Event-Free Survival in Relapsed and Refractory Rhabdomyosarcoma Treated on Cooperative Group Phase II Trials: A Report from the Children’s Oncology Group

Abstract

Background:

Novel therapies are needed for relapsed and refractory rhabdomyosarcoma (RRMS). Phase II clinical trials in RRMS have typically utilized radiologic response as the primary activity endpoint, an approach which poses several limitations in RRMS. In this analysis, we aimed to estimate an event-free survival (EFS) endpoint for RRMS that could be used as a benchmark for future studies.

Procedure:

We performed a retrospective study of patients with RRMS enrolling on 13 single-agent Phase II Children’s Oncology Group and legacy group trials from 1997 to 2016. All included trials used radiographic response as their primary activity endpoint. Six-month EFS was estimated from time of trial enrollment with 95% confidence intervals. Clinical characteristics, including trial of enrollment, sex, age, race, histology, number of prior chemotherapies, and radiographic response were evaluated for their impact on 6-month EFS.

Results:

We identified 175 patients across 13 trials. The 6-month EFS was 16.8% (11.6%-22.8%). No differences were seen in 6-month EFS based on age, sex, race, or histology. There were nonsignificant trends towards improved 6-month EFS for patients with ≤ 2 prior lines of therapy vs. > 2, for patients enrolled on trials that achieved their primary radiographic response endpoint vs. trials that did not, and for patients who achieved complete or partial response compare to those achieving stable disease.

Conclusions:

The prognosis of RRMS enrolled on single-agent Phase II trials is poor. This pooled 6-month EFS of RRMS on single-agent trials may be used as a RRMS-specific benchmark for future single-agent Phase II trials.

Precis:

Using Children’s Oncology Group and legacy group single-agent, nonrandomized Phase II trials, we estimated the 6-month event-free survival for patients with relapsed and refractory rhabdomyosarcoma. This data may serve as a rhabdomyosarcoma-specific historical benchmark for future Phase II trials.

BACKGROUND:

While rhabdomyosarcoma (RMS) cure rates exceed 70%, certain subsets of patients remain at higher risk of recurrence, including those with unfavorable histology, primary disease at unfavorable sites, and metastatic disease at diagnosis. ^1,2 Relapsed and refractory RMS (RRMS), excluding patients with initially low-risk features, carries a dismal prognosis, with a 5-year overall survival (OS) from first recurrence of less than 20%. ³ Ongoing investigation of novel therapies and treatment strategies is essential to improve survival for RRMS.

The Children’s Oncology Group (COG) and prior legacy groups (Children’s Cancer Group and Pediatric Oncology Group) have enrolled RRMS-specific cohorts on a succession of Phase II clinical trials to identify active agents worthy of further study in combination for RRMS or in the newly-diagnosed setting. In trials that have evaluated the activity of single agents in RRMS, the primary activity endpoint used has been objective radiographic response (ORR) based on World Health Organization (WHO) or Response Evaluation Criteria in Solid Tumors (RECIST, original or version 1.1) guidelines. ^4–6 Fourteen of these trials have been published since 1997. RRMS cohorts enrolled on these trials have only achieved the primary activity endpoint in two trials (P9963, which evaluated a rebeccamycin analogue and A09705, which studied vinorelbine). ^7,8 While the myelosuppressive toxicity seen with rebeccamycin analogue halted its further development, vinorelbine has been further developed in multiple RMS COG studies: a relapsed combination trial (ARST0921), a randomized phase 3 intermediate-risk trial where it was incorporated into maintenance chemotherapy with oral cyclophosphamide (ARST1431), and an ongoing high-risk randomized phase III clinical trial (ARST2031). ^2,9

The use of radiologic response as an activity endpoint in RRMS provides multiple limitations. First, several studies have shown conflicting findings between the relationship of an early radiographic response and long-term outcome in newly-diagnosed RMS. Studies from the Istituto Nazionale Tumori and the Cooperative Weichteilsarkom Studiengruppe showed early radiographic response as a favorable prognostic factor in RMS, however analyses of the Intergroup Rhabdomyosarcoma Study IV (IRS-IV) and the COG study, D9803, confirmed that early radiographic response to chemotherapy did not predict 5-year FFS. ^10–13 Likewise, a systematic review concluded that there was no evidence to support degree of response as a prognostic marker for survival in RMS, and future trials should not utilize early radiologic response to guide treatment escalation. ¹⁴ Second, the benchmark ORR rates required to demonstrate activity on COG/legacy group single agent Phase II studies have differed between trials. The benchmark ORR on these trials has also not been RRMS specific, as these trials have typically applied the same ORR activity threshold across disease histologies. Third, ORR as an activity endpoint does not account for the clinical benefit of prolonged stable disease. Finally, sarcoma-specific endpoints based on event-free survival for osteosarcoma and Ewing sarcoma have been published and have served as useful historical comparisons for subsequent sarcoma trials. ^15–19 Therefore, the primary aim of this study was to establish a benchmark EFS in patients with RRMS previously treated on COG or legacy group single-agent Phase II studies for use as an activity endpoint in future investigations.

PROCEDURES:

Patient Selection:

Patients with RRMS enrolled on completed Phase II single agent COG/legacy trials from 1997–2016 who had available event data were included in this analysis. The eligibility criteria for these studies have been previously published, and study details are outlined in Table 1. ^20–32 Informed consent/assent from the patient and/or parent/guardian, as appropriate, was obtained before enrollment, and all trials were approved by the institutional review board of participating institutions or by the Pediatric Central Institutional Review Board of the National Cancer Institute.

Table 1:

Characteristics of the 13 included single-agent phase II trials enrolling relapsed rhabdomyosarcoma.

Study	Years Open	Agent(Dose)	Radiographic Endpoint Criteria	Activity Endpoint Met?
CCG-0962	1997–2001	Docetaxel 125 mg/m2 every 21 days	WHO	No
A09705	1998–2002	Vinorelbine 33.75 or 30 mg/m2 weekly for 6 of every 8 weeks	WHO	Yes
A09713	1999–2003	Topotecan 0.3 mg/m2 continuous for 21 days every 28 days	WHO	No
P9761	1999–2001	Irinotecan 50mg/m2/day for 5 days every 21 days	WHO	No
P9963	2000–2004	Rebeccamycin analog 650 mg/m2 once every 21 days	RECIST	Yes
ADVL0221	2008–2010	Trabectedin 1.5 mg/m2 once every 21 days	RECIST	No
ADVL0421	2004–2006	Oxaliplatin 130 mg/m2 once every 21 days	RECIST	No
ADVL0524	2006–2007	Ixabepilone 8 mg/m2/day for 5 days every 21 days	RECIST	No
ADVL0525	2007–2009	Pemetrexed 1,910 mg/m2 once every 21 days	RECIST	No
ADVL0821	2009–2012	Cixutumumab 9 mg/kg weekly every 28 days	RECIST	No
ADVL0921	2011–2013	Alisertib 80 mg/m2 for 7 days every 21 days	RECIST 1.1	No
ADVL1121	2012–2013	Sorafenib 200 mg/m2/dose twice daily every 28 days	RECIST 1.1	No
ADVL1522	2015–2016	Lorvotuzumab, 110 mg/m2 on days 1 and 8 every 21 days	RECIST 1.1	No

For all trials included in the analysis, the primary outcome measure was radiographic response (WHO, RECIST, or RECIST 1.1). All patients with available response data were identified as a best response of complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD). Patients with response assessments deemed as having a “minor response” were reclassified as SD. For all patients, event information was collected prospectively, and patients were observed for status until disease progression, death, loss to follow-up, withdrawal of consent for further data submission, or for a minimum of five years after enrollment (whichever occurred first). Studies included used either two-stage or three-stage designs with null response rates ranging from 5% to 15% and alternative response rates ranging from 25 to 40%. The trials were designed to maintain a type I error rate of no more than 15% and a power of at least 80%.

For a comparative analysis, based on achieving the study’s prospective radiographic response endpoints, two trials were deemed “active” (A09705 using vinorelbine and P9963 using rebeccamycin), while all other trials were considered “inactive”. Other potential prognostic factors analyzed for their influence on EFS included specific trial enrolled upon, age at diagnosis (<1, 1–9, ≥10 years), age at trial enrollment (<1, 1–9, ≥10 years), race (white vs. nonwhite) sex (male vs. female), histology (embryonal, alveolar, spindle cell, NOS), and best response (CR, PR, SD, PD). For most patients, lines of prior chemotherapy were identified from the trial inclusion/exclusion criteria while for some patients it was found in the case reports forms. The number of prior chemotherapies were then coded as ≤2 or >2.

Statistical Methods:

All RRMS patients enrolled and initiated trial therapy were included in this analysis, including those not evaluable for the primary trial endpoint. EFS was defined as the time from trial enrollment until disease progression, death, or date of last contact. Those with disease progression or death were considered to have an event, while those without progression or death were censored at date of last follow-up. The Kaplan-Meier method was used to estimate event-free survival and the loglog transformation was used to estimate the 95% confidence interval of the Kaplan-Meier estimate. Univariate analysis of 6-month EFS and potential prognostic clinical factors was performed using the logrank test. P values of ≤0.05 were considered statistically significant. SAS 9.4 was used to conduct all analyses.

RESULTS

Thirteen trials met inclusion criteria which enrolled 176 patients with RRMS from 1997–2016. One patient enrolled but never received treatment and was excluded, leaving a total of 175 patients for analysis. Nine patients were enrolled on more than one trial. Each enrollment was retained as an independent observation for the purposes of this analysis. Overall patient demographics and clinical characteristics of patients are listed in Table 2.

Table 2:

Trial and Patient Characteristics and Event-Free Survival

		RRMS Patients		6-Month EFS (95% CI)	P Value*
	Variable	Number	(%)
Total Patients		175	100
Trial	CCG-0962	11	6.3	27.3% (6.5%, 53.9%)	<0.001
	A09705	11	6.3	34.1% (9.1%, 61.6%)
	A09713	9	5.1	11.1% (0.6%, 38.8%)
	P9963	21	12	25% (9.1%, 44.9%)
	P9761	19	10.9	15.8% (3.9%, 34.9%)
	ADVL0221	22	12.6	22.7% (8.3%, 41.4%)
	ADVL0421	10	5.7	0
	ADVL0524	10	5.7	0
	ADVL0525	9	5.1	0
	ADVL0821	17	9.7	35.3% (14.5%, 57%)
	ADVL0921	10	5.7	10% (0.6%, 35.8%)
	ADVL1121	10	5.7	0
	ADVL1522	16	9.1	6.3% (0.4%, 24.7%)
Sex	Male	86	49.1	14.4% (7.9%, 22.7%)	0.409
	Female	89	50.9	19.1% (11.6%, 27.9%)
Age, Diagnosis (years)	< 1	3	1.7	33.3% (0.9%, 77.4%)	0.798
	1 – 9	78	44.6	16.9% (9.4%, 26.1%)
	≥10	93	53.1	16.4% (9.7%, 24.6%)
	Unknown	1	0.6	0
Age, Enrollment (years)	< 1	0	0	0	0.825
	1–9	60	34.3	15.2% (7.3%, 25.7%)
	≥10	115	65.7	17.6% (11.3%, 25.1%)
Race	Nonwhite	36	20.6	20.8% (9.2%, 35.6%)	0.497
	White	126	72	15.8% (10.1%, 22.8%)
	Unknown	13	7.4	15.4% (2.5%, 38.8%)
Histology	Alveolar	79	45.1	15.3% (8.2%, 24.3%)	0.781
	Embryonal/Botryoid	32	18.3	9.7% (2.5%, 23.0%)
	NOS	39	22.3	20.5% (9.6%, 34.2%)
	Spindle Cell	4	2.3	0
	Unknown	21	12	28.6% (11.7%, 48.2%)
Number of Prior Lines of Chemotherapy Agents	≤2 lines	88	50.3	21.1% (13.1%, 30.3%)	0.22
	>2 lines	28	16	7.1% (1.3%, 20.4%)
	Unknown	59	33.7	15.3% (7.5%, 25.5%)
Best Response	Not Evaluable	1	0.6	NA	<0.001
	Progressive Disease	92	52.6	5.4% (2.0%, 11.4%)
	Stable Disease	32	18.3	34.7% (18.7%, 51.4%)
	Partial Response	12	6.9	50% (20.8%, 73.6%)
	Complete Response	4	2.3	75% (12.8%, 96.1%)
	Unknown	34	19.4	12.2% (3.8%, 25.6%)

Abbreviations: RRMS; relapsed and refractory rhabdomyosarcoma, EFS; event-free survival

^*Logrank test (unknowns are excluded from test)

Of the 175 patients included in the analysis, 145 (82.9%) experienced disease progression as first event, 24 (13.4%) experienced death as first event, and 6 (3.4%) had no event at last follow up. For the 24 patients with death as first event, 5 were removed from study for treatment-related toxicities, but none died of treatment-related toxicity. The 6-month EFS for the entire cohort was 16.8% (95% CI 11.6–22.8%, Figure 1A)

Figure 1:

Event-Free Survival in Relapsed and Refractory Rhabdomyosarcoma on Single-Agent Phase II Trials. A) Kaplan-Meier curve with 6-month EFS for the entire cohort. B) Kaplan-Meier curve with 6-month EFS for patients enrolled on active (red) vs. inactive (green) trials. C) Kaplan-Meier curve with 6-month EFS for patients experiencing CR/PR (red) vs. SD (green). Abbreviations: EFS; event-free survival, CR; complete response, PR; partial response, SD; stable disease.

Two trials (A09705; vinorelbine, P9963; rebeccamycin analogue) met their response-based primary endpoint and were deemed “active” in RMS, while the other 11 trials did not and were deemed inactive in RMS. Patient clinical characteristics between active and inactive trials were similar (Table 3). The 6-month EFS of the 43 patients enrolled on the active trials was 28.7% (95% CI 14.1%-45%) vs. 14.2% (95% CI 9.1%-20.5%) for the 143 patients enrolled on inactive trials (p=0.08, Figure 1B)

Table 3:

Clinical Characteristics of patients enrolled on trials by trial-specified outcome (active or inactive) and best response (CR/PR, SD, or PD)

		Active Trials		Inactive Trials		CR/PR		SD		PD
	Variable	Number	(%)	Number	(%)	Number	(%)	Number	(%)	Number	%
	Total Patients	32	-	143	-	16	-	32	-	92	-
Sex	Male	19	59.4	67	46.9	7	43.8	12	37.5	50	54.3
	Female	13	40.6	76	53.1	9	56.3	20	62.5	42	45.7
Age, Diagnosis (years)	< 1	1	3.1	2	1.4	1	6.3	0	0	1	1.1
	1 – 9	12	37.5	66	46.2	7	43.8	13	40.6	41	44.6
	≥10	19	59.4	74	51.7	8	50	19	59.4	49	53.3
	Unknown	0	0	1	0.7	0	0	0	0	1	1.1
Age, Enrollment (years)	1–9	10	31.2	50	35	6	37.5	6	18.8	32	34.8
	≥10	22	68.8	93	65	10	62.5	26	81.3	60	65.2
Race	Nonwhite	4	12.5	32	22.4	0	0	7	21.9	16	17.4
	White	27	84.4	99	69.2	15	93.8	21	65.6	69	75
	Unknown	1	3.1	12	8.4	1	6.3	4	12.5	7	7.6
Histology	Alveolar	12	37.5	67	46.9	8	50	17	53.1	47	51.1
	Embryonal/Botryoid	7	21.9	25	17.5	3	18.8	6	18.8	13	14.1
	NOS	4	12.5	35	24.5	3	18.8	5	15.6	23	25
	Spindle cell	3	9.4	1	0.7	0	0	1	3.1	3	3.3
	Unknown	6	18.8	15	10.5	2	12.5	3	9.4	6	6.5
Number of Prior Lines of Chemotherapy Agents	≤2 lines	32	100	56	39.2	11	68.8	18	56.3	40	43.5
	>2 lines	0	0	28	19.6	0	0	4	12.5	21	22.8
	Unknown	0	0	59	41.3	5	31.3	10	31.3	31	33.7
Best Response	Progressive Disease	14	43.8	78	54.5	0	0	0	0	92	100
	Stable Disease	10	31.3	22	15.4	0	0	32	100	0	0
	Partial Response	5	15.6	7	4.9	12	75	0	0	0	0
	Complete Response	2	6.3	2	1.4	4	25	0	0	0	0
	Not Evaluable	1	3.1	0	0	0	0	0	0	0	0
	Unknown	0	0	34	23.8	0	0	0	0	0	0

Abbreviations: CR: complete response, PR: partial response, SD: stable disease, PD: progressive disease, NOS: not otherwise specified

Four patients experienced a CR, 12 patients had a PR, 32 patients had SD, and 92 experienced PD. No major differences in clinical characteristics were noted between patients experiencing a response and those with SD or PD (Table 3). The 6-month EFS for the 16 patients with a radiographic response (CR+PR) was 56.3% (95%CI 29.5%-76.2%), while the 6-month EFS for 32 patients with SD was 34.7% (18.7%-51.4%, p=0.32, Figure 1C).

Six-month EFS was also estimated according to clinical and demographic variables for the entire cohort (Table 2). Significant differences in 6-month EFS were noted based on trial and best response. No statistically significant differences were noted based on sex, age at diagnosis, age at trial enrollment, race, or histology. Specifically, the 6-month EFS for alveolar histology was 15.3% (95% CI 8.2–24.3%) and for embryonal/botryoid histology was 9.7% (95% CI 2.5%-23%; p= 0.77) A trend was noted toward difference in 6-month EFS based on number of prior therapies (≤ or > 2 lines of therapy), but this was not statistically significant (p=0.22).

DISCUSSION

As confirmed in this study, the prognosis of RRMS is extremely poor and necessitates the need for discovery and implementation of novel agents to improve outcomes. Phase II clinical trials are the cornerstone of agent activity determination within the COG and in other cooperative groups, and the rarity of RRMS requires the design of trials with endpoints that can rapidly determine the activity of novel agents with high confidence. Here, we have identified the 6-month EFS for RRMS across COG single-agent Phase II trials to be 16.8% (95% CI 11.6–22.8%) that may serve as an alternative endpoint for future Phase II single agent trial design.

COG/legacy trials based on ORR have previously identified two agents, vinorelbine (A09705) and rebeccamycin analogue (P9963), as active in RRMS, with vinorelbine now incorporated into multiple clinical trials for newly-diagnosed rhabdomyosarcoma. A comparison of these trials to those that deemed tested agents to be inactive in RRMS demonstrated a non-statistically significant trend towards improved 6-month EFS. It is likely that both vinorelbine and rebeccamycin analogue would have also met a 6-month EFS-based activity endpoint in adequately powered studies. However, it is noteworthy that while many nonactive trials had a 0% 6-month EFS, several had patients with 6-month EFS comparable to those of A9705 and P9963, most notably ARST0821 (cixutumumab), with a 6-month EFS 35%. It is conceivable that agents outside of vinorelbine and rebeccamycin analogue may have been deemed active using a 6-month EFS as an alternative endpoint to ORR.

The 6-month EFS of patients achieving a radiographic response was also investigated. Although no statistically-significant difference between patients achieving CR/PR vs. SD was seen, it is notable that patients achieving SD experienced a 6-month EFS of 34.7%, suggesting that while amongst patients only achieving SD as best response, a substantial portion maintained prolonged disease control.

There are several limitations to this analysis. First, there was heterogenous disease evaluation timing on these trials. Most treatment cycles were either 21 or 28 days, however A09705 (vinorelbine) was administered via 8-week cycles. It has been previously demonstrated that variability in schedules of patient evaluation can affect analysis of progression-free survival, as progression occurring between visits is usually assigned to the timepoint at when progression is detected. ³³ Therefore we analyzed 6-month EFS, a timepoint at which 97.6% of events had already occurred, to minimize the effect of variations in disease assessment on our estimate. Secondly, this analysis only includes Phase II trials of single agents. Despite overall poor prognosis, RRMS often remains chemotherapy-responsive with multiple, usually multi-agent, regimens especially in first recurrence. This was well demonstrated in the COG trials ARST0121, which found a 1-year FFS rate > 35% using vincristine, irinotecan, doxorubicin, cyclophosphamide, etoposide, and ifosfamide, and ARST0921, which reported an ORR of 47% and a 6-month EFS of 69.1% for patients receiving cyclophosphamide, vinorelbine, and temsirolimus. ^9,34 Based on these data, the 6-month EFS endpoints from the above analysis would not serve as appropriate primary activity endpoints for multiagent trials in first recurrence. Thirdly, the population of patients treated was heterogenous and heavily pretreated. Patients with alveolar histology were more heavily represented than embryonal histology in contrast those with newly diagnosed rhabdomyosarcoma and this should be considered when agents are deemed active in the recurrent setting and potentially incorporated into trials for newly-diagnosed disease. The prolonged timespan of enrollments from 1997–2016 also may have introduced increased heterogeneity into the data given shifts in treatment that may have occurred over these two decades, including chemotherapy, surgery and radiation approaches for both newly diagnosed and recurrent disease. Finally, important clinical and biologic factors in rhabdomyosarcoma that may affect EFS were not available for this analysis. Molecular findings are rapidly becoming integral to risk stratification for newly-diagnosed RMS, however their role in RRMS is not yet investigated. ³⁵ Given the timespan of this study (1997–2016) preceded widespread molecular profiling of RMS, data on the presence of PAX::FOXO1 translocations and other well-known and potentially prognostic mutations (MYOD1, TP53, DICER1, etc.) were not available for this analysis As our study pooled data from many Phase II trials that enrolled many tumor types, the clinical data collection tools were not individualized for RMS; thus other potentially meaningful clinical data not available for this analysis include initial primary site of disease, initial stage and group, disease response to initial therapy (for example, complete response followed by relapse vs primary refractory disease) and type of relapse (local/regional recurrence vs metastatic recurrence). Further investigation of these clinical and molecular factors in future studies of RRMS will be very valuable.

Our data confirm a low EFS of 16.8% for RRMS patients enrolled on single-agent Phase II trials. The 6-month EFS data in this analysis may serve as a RRMS-specific, clinically meaningful benchmark for the determination of single-agent activity in future trials in RRMS. Appropriately-powered Phase II trials using this benchmark may help address the critical need to identify novel agents with activity in RRMS to improve outcomes in both the recurrent and up-front setting.

Abbreviations Table

RMS

Rhabdomyosarcoma

RRMS

Relapsed and Refractory Rhabdomyosarcoma

WHO

World Health Organization

RECIST

Response Evaluation Criteria in Solid Tumors

EFS

Event-Free Survival

OS

Overall Survival

FFS

Failure-Free Survival

COG

Children’s Oncology Group

CR

Complete Response

PR

Partial Response

SD

Stable Disease

NE

Not Evaluable

PD

Progressive Disease