Systematic review of efficacy of dose-dense versus non-dose-dense chemotherapy in breast cancer, non-hodgkin lymphoma, and non-small cell lung cancer

Transcription

1 Critical Reviews in Oncology/Hematology 81 (2012) Systematic review of efficacy of dose-dense versus non-dose-dense chemotherapy in breast cancer, non-hodgkin lymphoma, and non-small cell lung cancer Gary H. Lyman a,, Richard L. Barron b, Jaime L. Natoli c, Ross M. Miller c a Division of Medical Oncology, Department of Medicine, Duke University School of Medicine and the Duke Comprehensive Cancer Center, Durham, NC, USA b Amgen Inc., Thousand Oaks, CA, USA c Cerner LifeSciences, Beverly Hills, CA, USA Accepted 19 April 2011 Contents 1. Background Methods Search strategy and publication selection Data collection and analysis Results Search results and trial characteristics Breast cancer Non-Hodgkin lymphoma Non-small-cell lung cancer Meta-analysis feasibility Discussion Conflict of interest Role of the funding source Reviewers Appendix A. Supplementary data References Biographies Abstract Randomized controlled trials (RCTs) have suggested a potential advantage of dose-dense chemotherapy in improving disease-free and overall survival in patients with certain malignancies. This systematic review summarizes the literature on the efficacy of dose-dense chemotherapy across various cancers (breast cancer, non-hodgkin lymphoma [NHL], and non-small cell lung cancer) and chemotherapy regimens. Among the 17 trials identified, few reported statistically significant differences between dose-dense and standard chemotherapy, and most were small with limited statistical power. Statistically significant differences in overall survival favoring dose-dense schedules were apparent among large RCTs in potentially curative settings such as early-stage breast cancer and NHL. Clinical and treatment heterogeneity demonstrated the Corresponding author. Tel.: ; fax: address: gary.lyman@duke.edu (G.H. Lyman) /$ see front matter 2011 Elsevier Ireland Ltd. All rights reserved. doi: /j.critrevonc

2 G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012) flexibility of the dose-dense paradigm but also precluded quantitative meta-analysis of results. Further study of dose-dense schedules based on large RCTs is needed to demonstrate the consistency and generalizability of these findings Elsevier Ireland Ltd. All rights reserved. Keywords: Dose-dense chemotherapy; Breast cancer; Non-Hodgkin lymphoma; Non-small-cell lung cancer 1. Background In most human cancers, tumor cell growth follows a Gompertzian curve, which is characterized by an initial rapid growth of cells followed by a decrease in doubling rate as tumor size increases [1]. The Norton-Simon hypothesis suggests that chemotherapy efficacy can be enhanced by decreasing the interval between treatment cycles [2,3]. By decreasing the interval between treatment cycles, an approach known as dose-dense chemotherapy, cytotoxic agents interrupt this rapid growth phase and limit growth of tumor cells [2,3]. However, the clinical application of dose-dense schedules is limited by adverse events associated with myelosuppression, most notably chemotherapy-induced neutropenia (CIN) [4]. The advent of granulocyte colony stimulating factors (G-CSF) has decreased the incidence of CIN and enabled the delivery of dose-dense schedules [5]. A number of randomized controlled trials (RCTs) have supported the potential advantage of dose-dense chemotherapy enabled by G-CSF support in improving disease-free survival (DFS) and overall survival (OS) in patients with responsive and potentially curable malignancies, such as early-stage breast cancer (ESBC) and non-hodgkin lymphoma (NHL) [6 8]. CALGB 9741 [6] compared a dose-dense, every-two-week schedule of doxorubicin (Adriamycin), cyclophosphamide (Cytoxan), and paclitaxel (Taxol) to a conventional, every-three-week schedule of the same drugs in more than 2000 women with lymph nodepositive ESBC. Utilizing a 2 2 factorial design, CALGB 9741 examined both concurrent (AC-T) and sequential (A-T- C) approaches. Patients randomized to the dose-dense schedule received the same total dose of chemotherapy in 33% less time compared to a conventional, every-three-week schedule. After three years of follow-up, patients receiving dose-dense chemotherapy schedules demonstrated improved DFS (risk ratio [RR] = 0.74, p = 0.010) and OS (RR = 0.69, p = 0.013) compared to patients randomized to standard treatments. Favorable results for dose-dense chemotherapy also have been reported for RCTs in patients with NHL, while other trials including a mix of early-stage and advanced stage patients have provided mixed results [7,8]. A 2010 systematic review evaluated RCTs of patients with solid tumors or lymphoma who were randomized to chemotherapy with or without G- CSF support and that reported both occurrence of secondary malignancies and overall mortality [9]. The six eligible trials comparing dose-dense to standard treatment schedules did not demonstrate an increase in secondary cancers but did show a 16% (95% CI: 9 22%; p < 0.001) relative risk reduction for overall mortality at a median of nearly five years of follow-up in favor of the dose-dense schedule [9]. Nevertheless, it remains unclear as to whether dosedense chemotherapy is universally applicable and beneficial across a variety of cancers and chemotherapy regimens. In an effort to summarize the existing literature on the efficacy of dose-dense chemotherapy, we conducted a systematic review of comparative clinical trials of dose-dense chemotherapy in breast cancer, NHL, and non-small-cell lung cancer (NSCLC), as they represent the most common disease settings for dose-dense regimens [9]. The focus of this review was on the impact of dose-dense regimens on clinical outcomes relating to overall survival, progression, and tumor response. 2. Methods 2.1. Search strategy and publication selection A systematic review of clinical trials was undertaken including a search of Ovid-MEDLINE, CancerLit, EMBASE, and the Cochrane Library to identify Englishlanguage publications on dose-dense cancer chemotherapy published between January 1, 1995 and October 6, Since the nomenclature and terminology for dose-dense chemotherapy schedules are not standardized, a wide variety of terms related to dose-dense chemotherapy (e.g., dense, accelerated, compressed) were included in addition to search terms related to the specific cancers of interest. References provided in relevant clinical trials and review publications also were assessed for additional primary publications not captured by the search strategy. In an effort to capture all relevant trials on dose-dense chemotherapy for the three cancers of interest, no restriction was placed on specific chemotherapeutic agents, drug classes, or regimens. Trials were included if they compared a dose-dense regimen to a standard regimen; trials did not have to have a primary or secondary objective of testing a dosedense hypothesis. Dose-dense chemotherapy was defined as any regimen that decreased the interval between cycles while maintaining the same drugs and total dose per treatment as other regimen(s) in the trial. RCTs that reduced or increased the dose per treatment were excluded, even if the duration between intervals was shortened. No restrictions were made with regard to G-CSF use in any treatment arms (e.g., reactive administration, prophylaxis). Both randomized and non-randomized trials in patients with breast cancer, NHL, or NSCLC were included. Trials were required to report one or more of the following outcomes: OS, progression-free survival (PFS), time to

3 298 G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012) progression (TTP), overall response (OR), complete response (CR), or partial response (PR). Dose-finding and doseescalation trials, trials with regimens that involved peripheral blood stem cell transplantation, and trials conducted in exclusively pediatric populations were excluded. All abstracts were reviewed by one of two independent researchers. The two researchers concurrently reviewed a random sample of 200 abstracts, and inter-rater agreement was assessed using the kappa statistic. This process was repeated until a sufficient level of agreement was achieved (kappa 0.7), and the remaining abstracts were divided between the reviewers. For abstracts thought to contain relevant primary data or secondary information, the full-text publication was obtained and reviewed for inclusion and/or review of references. The final list of included publications was reviewed by all four co-authors, and any uncertainty about a publication s inclusion or exclusion was resolved by discussion and consensus. A flow diagram of the search results and article review process was created as described by the QUOROM guidelines for presentation of results from a systematic review, which states that authors should provide information on the number of trials identified, included, and excluded and the reasons for exclusion of trials [10] Data collection and analysis Relevant data from accepted publications were abstracted for the following data categories: publication information (e.g., authors, country), study methodology (e.g., randomization, phase), study population (e.g., number of patients, age of patients, cancer stage), treatment regimens (e.g., details of dosing and administration for all treatment arms), and outcomes (e.g., definition, time points, values, 95% CI, p values). All data for each trial were abstracted by a single researcher and checked for completeness and accuracy by two other independent researchers. Findings were qualitatively summarized and categorized by cancer type, outcome, and time point. Findings also were assessed to determine feasibility of quantitative meta-analysis. Since duration of follow-up and assessment for timedependent outcomes such as survival and progression varied across trials, data were collected for all reported time points. Therefore, multiple measures for a given outcome in a single trial were collected as necessary to allow comparison for similar time points across trials. We were unable to adjust for censoring since most trials did not report hazard ratios or other outcome measures that permitted such analysis. 3. Results 3.1. Search results and trial characteristics The initial search strategy elicited more than 14,000 unique publications. Of these, 425 abstracts were accepted and 59 additional publications were identified from review of relevant bibliographies. After full-text review, 16 publications [6 8,11 23] representing 15 unique trials were accepted. A QUOROM diagram of the systematic review process that depicts the reasons for inclusion and exclusion criteria of publications (and unique trials) is presented in Fig. 1. An overview of the 15 trials is provided in Table 1. All trials were prospective RCTs with sample sizes varying from 27 to 2005 patients; no non-rcts met the full inclusion criteria for this review. Twelve trials were conducted in Europe, two in the Western Pacific, and one in the Americas. We identified eight trials in patients with breast cancer, four in NHL, and three in NSCLC. Additional details on treatment regimens and dose intensity (e.g., treatment delay and/or dose reduction, relative dose intensity [RDI]) for both dosedense and non-dose-dense regimens are provided in an online supplement Breast cancer A summary of the eight breast cancer RCTs is provided in Table 2 [6,11 18]. Patient populations differed across trials, including four trials in patients with early or node-positive disease [6,12,13,18] and four in patients with advanced or metastatic disease [14 17]. Although most RCTs compared every-three-week versus every-two-week schedules, the specific drugs used varied among trials as did growth factor support and interval modifications. Sample size varied greatly across studies, ranging from as few as 27 patients [16] to more than 2000 subjects [6]. Most trials were underpowered to demonstrate a clinically relevant treatment effect; only two trials included more than 1000 subjects [6,12], and the remaining six trials each enrolled fewer than 200 subjects. Among the four RCTs in patients with early or nodepositive disease, OS was reported by three trials [6,12,13] at multiple time points ranging from 3 to 10 years. These three trials provided nine unique estimates for OS. Although all of the survival data were in favor of the dose-dense treatment arm, only one trial reported a statistically significant difference [6]. The largest of the trials, CALGB 9741 [6], found an improvement in four-year survival for two-weekly ATC therapy compared to three-weekly therapy (sequential and concurrent groups combined, RR = 0.69, p = 0.013). One trial [13] provided results for PFS but did not find a statistically significant difference between the two treatment arms (five-year PFS, 56% for dose-dense CEF/CMF q2w, 52% for conventional CEF/CMF q3w, p = 0.30). Response rates were reported by two of the trials in patients with early disease [13,18], neither of which found a statistically significant difference between dose-dense and standard schedules. Among the four trials in patients with advanced or metastatic disease, three reported median OS [14,16,17], three reported median PFS [14,16,17], and all four reported response rates. Although none of the trials found a statistically significant difference between dose-dense and

4 G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012) Fig. 1. Overall disposition of trials on dose-dense chemotherapy identified in Ovid-MEDLINE, CancerLit, EMBASE, and Cochrane, including reasons for exclusion during abstract and full-text review.

5 Table 1 Overview of clinical trials that evaluated dose-dense chemotherapy schedules in non-hodgkin lymphoma, non-small-cell lung cancer, and breast cancer. Author, Year (Trial) Country N randomized a Population Regimen(s) Outcomes b Breast Cancer, early disease Baldini [11,13] Italy 150 Stage IIIA/IIIB (locally advanced) Age Citron (CALGB 9741) [6] United States 2005 Node-positive Age range not noted Jones [18] United Kingdom 84 Early Age < 70 Venturini, 2005 [12] Italy 1214 Node-positive or high-risk node-negative Age range not noted Breast Cancer, advanced disease Capotorto, 2003 [14] Italy 89 Metastatic Age<70 Bonadonna, 1997 (CSF-001) Italy 94 Advanced [15] c Age range not noted De Boer, 2002 [16] United Kingdom 27 Metastatic or locally advanced (inoperable) Age Fountzilas, 1997 [17] Greece 167 Advanced Age 18 Non-Hodgkin Lymphoma Economopoulos, 2007 Greece 238 No stage specified (aggressive) NHL (HE22A99) [19] Previously untreated Age Pfreundschuh, 2004a (NHL-B1) [8] Pfreundschuh, 2004b (NHL-B2) [7] Germany 866 Stage I-IV (aggressive) NHL Previously untreated Age Germany 831 Stage I-IV (aggressive) NHL Previously untreated Age Zhang, 2007 [23] China 66 Stage I-IV DLBCL Newly diagnosed Non-small-cell Lung Cancer Font, 1999 (Spanish Lung Spain 126 Stage IIIA, IIIB, or IV (inoperable) Cancer Group A) [21] No prior chemotherapy Age<70 Masutani, 1999 [20] Japan 100 Stage IIIA, IIIB, or IV (locally advanced or metastatic) Age Soto Parra, 2002 [22] d Italy 107 Any stage Previously untreated Age Dose-dense Standard OS PFS/TTP OR CR PR CEF/CMF q2w CEF/CMF q3w 4 years 5 years Median 5 years X X * ATC q2w sequential ATC q3w sequential 3 years ATC q2w concurrent ATC q3w concurrent 4 years AC q2w AC q3w X X * FEC q2w FEC q3w 5 years 10 years FEC q2w FEC q3w Median Median X X X FNC q2w FNC q3w X ECF q2w ECF q3w Median Median X X * epirubicin q2w epirubicin q4w Median Median X X X CEOP-14 ± rituximab CEOP-21 ± rituximab Endpoint Endpoint X X X CHOP-14 CHOEP-14 CHOP-14 CHOEP-14 CHOP-21 CHOEP-21 CHOP-21 CHOEP-21 5 years * X X 3 years 5 years Endpoint CHOP-14 CHOP-21 1 year 3 years etoposide + cisplatin q3w etoposide + cisplatin q4w 1 year Median * X X * X X Median X X * PVM q3w PVM q4w Median Median X X X gemcitabine q3/4w + cisplatin q4w gemcitabine q2/3w + cisplatin q3w Median X X X CR, complete response; DLBCL, diffuse large B-cell lymphoma; NHL, non-hodgkin lymphoma; OR, overall response; OS, overall survival; PFS, progression-free survival; PR, partial response; TTP, time to progression. a Number randomized for relevant treatment regimens as noted in table. b An X indicates that the study reported the data, while a * indicates that the response rate was calculated by the research team from other data provided in the study. c Published only as a conference abstract. No full-text peer-reviewed publication was identified. d Soto Para was 90% NSCLC (any stage) and 10% other advanced epithelial neoplasms. 300 G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012)

6 Table 2 Summary of survival, progression, and response rates for trials that evaluated dose-dense chemotherapy schedules in breast cancer a. Study N randomized Time Point Dose-dense therapy Comparative therapy Statistical comparison b Regimen Value 95% CI Regimen Value 95% CI Early disease Overall survival Baldini Italian Study [11,13] 150 Median CEF/CMF q2w Not yet reached CEF/CMF q3w 7.8 years CALGB 9741 [6] years ATC q2w (sequential 92% % ATC q3w (sequential 90% % or concurrent) or concurrent) CALGB 9741 [6] years ATC q2w 92% ATC q3w 90% (concurrent) (concurrent) CALGB 9741 [6] years ATC q2w (sequential) 93% ATC q3w (sequential) 88% CALGB 9741 [6] years ATC q2w (sequential or concurrent) ATC q3w (sequential or concurrent) RR = % CI = p = Baldini Italian Study [11,13] years CEF/CMF q2w 76% CEF/CMF q3w 72% Baldini Italian Study [11,13] years CEF/CMF q2w 54% CEF/CMF q3w 52% p = 0.64 Venturini 2005 [12] years FEC q2w 91% 89 93% FEC q3w 89% 86 91% Venturini 2005 [12] years FEC q2w 80% 76 84% FEC q3w 78% 74 82% p = 0.35 Progression-free survival Baldini Italian Study [11,13] years CEF/CMF q2w 56% CEF/CMF q3w 52% p = 0.30 Overall Response Baldini Italian Study [11,13] 150 CEF/CMF q2w 61.6% 49 73% CEF/CMF q3w 62.3% 51 73% Jones 2009 [18] 84 AC q2w 70% AC q3w 79% p = 0.6 Complete response Baldini Italian Study [11,13] 150 (clinical) CEF/CMF q2w 1.4% CEF/CMF q3w 2.6% Baldini Italian Study [11,13] 150 (pathologic primary CEF/CMF q2w 8.2% CEF/CMF q3w 2.6% tumor only) Jones 2009 [18] 84 (pathologic) AC q2w 13% AC q3w 11% p = 0.9 Baldini Italian Study [11,13] 150 (pathologic) CEF/CMF q2w 4.1% % CEF/CMF q3w 2.6% % Advanced disease Overall survival De Boer 2002 [16] 27 Median ECF q2w 19.3 months 3 38 months (range) ECF q3w 18.3 months 1 41 months (range) Fountzilas 1997 [17] 167 Median Epirubicin q2w 14.9 months months (range) Epirubicin q4w 14.6 months months (range) p = Capotorto 2003 [14] 86 Median FEC q2w 25.6 months months FEC q3w 23.3 months months p = 0.55 Progression-free survival De Boer 2002 [16] 27 Median ECF q2w 10.6 months 1 25 months (range) ECF q3w 12 months 1 38 months (range) Fountzilas 1997 [17] 167 Median Epirubicin q2w 7.4 months months (range) Epirubicin q4w 7.2 months months (range) p = Capotorto 2003 [14] 86 Median FEC q2w 12 months months FEC q3w 12.7 months months NS Overall response De Boer 2002 [16] 27 ECF q2w 64% ECF q3w 77% Fountzilas 1997 [17] 167 Epirubicin q2w 53% 43 64% Epirubicin q4w 49% 38 60% p = CSF-001 [15] 94 FNC q2w 50% 36 64% FNC q3w 35% 21 49% Capotorto 2003 [14] 86 (evaluable patients) FEC q2w 71% 57 85% FEC q3w 42% 27 57% p = Capotorto 2003 [14] 86 (ITT) FEC q2w 69% FEC q3w 41% p = Complete response De Boer 2002 [16] 27 ECF q2w 36% ECF q3w 15% Fountzilas 1997 [17] 167 Epirubicin q2w 17% % Epirubicin q4w 5% % p = Capotorto 2003 [14] 86 (ITT) FEC q2w 16.6% FEC q3w 4.5% p = CI, confidence interval; NS, not significant; RR, risk ratio. a This table is organized by (1) time point, (2) outcome, (3) regimen, and then (4) study. b Comparisons that reached statistical significance are indicated in bold italics. G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012)

7 302 G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012) conventional schedules in terms of OS or PFS, two of the trials demonstrated statistically significant higher CR [14,17] and/or OR rates [14] in favor of the dose-dense schedule. Most of the breast cancer studies preferentially used G- CSF prophylaxis in dose-dense regimens. The CALGB 9741 trial [6] used filgrastim on days 3 10 only on dose-dense regimens, and the trial by Venturini and coworkers [12] also used filgrastim but on days 4 11 at a dose of 5 g/kg/d. Baldini et al. [11,13] used GM-CSF (dosing unspecified) in dose-dense regimens only. The Capotorto et al. study [14] selectively used lenograstim (G-CSF) at 150 g/kg for the dose-dense treatment arms. De Boer and colleagues [16] employed the same strategy as Capotorto et al. [14] in using lenograstim but at 300 g/d only on days in dose-dense treatment. Only the trial by Fountzilas et al. [17] employed a filgrastim prophylactic strategy similarly in both dose-dense and non-dose-dense regimens (days 2 12 at 5 g/kg/d) Non-Hodgkin lymphoma Details on the four trials in patients with NHL are provided in Table 3 [7,8,19,23]. The trials enrolled similar populations of patients at all or almost all stages of disease, and three trials [7,8,19] specifically mentioned aggressive NHL. Although most trials compared every-three-week versus every-twoweek schedules, the specific agents and combinations varied among trials as did growth factor support and interval modifications. Sample size varied across studies, ranging from 66 patients in a small Chinese study [23] to more than 800 subjects each in the NHL-B1 and NHL-B2 trials [7,8]. Data on OS were reported by all four trials, including two trials that reported survival for more than one time point. Almost all of the survival data at all time points favored the dose-dense schedule, and two trials [7,8] found a statistically significant improvement in OS for dose-dense therapy. The NHL-B2 trial [7] found that patients who received CHOP-14 were more likely to be alive at the end of the study (approximately 7.5 years) compared to patients who received CHOP-21 (RR = 0.58, p < 0.001). The NHL-B1 trial [8] showed a similar improvement for patients who were randomized to an every-two-week schedule (CHOP-14 or CHOEP-14) compared to a standard every-three-week schedule (CHOP-21 or CHOEP-21) (RR = 0.70, p = 0.044). Only one trial [19] reported data on PFS, with no significant differences reported. All four trials provided data on response rates. Response rates generally favored the dose-dense schedule, but statistical comparisons were made in only a few instances, including one trial (NHL-B2 [7]) that reported a significantly higher complete response rate for CHOP-14 compared to CHOP-21 (76.1% versus 60.1%, p = 0.001). The NHL studies demonstrated considerable variation in the treatment protocols. Three of the four trials utilized prophylactic G-CSF only in the dose-dense schedule while permitting physician discretion in the non-dose-dense schedules, which resulted in significantly less use of prophylactic G-CSF [7,8,19]. For example, in the NHL-B1 trial [8], more than 96% of patients in dose-dense arms received G-CSF prophylaxis in cycles 1 5, and although G-CSF use increased with each subsequent cycle with non-dose-dense CHOP or CHOEP, only 6 17% received prophylaxis after cycle 5. One trial allowed G-CSF prophylactic use in both schedules (Zhang [23]) Non-small-cell lung cancer An overview of the three RCTs in patients with NSCLC is presented in Table 4 [20 22]. Two of the trials included patients with stage IIIA, IIIB, or IV disease, and the third trial included patients at any stage of disease, of which 90% of patients had NSCLC while the other 10% had other advanced epithelial neoplasms. All three trials evaluated cisplatinbased combination regimens, but growth factor support and interval modifications varied across trials. Sample size was similar across trials and ranged from 100 to 126 subjects. All three trials reported median OS, and one trial also reported one-year survival data. One trial (Masutani 1999 [20]) demonstrated improved median survival with a threeweekly schedule of PVM compared to standard four-weekly therapy (57.0 weeks versus 37.7 weeks, p = 0.036). Median PFS was reported by two trials, with Masutani [20] demonstrating a significant improvement in favor of dose-dense therapy (23.7 weeks versus 13.9 weeks, p = 0.006). All three trials reported data on tumor response. Complete response rates were consistently low (0 8% for dose-dense treatment arms, 0 3% for standard therapy), and there were no differences in CR or OR rates among these three trials. Among the NSCLC trials, Font et al. [21] allowed use of GM-CSF only in the dose-dense regimen. In another study by Masutani and coworkers [20], prophylactic G-CSF was used only in the dose-dense regimen while reactive use was employed in the non-dose-dense regimen if ANC 2000/ l. In the trial by Jones et al. [18], pegfilgrastim was used prophylactically only in the dose-dense regimen Meta-analysis feasibility Quantitative meta-analyses were considered for each of the three cancer types, but significant clinical heterogeneity across studies precluded this approach. Meta-analysis of more homogeneous subgroups, such as those based on specific treatment regimens in specific populations for specific outcomes (e.g., five-year overall survival for CHOP-14 versus CHOP-21 in patients with NHL), also was considered, but such stratification resulted in too few studies for analysis. 4. Discussion While clinical trials of dose-escalating chemotherapy have demonstrated inconsistent results providing little clinical benefit, early modeling studies followed by preliminary clinical trials generated considerable interest in delivering

8 Table 3 Summary of survival, progression, and response rates for trials that evaluated dose-dense chemotherapy schedules in non-hodgkin lymphoma a. Trial N randomized Time point Dose-dense therapy Comparative therapy Statistical Comparison b Regimen Value 95% CI Regimen Value 95% CI Overall survival Zhang 2007 [23] 66 1 year (GCB subgroup) CHOP % CHOP % Zhang 2007 [23] 66 1 year (non-gcb subgroup) CHOP % CHOP % NHL-B2 [7] years CHOEP % % CHOEP % % NHL-B2 [7] years CHOP % % CHOP % % Zhang 2007 [23] 66 3 years (GCB subgroup) CHOP % CHOP % Zhang 2007 [23] 66 3 years (non-gcb subgroup) CHOP % CHOP % NHL-B1 [8] years CHOEP % % CHOEP % % NHL-B2 [7] years CHOEP % % CHOEP % % NHL-B1 [8] years CHOP % % CHOP % % NHL-B2 [7] years CHOP % % CHOP % % NHL-B1 [8] years CHOP-14 or CHOP % % CHOP-21 or CHOEP % % HE22A99 [19] years (endpoint) CEOP % CEOP % p = NHL-B2 [7] years (endpoint) CHOP-14 CHOP-21 RR = % CI = p < NHL-B1 [8] years (endpoint) CHOP-14 or CHOP-21 CHOP-21 or CHOEP-21 RR = % CI = p = Progression-free survival HE22A99 [19] years (endpoint) CEOP % CEOP % p = Overall response HE22A99 [19] 238 CEOP-14 79% CEOP-21 76% p = NHL-B1 [8] 362 CHOEP % CHOEP % NHL-B2 [7] 339 CHOEP % CHOEP % NHL-B1 [8] 348 CHOP % CHOP % NHL-B2 [7] 350 CHOP % CHOP % Zhang 2007 [23] 66 CHOP % CHOP % NHL-B1 [8] 710 CHOP-14 or CHOEP % CHOP-21 or CHOEP % Complete response HE22A99 [19] 238 CEOP-14 53% CEOP-21 51% p = NHL-B1 [8] 348 CHOEP % % CHOEP % % NHL-B2 [7] 339 CHOEP % % CHOEP % % NHL-B1 [8] 362 CHOP % % CHOP % % NHL-B2 [7] 350 CHOP % % CHOP % % OR = % CI = p = Zhang 2007 [23] 66 CHOP % CHOP % NHL-B1 [8] 710 CHOP-14 or CHOEP % % CHOP-21 or CHOEP % % p = CI, confidence interval; HR, hazard ratio; OR, odds ratio; RR, risk ratio. a This table is organized by (1) time point, (2) outcome, (3) regimen, and then (4) study. b Comparisons that reached statistical significance are indicated in bold italics. G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012)

9 Table 4 Summary of survival, progression, and response rates for trials that evaluated dose-dense chemotherapy schedules in non-small-cell lung cancer a. Trial N randomized Time point Dose-dense therapy Comparative therapy Statistical comparison b Overall survival Spanish Lung Cancer Group (Cis-Etop) [21] Soto Parra 2002 [22] Masutani 1999 [20] Spanish Lung Cancer Group (Cis-Etop) [21] Progression-free survival Spanish Lung Cancer Group (Cis-Etop) [21] Masutani 1999 [20] Overall response Spanish Lung Cancer Group (Cis-Etop) [21] 126 Median Etoposide q3w + Cisplatin q3w Regimen Value 95% CI Regimen Value 95% CI 9.0 months months Etoposide q4w + Cisplatin q3w 7.2 months months p = Median Gemcitabine 279 days Gemcitabine 367 days p = 0.49 q3/4w + Cisplatin q4w q2/3w + Cisplatin q3w 100 Median PVM q3w 57.0 weeks weeks PVM q4w 37.7 weeks weeks p = year Etoposide q3w + Cisplatin q3w 126 Median Etoposide q3w + Cisplatin q3w 34% % Etoposide q4w + Cisplatin q3w 7.0 months 5 10 months Etoposide q4w + Cisplatin q3w 20% % 5.6 months months p = Median PVM q3w 23.7 weeks weeks PVM q4w 13.9 weeks weeks p = Etoposide q3w + Cisplatin q3w 27% 22 32% Etoposide q4w + Cisplatin q3w 32% % p = 0.9 Soto Parra Gemcitabine 38% Gemcitabine 42% q3/4w + Cisplatin q4w q2/3w + Cisplatin q3w Masutani 1999 [20] 100 PVM q3w 52% PVM q4w 44% p = Complete response Spanish Lung Cancer Group (Cis-Etop) [21] Soto Parra 2002 [22] 126 Etoposide q3w + Cisplatin q3w 8% Etoposide q4w + Cisplatin q3w 2% 107 Gemcitabine 0% Gemcitabine 3% q3/4w + Cisplatin q4w q2/3w + Cisplatin q3w Masutani 1999 [20] 100 PVM q3w 2% PVM q4w 0% CI, confidence interval. a This table is organized by (1) time point, (2) outcome, (3) regimen, and then (4) study. b Comparisons that reached statistical significance are indicated in bold italics. 304 G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012)

10 G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012) conventional chemotherapy doses at more frequent intervals. The theoretical basis of such dose-dense chemotherapy delivery is evidence that chemotherapy-induced cytoreduction results in an increased growth rate of remaining cells and greater tumor cell kill with subsequent cytotoxic therapy. Larry Norton and Richard Simon developed mathematical models that predicted greater tumor cell kill with compression of the schedule of chemotherapy [1 3]. While dose-dense schedules increase the dose intensity of chemotherapy delivered, they generally must be supported by the administration of myeloid growth factors to enable the more frequent delivery of myelosuppressive agents. Although large, prospective RCTs of dose-dense schedules have reported improved survival in ESBC and NHL, the totality of evidence available to date on the comparative efficacy of dose-dense schedules, as discussed in this review, has not yet been fully summarized. The majority of studies of dose-dense schedules identified in this review were relatively small with limited power to demonstrate clinically relevant differences in treatment effect. Nevertheless, several studies demonstrated a trend toward improved survival for dose-dense schedules among larger phase III RCTs in potentially curative settings such as ESBC and NHL (e.g., CALGB 9741 [6], NHL-B1 [8], NHL- B2 [7]). Of note, none of the trials in this review found a statistical relationship in favor of conventional schedules. Considerable heterogeneity was encountered across the RCTs identified in this review in terms of the patient populations, disease stage, chemotherapy regimens, growth factor support, reported outcomes, and follow-up durations. For instance, RCTs of dose-dense therapy for breast cancer were conducted in patients at various stages of disease in both North America and Europe, and the specific agents used varied according to common practice. While the diversity of cytotoxic agents used demonstrates the flexible application of the dose-dense paradigm, it also limits any overall synthesis of evidence in this review and precludes quantitative metaanalysis, even among more similar subgroups. However, in a 2010 systematic review, Bonilla and coworkers [24] took the opposite approach and pursued a fixed-effects meta-analysis of dose-dense chemotherapy in non-metastatic breast cancer. The definition of conserved dose-dense chemotherapy used in their review, i.e., similar doses in both treatment arms but with a condensed schedule in one arm, is comparable to our definition, and their meta-analysis of overall survival included the same studies of ESBC patients that we identified (i.e., CALGB 9741, Baldini et al., Venturini et al.). Metaanalysis of these 3 studies found that conserved dose-dense chemotherapy was associated with greater overall survival (pooled HR = 0.84, 95% CI: ; p = 0.03) compared to conventional regimens. The findings from Bonilla and colleagues quantitative synthesis are in agreement with the findings from our qualitative approach. G-CSF has been shown to reduce the risk of neutropenic complications while enabling delivery of greater relative dose intensity in patients receiving cancer chemotherapy [5]. The utilization of hematopoietic growth factor support of patients randomized to dose-dense schedules varied considerably across trials. The variety of G-CSF prophylactic and reactive strategies make direct comparison of these approaches difficult; however, the studies in this review employed G- CSF only in dose-dense treatment arms with few exceptions [17,23]. Adherence to chemotherapy dose and schedule as well as delivered chemotherapy relative dose intensity was not reported in many trials. In a systematic review of RCTs of G-CSF reporting both secondary malignancies and survival, a significant association was observed between reduced allcause mortality and delivered chemotherapy relative dose intensity with G-CSF support [9]. The reduction in allcause mortality observed was inversely associated with both planned and delivered chemotherapy relative dose intensity. The strongest and most consistent association between dose intensity and survival was among the studies utilizing dose dense chemotherapy schedules with G-CSF support [25]. A relatively consistent finding across the trials reviewed here is the apparent safety of dose-dense schedules when supported by a myeloid growth factor. The majority of trials reported similar rates of neutropenic events between patients receiving dose-dense schedules with G-CSF support and control patients receiving conventional chemotherapy schedules without such support [6]. The four NHL studies illustrate the biased use of G-CSF prophylaxis in the dose-dense arms. These studies generally reported similar hematologic toxicity rates and infectious complications between dose-dense and non-dose-dense arms, although some reported slightly higher rates of hematologic toxicity or treatment for dose-dense regimens [7,8,23]. In the one breast cancer trial where there was appropriate and equivalent methodology for using prophylactic G-CSF in both dose-dense and non-dose-dense treatment arms, there were no significant differences in any toxicity between groups [17]. However, higher rates of most non-leukocyte hematologic toxicities and outcomes were reported in the dose-dense arm. In another trial [12], the incidence of thrombocytopenia and anemia were higher in the dose-dense arm compared to the non-dose-dense arm, while leukopenia occurred with lower frequency (i.e., seven cases of grade 3 leukopenia and two cases of grade 4 leukopenia in the non-dose-dense regimen versus three cases of grade 3 leukopenia in the dosedense regimen). Findings were similar in the NHL-B1 and NHL-B2 trials [7,8] (i.e., higher rates of anemia and thrombocytopenia for dose-dense regimens). CALBG 9741 [6], which selectively used G-CSF only for dose-dense regimens, found that significantly more dose delays were due to hematologic toxicity among patients who received non-dose-dense regimens (38%) compared to dose-dense regimens (15%). In addition, grade 4 granulocytopenia (<500/ l) occurred with greater frequency in non-dose-dense regimens (33% versus 6%, p < ). Toxicities in the Baldini study [11,13] were similar in both arms. The Capotorto study [14] showed lower grade 3 4 leukopenia in the dose-dense arm that had G-CSF

11 306 G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012) prophylaxis (6.7 versus 12.3%, p < 0.001), while thrombocytopenia and anemia were higher in the dose-dense arm, which is similar to results reported previously. The De Boer study [16] also demonstrated positive effects of using G-CSF in the dose-dense regimen and went so far as to report a positive impact of G-CSF on haematological recovery. The Masutani et al. NSCLC study [20] also demonstrated that the G-CSF prophylactic support in the dose-dense arm led to an increase in WBC counts, although patients experienced greater chemotherapy-related thrombocytopenia and anemia in the dose-dense arm. The Jones et al. study [18] also showed a higher incidence of grade 3 4 neutropenia in the non-g-csf, non-dose-dense regimens (p = 0.001) with no difference in other hematologic toxicities but notably more RBC and platelet transfusions in the dose-dense arm. In clinical practice, patients and clinicians need to weight the benefits, harms, and costs of dose-dense chemotherapy regimens. On one hand, patients have to return for treatment more frequently, and the use of G-CSF agents requires next-day or daily injections and adds to the overall cost of treatment. On the other hand, the use of G-CSF may reduce the toxicity associated with dose-dense regimens to levels that are equal to or less than standard schedules. In addition, dose-dense schedules are attractive to many patients since chemotherapy is completed in less time, which, in some cases, also permits earlier surgery or radiation treatment. Ultimately, the value of dose-dense chemotherapy hinges on whether or not these condensed regimens improve survival in specific populations. A number of limitations to this review should be considered. First, although dose-dense chemotherapy schedules have been applied to a number of malignancies, this review was limited to patients with NHL and cancers of the breast and lung. Results for these tumor types may or may not be applicable to dose-dense schedules for other cancer types, such as ovarian or endometrial cancer. Second, several aspects of the search strategy may have limited our findings. Two relevant conference abstracts without corresponding full-text, peerreviewed publications were identified through the search of relevant bibliographies. However, abstracts from conferences where clinical trials of dose-dense chemotherapy are commonly reported were not systematically searched. Another challenge to the publication review process was the lack of a consistent nomenclature for dose-dense chemotherapy. In order to overcome the lack of consistent terminology in this area, very broad search terms were employed in an effort to capture trials of interest. Finally, the a priori definition for dose-dense therapy utilized in this review was relatively stringent (i.e., any regimen that decreased the interval between cycles while maintaining the same drugs and total dose per treatment as other regimen(s) in the trial). As such, several trials that provide clinical support for the benefits of dose-dense regimens were excluded. For example, two secondary analyses, including a retrospective subgroup analysis [26] and a meta-analysis [24], have suggested that the benefits of dose-dense therapy observed in CALGB 9741 and Venturini et al. were limited to patients with hormone receptor-negative breast cancer, but the advantages of dose-dense chemotherapy were independent of hormone receptor status in a recent trial reported by Moebus and coworkers [27]. This study, however, failed to meet our inclusion criteria since the dose per treatment varied between the study arms (e.g., paclitaxel 175 mg/m 2 q3w versus paclitaxel 225 mg/m 2 q2w). In their meta-analysis of RCTs in patients with non-metastatic breast cancer, Bonilla and coworkers [24] included a separate analysis of trials that examined modified dose-dense schedules, i.e., dosedense regimens that modified specific agents and/or dose per treatment, such as in Moebus et al. They found statistically significant survival benefits in favor of these modified dose-dense schedules compared to standard schedules. The combined findings from the CALGB 9741 and Moebus trials provide support for the importance of dose intensity in women with early-stage breast cancer receiving adjuvant chemotherapy. While premised upon intriguing mathematical models and very encouraging early results, the strongest support for the efficacy and safety of dose-dense chemotherapy comes from a few large prospective randomized trials in patients with ESBC or aggressive NHL in the curative setting with G-CSF support. Conflict of interest Dr. Lyman is principal investigator on a research grant to Duke University from Amgen, Inc. Mr. Barron is an employee of Amgen, Inc. and therefore receives personal compensation from and has an equity/ownership (e.g., company stock) in the study sponsor. Dr. Miller and Ms. Natoli are employees of Cerner Life- Sciences, which received research funding for this project. Role of the funding source Representatives from Amgen, the study sponsor, were involved in the study design, data analysis, and interpretation of data. Representatives from the study sponsor also contributed as co-authors of the manuscript and were involved in the decision to submit the manuscript for publication. Reviewers Professor Christoph C. Zielinski, Medical University of Vienna, Clinical Division of Oncology, Dept. of Medicine I, Währinger Gürtel 18-20, A-1090 Vienna, Austria. Professor Nagahiro Saijo, Kinki University School of Medicine, 377-2, Ohono-Higashi, Osaka-Sayama, Osaka, , Japan.

12 G.H. Lyman et al. / Critical Reviews in Oncology/Hematology 81 (2012) Sibylle Loibl, M.D., Ph.D., German Breast Group, Medicine and Research, Martin-Behaim-Str. 12, D Neu-Isenburg, Germany. Professor Lorenz Trümper, Chairman, Georg-August University, Göttingen, Dept. of Hematology and Oncology, Robert-Koch-Str. 40, D Göttingen, Germany. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi: /j.critrevonc References [1] Norton L. A Gompertzian model of human breast cancer growth. Cancer Res 1988;48(24 Pt 1): [2] Norton L, Simon R. Tumor size, sensitivity to therapy, and design of treatment schedules. Cancer Treat Rep 1977;61(7): [3] Norton L, Simon R. The Norton-Simon hypothesis revisited. Cancer Treat Rep 1986;70(1): [4] Lyman GH. Impact of chemotherapy dose intensity on cancer patient outcomes. J Natl Compr Canc Netw 2009;7(1): [5] Kuderer NM, Dale DC, Crawford J, Lyman GH. Impact of primary prophylaxis with granulocyte colony-stimulating factor on febrile neutropenia and mortality in adult cancer patients receiving chemotherapy: a systematic review. J Clin Oncol 2007;25(21): [6] Citron ML, Berry DA, Cirrincione C, et al. Randomized trial of dose-dense versus conventionally scheduled and sequential versus concurrent combination chemotherapy as postoperative adjuvant treatment of node-positive primary breast cancer: first report of Intergroup Trial C9741/Cancer and Leukemia Group B Trial J Clin Oncol 2003;21(8): [7] Pfreundschuh M, Trumper L, Kloess M, et al. Two-weekly or 3-weekly CHOP chemotherapy with or without etoposide for the treatment of elderly patients with aggressive lymphomas: results of the NHL-B2 trial of the DSHNHL. Blood 2004;104(3): [8] Pfreundschuh M, Trumper L, Kloess M, et al. Two-weekly or 3-weekly CHOP chemotherapy with or without etoposide for the treatment of young patients with good-prognosis (normal LDH) aggressive lymphomas: results of the NHL-B1 trial of the DSHNHL. Blood 2004;104(3): [9] Lyman GH, Dale DC, Wolff DA, et al. Acute myeloid leukemia or myelodysplastic syndrome in randomized controlled clinical trials of cancer chemotherapy with granulocyte colony-stimulating factor: a systematic review. J Clin Oncol 2010;28: [10] Nistico C, Garufi C, Barni S. Phase II study of epirubicin and vinorelbine with granulocyte colony-stimulating factor: a high-activity, dose-dense weekly regimen for advanced breast cancer. Ann Oncol 1999;10(8): [11] Baldini E, Gardin G, Giannessi P, et al. Accelerated CEF/CMF chemotherapy versus standard CEF/CMF: a randomized trial in locally advanced breast cancer. Proc Am Soc Clin Oncol 2001;20(Pt 1): 37a. [12] Venturini M, Del Mastro L, Aitini E. Dose-dense adjuvant chemotherapy in early breast cancer patients: results from a randomized trial. J Natl Cancer Inst 2005;97(23): [13] Baldini E, Gardin G, Giannessi PG, et al. Accelerated versus standard cyclophosphamide, epirubicin and 5-fluorouracil or cyclophosphamide, methotrexate and 5-fluorouracil: a randomized phase III trial in locally advanced breast cancer. Ann Oncol 2003;14(2): [14] Capotorto AM, Pavesi L, Pedrazzoli P, et al. Randomized, controlled, multicenter phase III trial of standard-dose fluorouracilepirubicin-cyclophosphamide (FEC), compared with time-intensive FEC (FEC-G) and mitoxantrone methotrexate mitomycin C (MMM- G) in metastatic breast carcinoma. J Chemother 2003;15(2): [15] Bonadonna G, Pannuti F, Robustelli della cuna G, et al. Phase III study comparing standard dose FNC (5-fluorouracil, mitoxantrone, cyclophosphamide) versus dose-intensive FNC + rhug-csf (lenograstim) versus time-intensive FNC + lenograstim in advanced breast cancer (ABC) patients: preliminary results of the (CSF-001) study. Proc Am Soc Clin Oncol Abstract 523. [16] De Boer RH, Eisen TG, Ellis PA, et al. A randomised phase II study of conventional versus accelerated infusional chemotherapy with granulocyte colony-stimulating factor support in advanced breast cancer. Ann Oncol 2002;13(6): [17] Fountzilas G, Athanassiades A, Giannakakis T, et al. A randomized study of epirubicin monotherapy every four or every two weeks in advanced breast cancer. A Hellenic Cooperative Oncology Group study. Ann Oncol 1997;8(12): [18] Jones RL, Walsh G, Ashley S, et al. A randomised pilot Phase II study of doxorubicin and cyclophosphamide (AC) or epirubicin and cyclophosphamide (EC) given 2 weekly with pegfilgrastim (accelerated) vs 3 weekly (standard) for women with early breast cancer. Br J Cancer 2009;100(2): [19] Economopoulos T, Psyrri A, Dimopoulos MA, et al. CEOP-21 versus CEOP-14 chemotherapy with or without rituximab for the firstline treatment of patients with aggressive lymphomas: results of the HE22A99 trial of the Hellenic Cooperative Oncology Group. Cancer J 2007;13(5): [20] Masutani M, Tsujino I, Fujie T, et al. Moderate dose-intensive chemotherapy for patients with non-small cell lung cancer: randomized trial, can it improve survival of patients with good performance status? Oncol Rep 1999;6(5): [21] Font A, Moyano AJ, Puerto JM, et al. Increasing dose intensity of cisplatin-etoposide in advanced nonsmall cell lung carcinoma: a phase III randomized trial of the Spanish Lung Cancer Group. Cancer 1999;85(4): [22] Soto Parra HJ, Cavina R, Latteri F, et al. Three-week versus four-week schedule of cisplatin and gemcitabine: results of a randomized phase II study. Ann Oncol 2002;13(7): [23] Zhang Q, Wang J, Biweekly CHOP. therapy improves therapeutic effect in the non-gcb subtype of diffuse large B-cell lymphoma. Cent Eur J Med 2007;2(4): [24] Bonilla L, Ben Aharon I, Vidal L, Gafter-Gvili A, Leibovici L, Stemmer SM. Dose-dense chemotherapy in nonmetastatic breast cancer: a systematic review and meta-analysis of randomized controlled trials. J Natl Cancer Inst 2010;102(24): [25] Chang J. Chemotherapy dose reduction and delay in clinical practice. Evaluating the risk to patient outcome in adjuvant chemotherapy for breast cancer. Eur J Cancer 2000;36(Suppl 4): S11 4. [26] Berry DA, Cirrincione C, Henderson IC, Citron ML, Budman DR, Goldstein LJ. Effects of improvements in chemotherapy on diseasefree and overall survival of estrogen-receptor negative, node-positive breast cancer. 20-year experience of the CALGB and US Breast Intergroup. Presented at the 27th Annual San Antonio Breast Cancer Symposium. [27] Moebus V, Jackisch C, Lueck HJ, et al. Intense dose-dense sequential chemotherapy with epirubicin, paclitaxel, and cyclophosphamide compared with conventionally scheduled chemotherapy in high-risk primary breast cancer: mature results of an AGO phase III study. J Clin Oncol 2010;28(17):