Influence of non-measurable disease on progression-free survival in patients with metastatic breast cancer

Cancer Treatment Reviews 59 (2017) 46–53

Contents lists available at ScienceDirect

Cancer Treatment Reviews journal homepage: www.elsevierhealth.com/journals/ctrv

Systematic or Meta-analysis Studies

Influence of non-measurable disease on progression-free survival in patients with metastatic breast cancer Hadar Goldvaser a,b, Domen Ribnikar a, Rouhi Fazelzad a, Bostjan Seruga c, Arnoud J. Templeton d, Alberto Ocana e, Eitan Amir a,⇑ a

Division of Medical Oncology, University of Toronto and Princess Margaret Cancer Centre, 610 University Ave, Toronto, ON, M5G 2M9, Canada Sackler Faculty of Medicine, Tel Aviv University, POB 39040, Tel Aviv 6997801, Israel Department of Medical Oncology, Institute of Oncology Ljubljana, 2 Zaloska cesta, Ljubljana 1000, Slovenia d Department of Oncology and Hematology, St. Claraspital, Faculty of Medicine, University of Basel, Kleinriehenstrasse 30, Basel 4058, Switzerland e Translational Research Unit, Centro Regional de Investigaciones Biomedicas Universidad de Castilla La Mancha, Albacete University Hospital, calle Francisco Javier de Moya 02006, Albacete, Spain b c

a r t i c l e

i n f o

Article history: Received 17 May 2017 Received in revised form 26 June 2017 Accepted 28 June 2017

Keywords: Bone-only disease Measurable disease Endpoints Metastatic breast cancer Progression-free survival

a b s t r a c t Background: The presence of non-measurable disease is common in metastatic breast cancer. It is unknown whether treatment effect on progression free survival (PFS) is consistent among patients with measurable and non-measurable disease. Methods: A systematic literature search identified phase III randomized controlled trials (RCTs) in metastatic breast cancer that reported outcomes in patients with non-measurable and measurable disease. Hazard ratios (HRs) and 95% confidence intervals were computed to compare the individual trial treatment effect on PFS in non-measurable versus measurable disease. Analyses were repeated for boneonly compared to non-bone-only disease and based on drug mechanism of action. Results: Among 82 RCTs that enrolled patients with non-measurable disease, data were available from 16 trials comprising 8516 patients. Treatment effect on PFS was similar in patients with non-measurable and measurable disease (HR for intra-study comparison = 1.01, p = 0.82). However, compared to non-boneonly disease, a significantly greater effect on PFS was seen in those with bone-only disease (HR 0.83, p = 0.03). Compared to patients with measurable disease, there was a greater effect on PFS in those with non-measurable disease in RCTs of signal transduction inhibitors and endocrine therapy (HR 0.74, p = 0.01) and a lesser effect on PFS in RCTs of antiangiogenic drugs (HR 1.34, p = 0.02). Comparable effect on PFS was shown in RCTs evaluating endocrine therapy (HR 1.13, p = 0.23) and chemotherapy (HR 0.73, p = 0.22). Conclusions: There is variability in treatment effect on PFS in patients with measurable and nonmeasurable disease, especially those with bone-only disease. Standardization of PFS determination in these patients is warranted. Ó 2017 Elsevier Ltd. All rights reserved.

Introduction Despite its lack of established surrogacy with overall survival (OS), progression-free survival (PFS) has become the dominant primary endpoint in phase III randomized controlled trials (RCTs) in metastatic breast cancer, including trials which support drug

⇑ Corresponding author at: Princess Margaret Cancer Centre, 700 University Ave, 7-721, Toronto, ON M5G 2M9, Canada. E-mail addresses: [email protected] (H. Goldvaser), [email protected] (D. Ribnikar), [email protected] (R. Fazelzad), [email protected] (B. Seruga), [email protected] (A.J. Templeton), [email protected] (A. Ocana), [email protected] (E. Amir). http://dx.doi.org/10.1016/j.ctrv.2017.06.007 0305-7372/Ó 2017 Elsevier Ltd. All rights reserved.

approval [1,2]. Possible explanations for the poor performance of PFS as a surrogate for OS include trial and patient-related characteristics such as cross-over and both the duration of survival and exposure to treatment post-progression [2]. An advantage of the use of PFS as a primary endpoint includes a smaller sample size requirement and shorter duration of follow-up compared to OS. Unlike OS where the measurement of outcome is rarely subject to variability, PFS and other tumor response-based endpoints are subject to measurement error [3,4]. The Response Evaluation Criteria in Solid Tumors (RECIST) criteria were developed with the aim of standardizing the measurement of target lesions and their change over time [5]. However, even with the establishment of these criteria, inconsistencies in the measurement of PFS such as

H. Goldvaser et al. / Cancer Treatment Reviews 59 (2017) 46–53

differences between investigator assessment and blinded independent radiology review are well described in a variety of solid tumors including breast cancer [6]. Discordance in radiologic response assessment have been described both in trials with a low proportion of response [7] and in those with a high proportion of response [8]. This discordance can have a substantial effect on the interpretation of trial results [8] which likely explains why typically, regulators advise the use of both investigator-assessed and blinded independent review of imaging data [9]. Patients with metastatic breast cancer can present with nonmeasurable disease such as bone-only involvement. In this population, the assessment of PFS can be challenging. Radiologic abnormalities such as lytic or sclerotic lesions, or increased uptake on bone scintigraphy, are not direct measures of tumor burden, but rather measures of response of native bone cells to cancer. As progressive disease and flare resulting from early treatment efficacy can appear similar [10,11], interpretation of bone imaging may be subjective and prone to inter-observer variability. The influence of inclusion of patients with non-measurable disease in RCTs in metastatic breast cancer is not well established. Additionally, there are limited data on whether treatment effect on PFS is consistent among patients with non-measurable and measurable disease. Here, we report on a systematic review and meta-analysis which aimed to explore the effect of treatment on PFS in patients with non-measurable and measurable metastatic breast cancer enrolled in phase III trials. We hypothesized that there would be significant variability in treatment effect on PFS, especially in patients with bone-only disease. Materials and methods Literature review and study identification A comprehensive search strategy was employed in MEDLINE (Host: OvidSP: 1946 to October Week 3 2016), Epub Ahead of Print and In-Process & Other Non-Indexed Citations (Host: OvidSP, November 07, 2016), EMBASE (Host: OvidSP, 1974–2016 November 09) and Cochrane Central Register of Controlled Trials (Host: OvidSP, October 2016) to identify all phase III RCTs in metastatic breast cancer published between 2002 (the year of the last revision of the American Joint Committee on Cancer definition of metastatic breast cancer) and week 1, November 2016 (see Supplementary Appendix S1 for search strategy). Databases of the Annual Meetings of the American Society of Clinical Oncology and the San Antonio Breast Cancer Symposium were searched for abstracts presented between 2014 and 2016. Studies reported as abstracts before 2014 were expected to have been published fully and captured in the main literature search. The search was restricted to English language articles. Eligible trials were all phase III RCTs in metastatic breast cancer which reported outcomes in subgroups of patients with non-measurable and measurable disease. If data on non-measurable disease were not explicitly reported, boneonly disease was used as a surrogate for non-measurable disease. Data extraction The following data were extracted from eligible studies and recorded in predesigned electronic forms: name of first author, year of publication, type of experimental therapy (cytotoxic, endocrine, signal transduction inhibitors, antiangiogenic), primary endpoint and interval of re-imaging. We then extracted the hazard ratios (HRs) and respective 95% confidence intervals (CI) for subgroups of patients with non-measurable and measurable disease, bone-only and non-bone-only disease, and visceral and nonvisceral disease. When the HRs for subgroups were not reported

47

explicitly, they were estimated from forest plots. Finally, we collected data on the proportion of patients with non-measurable disease and those with hormone receptor positive disease in each individual study. Data were collected independently by two reviewers (HG and DR). Discrepancies were resolved by a third reviewer (EA). All data were extracted from primary publications and their associated online appendices. Data synthesis The primary analysis compared the effect of treatment on PFS in patients with non-measurable compared to measurable disease. Individual HRs for patients with non-measurable and measurable disease were utilized to calculate a single HR and its associated 95% CI comparing the effect of individual study treatment on PFS among patients non-measurable to measurable disease. A HR less than 1.0 corresponds to a greater effect of treatment on PFS in patients with non-measurable disease while a HR of greater than 1.0 corresponds to a lesser effect of treatment on PFS in patients with non-measurable disease. This analysis was repeated comparing bone-only to non-bone-only and for visceral and non-visceral disease and for different types of experimental treatment. Metaregression was used to explore the influence of the proportion of patients with non-measurable disease, the proportion of patients with hormone receptor positive disease and the interval of reimaging on the results. Sensitivity analyses were conducted to assess whether the use of bone-only disease as a surrogate for non-measurable disease and whether data estimated from forest plots rather than extracted directly from trial reports affected the results. Additionally, as human epidermal growth factor receptor 2 positive (HER2) disease represent a distinct entity, we performed post-hoc sensitivity analyses excluding studies conducted exclusively in patients with HER2 positive disease. Statistical analysis Data from individual trials were pooled in a meta-analysis using RevMan 5.3 (The Cochrane Collaboration, Copenhagen, Denmark). HRs for PFS (or time to progression, if PFS was not reported) were weighted and pooled using the generic inverse variance and random-effect model. Meta-regression comprised a linear regression weighted by individual study sample size exploring the association between the natural log of the HR for treatment effect on PFS and the proportion of patients with non-measurable disease, the proportion of patients with hormone receptor positive disease and the interval of re-imaging. Meta-regression was performed using SPSS version 21 (IBM Corp, Armonk, NY, USA). Statistical tests were two-sided, and statistical significance was defined as p < 0.05. No correction was made for multiple statistical testing.

Results The search identified 9341 titles. After exclusion of duplicates, 4373 studies were assessed and 4050 were excluded on the basis of the title and a further 175 excluded on the basis of the abstract or full text assessment. Among the remaining 148 studies, 65 studies did not include patients with non-measurable disease or did not report explicitly if such patients were included. An additional 66 studies did not report results on the basis of subgroups of patients with measurable and non-measurable disease (or the surrogate subgroup of bone-only versus non-bone-only disease). Finally, one additional study was excluded as it was an earlier report of a previously included study. Consequently, a total of 16 studies comprising 8516 patients were eligible for analysis

48

H. Goldvaser et al. / Cancer Treatment Reviews 59 (2017) 46–53

(Fig. 1). The characteristics of included studies are shown in Table 1. Among the 16 included studies, 7 studies did not report data explicitly based on subgroups of measurable and non-measurable disease, but reported subgroup data for bone-only versus nonbone-only disease [12–18]. Outcomes for bone-only disease were reported in 10 trials [12–21] while 11 trials reported outcomes of patients with visceral and non-visceral disease [14–24]. In 7 studies, data for subgroups were not reported explicitly, but were estimated from forest plots [12,13,15–18,20]. The primary endpoint was PFS in 14 studies [12–23,25,26] and time to progression in 2 studies [24,27]. Results of individual study subgroup analyses comparing measurable versus non-measurable disease as well as bone-only versus non-bone-only disease are shown in Table 2. In the pooled analysis, the effect of treatment on PFS was comparable between patients with non-measurable and measurable disease (HR 0.72, 95% CI 0.64–0.80 and HR 0.69, 95% CI 0.65–0.73 respectively), HR for intra-study comparison 1.01, 95% CI 0.89–1.15, p = 0.82, see Fig. 2A). Sensitivity analyses showed similar results when excluding studies in which bone-only disease was used as a surrogate for non-measurable disease (intra-study HR 1.12, 95% CI 0.96–

1.31), when excluding studies in which subgroup data were estimated from forest plots rather than extracted directly from trial reports (intra-study HR 1.17, 95% CI 0.97–1.42) and when excluding studies that included only patients with HER2 positive disease [16,25] (intra-study HR 0.99, 95% CI 0.87–1.13). Compared to patients with non-bone-only disease, those with bone-only disease had significantly higher effect of treatment on PFS (intra-study HR 0.83, 95% CI 0.70–0.98, p = 0.03, see Fig. 2B). Sensitivity analysis excluding studies in which subgroup data were estimated from forest plots rather than extracted showed similar results (intra-study HR 0.91, 95% CI 0.72–1.14). The effect of treatment on PFS was similar among patients non-visceral and visceral disease (intra-study HR 0.96, 95% CI 0.85–1.10, p = 0.59). Again, sensitivity analysis did not influence the results. Excluding studies in which bone-only disease was used as a surrogate for nonmeasurable disease (intra-study HR 0.95, 95% CI 0.78–1.16) and in which subgroup data were estimated from forest plots rather than extracted (intra-study HR 1.00, 95% CI 0.85–1.18) yielded similar effect sizes. Meta-regression did not identify any factors which influenced the treatment effect on PFS in patients with measurable and nonmeasurable disease (see Table 3). Only 4 studies reported PFS data

Fig. 1. Study selection schema.

49

H. Goldvaser et al. / Cancer Treatment Reviews 59 (2017) 46–53 Table 1 Characteristics and design of included studies. Trial

Number of patients

Bone-only as a surrogate for non-measurable disease

Experimental arm

Control arm

Primary endpoint

Nonmeasurable disease (%)

Hormone receptor + (%)

HER2 + (%)

Triple negative (%)

Interval of re-imaging (weeks)

Baselga et al. [22]

229

No

16%

75.5%

0

23.5%

6

693 736 348

No No Yes

Capecitabine and placebo Exemestane Fulvestranta Letrozole

PFS

Chia et al. [24] Di Leo et al. [23] Dickler et al. [12]

TTP PFS PFS

22.1% NR 37.9%

98.3% 100% 98%

NR NR 4.3%

NR 0 NR

8 12 9

Finn et al. [19]

666

No

23.6%

100%

0

0

12

424

No

Letrozole and placebo Trastuzumab and docetaxel

PFS

Gianni et al. [25]

PFS

15.3%

52.4%

100%

0

9

Hortobagyi et al. [20] Keller et al. [13]

668

Yes

PFS

22%

99.6%

0

NR

8

301

Yes

Capecitabine and sorafenib Fulvestrant Fulvestranta Letrozole and bevacizumab Letrozole and palbociclib Trastuzumab, docetaxel and bevacizumab Letrozole and ribociclib Liposomal doxorubicin

PFS

10%

47.5%

NR

NR

8

Metha et al. [14]

707

No

PFS

45.8%

100%

9.50%

0

12

Miles et al. [26]

736

No

PFS

16.6%

77.3%

0

22.7%

9

Miller et al. [21]

722

No

PFS

26.9%

61.3%

1.3%

NR

12

O’Brien et al. [15]

509

Yes

PFS

19.4%

37.7%

NR

NR

12

Schwartzberg et al. [16] Welt et al. [17]

219

Yes

PFS

15.5%

100%

100%

0

12

600

Yes

PFS

42.4%

79.2%

0.7%

20.60%

9

Xu et al. [27]

234

No

TTP

29.1%

100%

NR

0

NR

Yardley et al. [18]

724

Yes

PFS

30.7%

100%

0

0

6

Anastrozole and fulvestrant Docetaxel and bevacizumab Paclitaxel and bevacizumab Pegylated liposomal doxorubicin Letrozole and lapatinib Capecitabine, bevacizumab and vinorelbine Fulvestrant 250 mg Exemestane and everolimus

Letrozole and placebo Vinorelbine or mitomycin C and vinblastine Anastrozole Docetaxel and placebo Paclitaxel and placebo Doxorubicin

Letrozole and placebo Capecitabine and bevacizumab Anastrozole Exemestane and placebo

HER2 – human epidermal growth factor receptor 2, NR – not reported, PFS – progression free survival, TTP – time to progression. a The study investigated different doses of fulvestrant (500 mg versus 250 mg).

based on both blinded independent central review and investigator assessment. Among these, two studies showed lower effect sizes with blinded independent review [19,20] and two trials reported a higher effect size [18,25]. However, these differences did not meet statistical significance (see Supplementary Table S1). Only 2 trials reported OS in subgroups [13,22], therefore, no further analysis for OS was performed. Subgroup analyses based on experimental drug type Results of subgroup analyses based on experimental drug type are shown in Table 4. Among the 4 studies evaluating signal transduction inhibitors (all of which were added to endocrine therapy) [16,18–20], a higher effect of treatment on PFS was observed in patients with non-measurable disease (intra-study HR 0.74, 95% CI 0.59–0.94, p = 0.01, see Supplementary Fig. S1). Similar results were observed for the comparison of bone-only to non-bone-only disease (intra-study HR 0.73, 95% CI 0.58–0.92, p = 0.007), while the effect on non-visceral compared to visceral disease was comparable (intra-study HR 0.98, 95% CI 0.78–1.25, p = 0.90). Sensitivity analysis excluding the single study that included only patients with HER2-positive disease [16] showed similar results when comparing non-measurable and measurable disease (intra-study HR 0.71, 95% CI 0.56–0.91), bone-only and none bone-only disease (intra-study HR 0.72, 95% CI 0.56–0.92) and non-visceral to visceral disease (intra-study HR 0.86, 95% CI 0.67–1.11). Conversely, among the 5 studies evaluating anti-angiogenic agents (which were added to either chemotherapy [21,22,25,26]

or endocrine therapy [12]), a lesser effect in treatment on PFS was observed in patients with non-measurable disease (intrastudy HR 1.34, 95% CI 1.05–1.71, p = 0.02). Sensitivity analysis excluding the single study that included only patients with HER2 positive disease [25] showed similar results when comparing non-measurable and measurable disease (intra-study HR 1.27, 95% CI 0.98–1.65). Similar effect was observed for the comparison of bone-only and non-bone-only disease (intra-study HR 1.26, 95% CI 0.82–1.92, p = 0.29) [12,21] and in patients with non-visceral and visceral disease (intra-study HR 1.14, 95% CI 0.76–1.73, p = 0.53) [21,22]. There were no apparent differences in the magnitude of treatment effect on PFS between non-measurable compared to measurable disease in studies comparing different endocrine therapies [14,23,24,27] (intra-study HR 1.13, 95% CI 0.92–1.40, p = 0.23) or cytotoxic chemotherapeutics [13,15,17] (intra-study HR 0.73, 95% CI 0.44–1.21, p = 0.22). Treatment effect was also similar for patients with bone-only disease compared to patients with nonbone-only disease in studies of cytotoxic chemotherapy (intrastudy HR 0.71, 95% CI 0.45–1.10, p = 0.13). Only one study [14] comparing different endocrine therapies reported subgroup data in patients with bone-only and none-bone-only disease, thereby precluding further analysis of this subgroup. Discussion Most phase III RCTs in metastatic breast cancer use progressionbased primary endpoints rather than OS [1,2]. The surrogacy of

(0.51–0.93)

(0.57–2.01) (0.48–2.11) (0.41–1.55) (0.26–1.00)

(0.66–2.46) (0.27–2.06) (0.64–1.42)

– – – 1.43 0.55 – 1.28 0.75 0.95 – 1.07 1.01 0.80 0.51 – 0.69 (0.25–0.55)

(0.33–1.11) (0.42–1.68) (0.41–1.31) (0.24–0.88)

(0.38–1.25) (0.35–2.52) (0.54–1.11)

(0.59–1.64) (0.22–0.59)

– – – 0.69 (0.53–0.91) 0.65 (0.51–0.84) – 0.54 (0.41–0.72) 1.2 (0.85–1.6) 0.81 (0.68–0.97) – 0.57 (0.48–0.68) 1.01 (0.82–1.24) 0.70 (0.50–0.96) 0.91 (0.75–1.10) – 0.48 (0.43–0.61) – – – 0.99 0.36 – 0.69 0.90 0.77 – 0.61 0.86 0.73 0.46 – 0.33 1.21 (0.48–3.07) 1.24 (0.83–1.87) 0.87 (0.61–1.24) 1.43a (0.81–2.55) 0.53 (0.31–0.92) 1.94 (0.97–3.91) 1.28a (0.66–2.47) 0.78a (0.28–2.22) 1.35 (0.97–1.88) 1.24 (0.73–2.10) 1.24 (0.84–1.81) 0.85a (0.42–1.74) 0.80a (0.41–1.55) 0.40a (0.20–0.78) 3.86 (1.72–8.65) 0.69a (0.51–0.93) CI – confidence interval, HR – hazard ratio, PFS – progression-free survival. a Bone-only was used as a surrogate for non-measurable disease.

0.56 (0.39–0.80) 0.92 (0.77–1.12) 0.85 (0.70–1.03) 0.69a (0.53–0.91) 0.66 (0.52–0.85) 0.73 (0.57–0.93) 0.54a (0.41–0.72) 1.20a (0.85–1.6) 0.69 (0.55–0.86) 0.74 (0.61–0.90) 0.55 (0.46–0.67) 1.01a (0.82–1.24) 0.70a (0.50–0.96) 0.91a (0.75–1.10) 1.16 (0.80–1.69) 0.48a (0.43–0.61) 0.68 (0.29–1.58) 1.15 (0.8–1.64) 0.74 (0.56–1.00) 0.99a (0.59–1.64) 0.35 (0.22–0.57) 1.42 (0.74–2.73) 0.69a (0.38–1.25) 0.90a (0.35–2.52) 0.93 (0.73–1.19) 0.91 (0.58–1.6) 0.68 (0.49–0.95) 0.86a (0.42–1.68) 0.73a (0.41–1.31) 0.46a (0.24–0.88) 1.85 (0.98–3.90) 0.33a (0.25–0.55) 0.58 0.93 0.80 0.75 0.58 0.82 0.56 1.26 0.80 0.77 0.60 1.00 0.71 0.84 1.31 0.45 Baselga et al. [22] Chia et al. [24] Di Leo et al. [23] Dickler et al. [12] Finn et al. [19] Gianni et al. [25] Hortobagyi et al. [20] Keller et al. [13] Metha et al. [14] Miles et al. [26] Miller et al. [20] O’Brien et al. [15] Schwartzberg et al. [16] Welt et al. [17] Xu et al. [27] Yardley et al. [18]

(0.41–0.81) (0.82–1.13) (0.68–0.94) (0.59–0.96) (0.46–0.72) (0.65–1.02) (0.43–0.72) (0.98–1.62) (0.68–0.94) (0.64–0.93) (0.51–0.7) (0.82–1.22) (0.53–0.96) (0.70–1.01) (0.95–1.82) (0.38–0.54)

PFS non-bone-only HR (95% CI) PFS bone-only HR (95% CI) Treatment effect on PFS in non-measurable vs. measurable disease HR (95% CI) PFS measurable disease HR (95% CI) PFS non-measurable disease HR (95% CI) PFS all HR (95% CI) Trial

Table 2 Individual trial hazard ratios for PFS for measurable and non-measurable, bone-only and non-bone-only.

(0.81–2.55) (0.32–0.96)

H. Goldvaser et al. / Cancer Treatment Reviews 59 (2017) 46–53

Treatment effect on PFS in non-bone-only vs. bone-only HR (95% CI)

50

progression-based endpoints with OS has not been established [1,2,28], and the value of such endpoints as measures of patient benefit is uncertain. In several studies, even a large magnitude of improvement in PFS has not translated into improved OS [1,28]. Moreover, progression-based endpoints are limited by intra- and inter-observer variability. Such variability is challenging particularly in patients with non-measurable or bone-only disease. In this article we have shown that of the 148 phase III trials in metastatic breast cancer published over the last 15 years, over half included patients with non-measurable disease. Among these, only a small proportion reported outcome in subgroups of patients defined by presence or absence of measurable disease. This finding is of interest in view of data which show that there are inconsistencies in the effect of treatment on PFS among patients with measurable and non-measurable disease especially among those with bone-only and non-bone-only disease. Of interest, inconsistencies in treatment effect were seen more often in trials of signal transduction inhibitors when added to endocrine therapy where patients with non-measurable disease seemed to gain a greater treatment effect. Conversely, in trials of antiangiogenic drugs a lower magnitude of treatment effect was observed in those with non-measurable disease. No difference in treatment effect was observed in trials comparing two types of similar therapy (e.g. two regimens of cytotoxic chemotherapy or two types of endocrine therapy). There are a number of possible explanations for our findings. First, it is possible that the different treatment effects are influenced by the differing underlying biology of breast cancer which results in measurable and non-measurable disease. However, a substantial inter-observer variability in PFS assessment in patients with non-measurable disease was reported in some trials [8,19] suggesting that this is less likely. Another possible explanation is that targeted therapy (whether signal transduction inhibitors or anti-angiogenic agents) when combined with standard therapy can have a variable effect on different tissues. The effect of these drugs on the interpretation of imaging of metastatic bone disease is therefore of interest. Radiologic and scintigraphic changes in bone reflect the reaction of native bone cells to cancer rather than being a direct measure of cancer burden [10,11]. Some targeted agents such as everolimus can have an effect on bone which is independent of their anti-cancer effect [29]. It is possible that changes seen in bone disease after exposure to targeted drugs may reflect changes to normal bone biology rather than an anticancer effect. Similarly, the lower magnitude of treatment effect of anti-angiogenic agents in bone-only and non-measurable disease may reflect their lack of effect on bone remodeling. However, the differing mechanism of action of drugs evaluated in the included studies (e.g. CDK4/6 inhibitors, HER2 small molecule inhibitors and mTOR inhibitors) make it less likely that a consistent differential effect on different target tissues would occur. Improvement in PFS should be judged in caution. New anticancer agents that led to improvements in progression-based end points and were approved by the US Food and Drug Administration (FDA) between 2000 and 2010 were found to increase morbidity and treatment-related mortality [30]. Additionally, many of these drugs do not improve OS. For example, bevacizumab received accelerated approval by the FDA in 2008 for first-line treatment in HER2 negative metastatic breast cancer in combination with paclitaxel, based on significant improvement in PFS compared to paclitaxel and placebo [21]. Confirmatory studies failed to show that the addition of bevacizumab to chemotherapy improved OS [26,31], although analysis for OS was limited by a high proportion of cross-over as well as variability in the duration of survival postprogression [32,33]. Additionally, the magnitude of improvement in PFS in confirmatory studies was much smaller while toxicity remained substantial. As a result, in 2011 the FDA removed the

51

H. Goldvaser et al. / Cancer Treatment Reviews 59 (2017) 46–53

Fig. 2. Forest plots of progression-free survival. (A) Showing comparison between non-measurable and measurable disease. (B) Showing comparison between bone-only and none-bone-only disease.

Table 3 Results of meta-regression exploring the association of PFS effect of pre-specified variables. b

Variable PFS effect Proportion of patients with non-measurable disease Proportion of patients with hormone receptor positive disease Interval of re-imaging

p 0.220 0.041

0.45 0.88

0.033

0.91

PFS – progression-free survival.

metastatic breast cancer indication from the bevacizumab label [34]. In our analysis, 4 studies explored the addition of bevacizumab to standard treatment. Patients with non-measurable

disease comprised between 15% and 38% of included patients in these trials and in these patients, there appeared to be a lower effect of treatment on PFS. This finding may explain some of the variability in effect of bevacizumab observed in different trials. Interpretation of changes in bone imaging has been identified as challenging in prostate cancer. Similar to data in breast cancer, progressive disease and flare resulting from early treatment efficacy can be hard to distinguish in prostate cancer [10,11]. To overcome the problem of inaccurate classification of progressive disease, the Prostate Cancer Working Group 3 (PCWG3) has recommended that progression be confirmed with a repeat assessment according to standardized criteria [35]. Consideration of treatment failure during the first 12 weeks, even in the presence of radiological changes, is discouraged and confirmation of progression

Table 4 Intra-study comparison, according to evaluated drug mechanism. Cohort/investigational drug

No. studies included

Measurable HR (95% CI)

Non measurable HR (95% CI)

Intra- study comparison HR (95% CI)

P – for intra-study comparison

All Chemotherapy Endocrine treatment Signal transduction inhibitors Anti-angiogenetic agents

16 3 4 4 5

0.69 0.99 0.86 0.52 0.66

0.72 0.67 0.94 0.41 0.84

1.01 0.73 1.13 0.74 1.34

0.82 0.22 0.23 0.01 0.02

CI – confidence interval, HR – hazard ratio.

(0.65–0.73) (0.87–1.13) (0.77–0.96) (0.48–0.57) (0.59–0.73)

(0.64–0.80) (0.44–1.02) (0.80–1.10) (0.33–0.50) (0.67–1.04)

(0.89–1.15) (0.44–1.21) (0.92–1.40) (0.59–0.94) (1.05–1.71)

52

H. Goldvaser et al. / Cancer Treatment Reviews 59 (2017) 46–53

mandates evidence of new lesions in 2 sequential scans [35]. Similar to prostate cancer, a significant proportion of patients with metastatic breast cancer have bone-only disease, and they are often included in phase III trials. Our finding of variable effect of investigational drugs on PFS in patients with bone-only compared to non-bone-only disease, support the challenge in evaluating patients with bone-only disease. Adopting a modification of the PCWG3 criteria for the assessment of bone disease progression in metastatic breast cancer patients should be considered. Our results are also of relevance when attempting to translate results of clinical trials into clinical practice. RCTs are the gold standard tool for establishing the benefit of any new treatment. However, differences in characteristics between patients treated on trials and those treated in routine practice as well as differences in follow-up (e.g. differences in the frequency of radiologic assessments of disease status) mean that benefit observed in clinical trials may not translate into real world benefit. Additionally, clinicians may use different parameters to determine treatment failure (e.g. symptom burden in place of imaging or non-RECIST-based imaging assessments) in routine practice than are allowed on clinical trials. These may lead to additional differences in outcomes observed between clinical trials and clinical practice. Clinicians should tailor treatment decision, including assessment of treatment failure based on individual patients and include comorbidities, burden of the disease and tolerability of treatment. Our study has several limitations. First, this is a literature-based rather than an individual patient-based meta-analysis. Consequently, it is subject to publication bias. Additionally, although patients with non-measurable metastatic breast cancer are often included in RCTs, only a minority of trials reported outcomes in subgroups defined by measurable disease. This may not be representative of all trials. Second, this meta-analysis is heterogeneous. Trials included different sub-types of breast cancer and variable proportions of patients with non-measurable disease. Moreover, the investigational drugs and the re-imaging interval in each trial were different. However, meta-regression showed that neither the proportion of patients with non-measurable disease or hormone receptor positive disease nor the interval of re-imaging influenced results. In addition, sensitivity analyses excluding studies that included only patients with HER2 positive disease did not affect results. One factor which did influence results was the type of experimental drug therapy with greater variability seen in studies adding targeted therapy to either chemotherapy or endocrine therapy. Third, as only two trials reported OS results in subgroups defined by measurable disease [13,22], analysis of treatment effect on OS was not possible. Finally, despite plans to explore intrastudy differences by investigating differences between investigator assessment of PFS and that of blinded independent central review, such data were available only in 4 trials [18–20,25] so were explored descriptively only. In conclusion, there is significant variability in treatment effect on PFS in patients with measurable and non-measurable disease. Patients with non-measurable disease had greater effect on PFS when treated with signal transduction inhibitors and decreased effect on PFS with the addition of anti-angiogenic therapy to standard treatment. Furthermore, patients with bone-only disease had greater treatment effect on PFS compared to non-bone-only. These inconsistencies emphasize the need for standardization of PFS determination in patients with non-measurable and bone-only disease. Rethinking the assessment and definition of bone progression in breast cancer patients is warranted. Moreover, newer approaches to endpoint selection such as composite outcome endpoints comprising radiologic progression and changes in quality of life should be considered as they may better represent net treatment effect.

Conflict of interests The authors declare no potential conflict of interest.

Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ctrv.2017.06.007. References [1] Verma S, McLeod D, Batist G, Robidoux A, Martins IR, Mackey JR. In the end what matters most? A review of clinical endpoints in advanced breast cancer. Oncologist 2011;16(1):25–35. [2] Saad ED, Katz A, Buyse M. Overall survival and post-progression survival in advanced breast cancer: a review of recent randomized clinical trials. J Clin Oncol 2010;28(11):1958–62. [3] Warr D, McKinney S, Tannock I. Influence of measurement error on assessment of response to anticancerchemotherapy: proposal for new criteria of tumor response. J Clin Oncol 1984;2(9):1040–6. [4] Hopper KD, Kasales CJ, Van Slyke MA, Schwartz TA, TenHave TR, Jozefiak JA. Analysis of interobserver and intraobserver variability in CT tumor measurements. AJR Am J Roentgenol 1996;167(4):851–4. [5] Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45(2):228–47. [6] Tang PA, Pond GR, Chen EX. Influence of an independent review committee on assessment of response rate and progression-free survival in phase III clinical trials. Ann Oncol 2010;21(1):19–26. [7] Perez EA, Lerzo G, Pivot X, Thomas E, Vahdat L, Bosserman L, et al. Efficacy and safety of ixabepilone (BMS-247550) in a phase II study of patients with advanced breast cancer resistant to an anthracycline, a taxane, and capecitabine. J Clin Oncol 2007;25(23):3407–14. [8] Center for drug evaluation and research application number: 207103Orig1s000. . [9] US Food and Drug Administration: guidance for industry clinical trial endpoints for the approval of cancer drugs and biologics. Washington (DC): US Food and Drug Administration; 2007. p. 1–19. Available at: . [10] Bares R. Skeletal scintigraphy in breast cancer management. Q J Nucl Med 1998;42(1):43–8. [11] Body JJ. Metastatic bone disease: clinical and therapeutic aspects. Bone 1992;13(suppl 1):S57–62. [12] Dickler MN, Barry WT, Cirrincione CT, Ellis MJ, Moynahan ME, Innocenti F, et al. Phase III trial evaluating letrozole as first-line endocrine therapy with or without bevacizumab for the treatment of postmenopausal women with hormone receptor-positive advanced-stage breast cancer: CALGB 40503 (Alliance). J Clin Oncol 2016;34(22):2602–9. [13] Keller AM, Mennel RG, Georgoulias VA, Nabholtz JM, Erazo A, Lluch A, et al. Randomized phase III trial of pegylated liposomal doxorubicin versus vinorelbine or mitomycin C plus vinblastine in women with taxanerefractory advanced breast cancer. J Clin Oncol 2004;22(19):3893–901. [14] Mehta RS, Barlow WE, Albain KS, Vandenberg TA, Dakhil SR, Tirumali NR, et al. Combination of anastrazole and fluvestrant in metastatic breast cancer. N Engl J Med 2012;367(5):435–44. [15] O’Brien ME, Wigler N, Inbar M, Rosso R, Grischke E, Santoro A, et al. Reduced cardiotoxicity and comparable efficacy in a phase III trial of pegylated liposomal doxorubicin HCL (CAELYX/Doxil) versus conventional doxorubicin for firstline treatment of metastatic breast cancer. Ann Oncol 2004;15(3):440–9. [16] Schwartzberg LS, Franco SX, Florance A, O’Rourke L, Maltzman J, Johnston S, et al. Lapatinib plus letrozole as first-line therapy for HER-2+ hormone receptorpositive metastatic breast cancer. Oncologist 2010;15(2):122–9. [17] Welt A, Marschner N, Lerchenmueller C, Decker T, Steffens CC, Koehler A, et al. Capecitabine and bevacizumab with or without vinorelbine in firstline treatment of HER2/neu-negative metastatic or locally advanced breast cancer: final efficacy and safety data of the randomised, open-label superiority phase 3 CARIN trial. Breast Cancer Res Treat 2016;156(1):97–107. [18] Yardley DA, Noguchi S, Pritchard KI, Burris 3rd HA, Baselga J, Gnant M, et al. Everolimus plus exemestane in postmenopausal patients with HR(+) breast cancer: BOLERO-2 finalprogression-free survival analysis. Adv Ther 2013;30 (10):870–84. [19] Finn RS, Martin M, Rugo HS, Jones SI, Im SA, Gelmon K, et al. Palbociclib and letrozole in advanced breast cancer. N Engl J Med 2016;375(20):1925–36.

H. Goldvaser et al. / Cancer Treatment Reviews 59 (2017) 46–53 [20] Hortobagyi GN, Stemmer SM, Burris HA, Yap YS, Sonke GS, Paluch-Shimon S, et al. Ribociclib as first-line therapy for HR-positive, advanced breast cancer. N Engl J Med 2016;375(18):1738–48. [21] Miller K, Wang M, Gralow J, Dickler M, Cobleigh M, Perez EA, et al. Paclitaxel plus bevacizumab versus paclitaxel alone for metastatic breast cancer. N Engl J Med 2007;357(26):2666–76. [22] Baselga J, Segalla JG, Roché H, Del Giglio A, Pinczowski H, Ciruelos EM, et al. Sorafenib in combination with capecitabine: an oral regimen for patients with HER2negative locallyadvanced or metastatic breast cancer. J Clin Oncol 2012;30 (13):1484–91. [23] Di Leo A, Jerusalem G, Petruzelka L, Torres R, Bondarenko IN, Khasanov R, et al. Results of the CONFIRM phase III trial comparing fulvestrant 250 mg with fulvestrant 500 mg in postmenopausal women with estrogen receptor positive advanced breast cancer. J Clin Oncol 2010;28(30):4594–600. [24] Chia S, Gradishar W, Mauriac L, Bines J, Amant F, Federico M, et al. Doubleblind randomized placebo controlled trial of fulvestrant compared with exemestane after priornonsteroidal aromatase inhibitor therapy in postmenopausal women with hormone receptor-positive, advanced breast cancer: results from EFECT. J Clin Oncol 2008;26(10):1664–70. [25] Gianni L, Romieu GH, Lichinitser M, Serrano SV, Mansutti M, Pivot X, et al. AVEREL: randomized phase III trial evaluating bevacizumab in combination with docetaxel and trastuzumab as first-line therapy for HER2-positive locally recurrent/metastatic breast cancer. J Clin Oncol 2013;31(14):1719–25. [26] Miles DW, Chan A, Dirix LY, Cortés J, Pivot X, Tomczak P, et al. Phase III study of bevacizumab plus docetaxel compared with placebo plus docetaxel for the first-line treatment of human epidermal growth factor receptor 2negative metastatic breast cancer. J Clin Oncol 2010;28(20):3239–47.

53

[27] Xu B, Jiang Z, Shao Z, Wang J, Feng J, Song S, et al. Fulvestrant 250 mg versus anastrazole for Chinese patients with advanced breast cancer: results of a multicentre, double-blind, randomized phase III. Cancer Chemother Pharmacol 2011;67(1):223–30. [28] Tannock IF, Amir E, Booth CM, Niraula S, Ocana A, Seruga B, et al. Relevance of randomised controlled trials in oncology. Lancet Oncol 2016;17(12):e560–7. [29] Hadji P, Coleman R, Gnant M. Bone effects of mammalian target of rapamycin (mTOR) inhibition with everolimus. Crit Rev Oncol Hematol 2013;87 (2):101–11. [30] Niraula S, Seruga B, Ocana A, Shao T, Goldstein R, Tannock IF, et al. The price we pay for progress: a meta-analysis of harms of newly approved anticancer drugs. J Clin Oncol 2012;30(24):3012–9. [31] Robert NJ, Diéras V, Glaspy J, Brufsky AM, Bondarenko I, Lipatov ON, et al. RIBBON-1: randomized, double-blind, placebo-controlled, phase III trial of chemotherapy with or without bevacizumab for first-line treatment of human epidermal growth factor receptor 2- negative, locally recurrent or metastatic breast cancer. Clin Oncol 2011;29(10):1252–60. [32] Broglio KR, Berry DA. Detecting an overall survival benefit that is derived from progression-free survival. J Natl Cancer Inst 2009;101(23):1642–9. 2. [33] Cortés J, Calvo E, González-Martín A, Dawood S, Llombart-Cussac A, De Mattos-Arruda L, et al. Progress against solid tumors in danger: the metastatic breast cancer example. J Clin Oncol 2012;30(28):3444–7. [34] Carpenter D, Kesselheim AS, Joffe S. Reputation and precedent in the bevacizumab decision. N Engl J Med 2011;365(2):e3. [35] Scher HI, Morris MJ, Stadler WM, Higano C, Basch E, Fizazi K, et al. Trial design and objectives for castration-resistant prostate cancer: updated recommendations from the prostate cancer clinical trials working group 3. J Clin Oncol 2016;34(12):1402–18.