCyR, n =8 PCyr, n =5 CHR 11.5 months RBC-TI 6.5 (range 2 18) months Median TTF DFS Median OS Of 4 CR/CRi, 3 relapsed at 1, 2 and 4 mo The 4th died at 13 mo w/o relapse Of CR/CRi 14 relapsed + 2 deaths in CR1 26.5% @1y 8.9 months 5.6 in nonresponders 18.4 mo in HR NR @ 10.8 mo in CHR 5.5 mo with 1 add abn 2 months (range 1 4) 30% @ 2 y Comments Factors predictive for response: Isolated del5q PLTS > 100 14 patients (38%) completed induction therapy Abbreviations: AML, acute myeloid leukemia; ara-c, cytosine arabinoside; ASH, American Society of Hematology; del, deletion; BMR, bone marrow resolution (of dysplasia); Cy, cytogenetic; DFS, disease free survivial; DNR, daunorubicin; E, erythroid; HI, hematologic improvement; int, intermediate; IWG, International Working Group; len, lenalidomide; M, major; m, minor; MDS, myelodysplastic syndrome; mo, months; N, number; NR, not reached; ORR, overall response rate; OS, overall survival; PLTS, platelet count ( 10 9 /L); RAEB, refractory anemia with excess blasts; RBC, red blood cell; TD, transfusion dependent; TI, transfusion independent; TTF, time to treatment failure; R, response; RBC, red blood cell; w/o, without; y, years. a IWG 2006 criteria. b Response criteria not specified. c No cytokines or cytotoxic therapy allowed. d For responders, 6 consolidation courses of DNR 45 mg/m 2 d1 + arac 120 mg/m 2 d1-5 + len 10 mg d1-15, then maintenance len 14 days/month until progression. e AML criteria. 172 H.A. Leitch et al. / Critical Reviews in Oncology/Hematology 85 (2013) 162 192