Reduction of cyclophosphamide dose for patients with subset 2 low-risk rhabdomyosarcoma is associated with increased risk of recurrence: A report from the Soft Tissue Sarcoma Committee of the Children’s Oncology Group

Abstract

Background

Failure-free survival (FFS) and overall survival (OS) rates improved on Intergroup Rhabdomyosarcoma Study (IRS)-IV compared with IRS-III for patients with subset 2 (stage 1, group III non-orbit or stage 3, group I/II) low-risk embryonal rhabdomyosarcoma (ERMS) with the addition of cyclophosphamide (total cumulative cyclophosphamide dose 26.4 g/m²) to vincristine and dactinomycin (VAC). The goal of Children’s Oncology Group ARST0331 for subset 2 low-risk patients was to reduce the total cumulative cyclophosphamide dose without compromising FFS.

Methods

Therapy included 4 VAC cycles (total cumulative cyclophosphamide dose 4.8 g/m²) followed by 12 cycles of vincristine and dactinomycin over 46 weeks. Patients with group II or III tumors received radiotherapy, except girls with group III vaginal tumors who enrolled before September 2009 and achieved a complete response with chemotherapy with or without delayed resection.

Results

Amongst 66 eligible patients that were followed for a median of 3.5 years, there were 20 failures vs. 10.53 expected failures. Estimated 3-year FFS and OS rates were 70% (95% CI: 57%, 80%) and 92% (95% CI: 83%, 97%), respectively. The estimated 3-year FFS was 57% (95% CI: 33%, 75%) for girls with subset 2 genital tract ERMS (n=21) and 77% (95% CI: 61%, 87%) for all other subset 2 patients (n=45) (p=0.02).

Conclusions

We observed suboptimal FFS of patients with subset 2 low-risk RMS using reduced total cyclophosphamide. Eliminating radiotherapy for girls with group III vaginal tumors in combination with reduced total cyclophosphamide contributed to the suboptimal outcome.

Introduction

The Soft Tissue Sarcoma Committee of the Children’s Oncology Group (COG) has defined two subsets of patients with low-risk rhabdomyosarcoma (RMS) each with an expected 5-year failure-free survival (FFS) rate of ≥ 83% (1). For patients in subset 1, this outcome was achieved with a chemotherapy regimen of vincristine and dactinomycin (VA) on Intergroup Rhabdomyosarcoma Study (IRS)-III (2). Subset 1 includes patients with localized embryonal RMS (ERMS) that 1) has been grossly resected before initiation of chemotherapy (group I/II) and arises in either a favorable primary site (stage 1; orbit, non-parameningeal head/neck, non-bladder/prostate genitourinary tract, or hepatobiliary) or in an unfavorable primary site (all sites that are not classified as favorable) and is ≤ 5 cm in size and without nodal involvement (stage 2), or 2) has not been resected before initiation of chemotherapy (group III) and arises in the orbit. For patients in subset 2, a 5-year FFS rate of ≥ 83% was achieved with vincristine, dactinomycin and cyclophosphamide (VAC, total cumulative cyclophosphamide dose 26.4 g/m²) on IRS-IV (3). Subset 2 includes patients with localized ERMS that 1) has been grossly resected before initiation of chemotherapy (group I/II) and arises in an unfavorable primary site and is > 5 cm in size or with regional nodal involvement (stage 3), or 2) has not been resected before initiation of chemotherapy (group III) and arises in a favorable primary site other than the orbit (stage 1 non-orbit). We previously reported outcome using shorter duration therapy for subset 1 low-risk patients on COG ARST0331 (4) and now report outcome for low-risk patients in subset 2 of ARST0331.

The 5-year FFS rate for subgroup B low-risk patients on Intergroup RMS Study Group (IRSG) D9602 (subset 2 patients together with patients with stage 1, group IIB/C or stage 2 group II ERMS) was 85% when treated with VAC (total cumulative cyclophosphamide dose 28.6 g/m²) and a reduced radiotherapy (RT) dose (36 Gy) compared with IRS-IV (41.4 Gy) for patients with group IIA tumors (microscopic positive margins without regional lymph node involvement) (3, 5). Although IRS-IV and D9602 therapies were effective for subset 2 patients, their potential long-term toxicities, which can be attributed largely to the total cumulative cyclophosphamide and RT exposures, are concerning for low-risk patients (6-8).

The primary goal for patients in subset 2 on COG ARST0331 was to estimate the FFS rate using reduced total cyclophosphamide (4.8 g/m²) in combination with VA. A secondary goal was to estimate the local control rate for subset 2 patients with group IIA tumors who received a radiation dose of 36 Gy in the setting of reduced cyclophosphamide.

Patients and Methods

Eligibility criteria for subset 1 of ARST0331 and required staging studies on ARST0331 have been previously published (4). Eligibility for subset 2 differed from subset 1 only by specific stage and group requirements. We defined stage and group according to IRSG guidelines (3). Patients eligible for subset 2 had stage 1, group III non-orbit or stage 3, group I/II ERMS. Signed written informed consent by the patient or guardian following institutional guidelines with IRB approval according to the declaration of Helsinki was required.

The treatment plan for subset 2 of ARST0331 is shown in Table 1. Delayed resection was encouraged if a group III tumor could be completely resected at week 13. RT dose was reduced based on the extent of resection (Table 2) (9). For children ≤ 2 years of age the study permitted RT protocol deviations at the discretion of the treating clinicians based on concerns of long-term morbidity of RT.

Table 1.

ARST0331 Regimen for Patients in Subset 2

Week
1	2	3	4	5	6	7	8	9	10	11	12	E
V	V	V	V	V	V	V	V	V				V
A	-	-	A	-	-	A	-	-	A	-	-	A
C	-	-	C	-	-	C	-	-	C	-	-	L
Week
13	14	15	16	17	18	19	20	21	22	23	24	E
V	V	V	V	V	V	V	V	V	-	-	-	V
A	-	-	A*	-	-	A	-	-	A	-	-	A
RT (if Group II or III)**												L
Week
25	26	27	28	29	30	31	32	33	34	35	36	E
V	V	V	V	V	V	V	V	V	-	-	-	V
A	-	-	A	-	-	A	-	-	A	-	-	A
												L
Week
37	38	39	40	41	42	43	44	45	46	47	48	E
V	V	V	V	V	V	V	V	V	-	-	-	V
A	-	-	A	-	-	A	-	-	A	-	-	A
												L

Abbreviations: V = Vincristine: age < 1 year: 0.025 mg/kg, age 1 - < 3 years: 0.05 mg/kg, age ≥ 3 years: 1.5 mg/m² (maximum dose 2 mg); A = Dactinomycin: age < 1 year: 0.025 mg/kg, age ≥ 1 year: 0.045 mg/kg (maximum dose 2.5 mg); C = Cyclophosphamide: age < 3 years: 40 mg/kg, age ≥ 3 years: 1.2 g/m², given with MESNA; EVAL = Evaluate extent of disease; RT = Radiotherapy (group IIA, 36 Gy; group IIB, 41.4 Gy; group III, 50.4 Gy);

^*= Omit dactinomycin at week 16 in patients receiving RT;

^**= For patients with group III vaginal tumors who enrolled before September 7, 2009, local therapy was delayed until week 24 if the tumor was shrinking or undetectable at week 12. If RT was initiated at week 24, dactinomycin was omitted during RT.

Table 2.

RT Dose Reductions for Patients with Group III Tumors who Underwent Delayed Resection

Extent of Resection	RT dose (Gy)*
Complete resection with negative regional nodes	36
Microscopic positive margin with negative regional nodes	36
Positive regional nodes +/- microscopic positive margins	41.4
Gross residual or no resection	50.4

^*There were two exceptions to this treatment plan: RT was not administered to patients with group III vaginal tumors who enrolled before September 7, 2009 and met the criteria described in the text or to patients with group III uterine or cervical tumors who underwent delayed complete resection with hysterectomy at week 13.

Before September 7, 2009, patients with group III node-negative vaginal tumors were an exception to the RT plan used for patients with tumors at other sites on ARST0331. For these patients, RT was delayed if imaging or vaginoscopy at week 12 of therapy showed that the tumor was shrinking or undetectable. RT was not used if resection did not show tumor or if any residual tumor was completely removed at week 24. If residual tumor remained in the vagina at week 24, RT dose was based on the extent of resection as shown in Table 2. Following September 7, 2009, the study was amended so that girls with group III vaginal RMS followed the same RT guidelines as other patients because increased local failures had been observed (10).

Statistical Analysis

ARST0331 was a non-randomized, non-inferiority clinical trial. The statistical design has been previously published for subset 1 of ARST0331 (4). The design for subset 2 differed by comparing the primary endpoint of FFS with the fixed expected outcome for patients in subgroup B of D9602, excluding those with stage 1, group IIB/C or stage 2, group II tumors, whose long-term FFS was expected to be 83% at the time that ARST0331 was written, using the formula S(t) = 0.83 + 0.17*exp(-1.00*t) (5). A cure model was used to develop the required sample size. One hundred subset 2 patients would give 80% power (testing at the 15% significance level, one-sided) to detect a decrease in FFS from 83% to 75% associated with the reduction in therapy. We compared FFS and overall survival (OS) curves for subset 2 using the log-rank test (11). Confidence intervals (CI) were determined by Greenwood’s formula (12). Results were based on data available as of June 30, 2014.

Results

The study accrued 85 subset 2 patients from September 20, 2004-September 23, 2011. Accrual to subset 2 was temporarily suspended February 14, 2008 due to a high rate of local recurrences among girls with group III vaginal tumors who did not receive RT (10). The protocol was therefore amended to require RT for patients with group III vaginal tumors and re-opened to accrual September 7, 2009. Interim monitoring subsequently showed suboptimal FFS for all patients in subset 2, and further enrollment onto subset 2 was permanently suspended on September 23, 2011.

There were 19 ineligible patients: 7 with ineligible pathology, 4 with no tissue for pathology review, and 8 with ineligible stage/group. The median follow-up for the 66 eligible subset 2 patients at the time of the data freeze was 3.5 years (range 1.5 to 8.2 years).

Patients

Patient characteristics are shown in Table 3. Most were < 5 years of age (n=46, 70%), female (n=44, 67%) and had stage 1 (n=45, 68%), group III (n=45, 68%) tumors that were noninvasive (n=44, 67%), > 5 cm (n=39, 60%) and known to be without nodal involvement (n=53, 80%). The most common primary sites were female genital tract (n=21 [18 vagina, 2 uterus, 1 cervix], 32%), non-parameningeal head/neck (n=12, 18%) and hepatobiliary (n=9, 14%). Fifty-nine patients (89%) had group II or III tumors, of whom 24 (41%) did not receive RT; 14 of these did not receive RT as allowed by protocol guidelines (12 with tumors of the female genital tract and 2 ≤ 2 years of age), 3 did not receive RT against protocol guidelines, and 7 were taken off protocol therapy before scheduled RT. Five patients with group III vaginal tumors enrolled after September 7, 2009; four received RT and one was taken off protocol therapy before scheduled RT. Among the 13 patients with group IIA RMS, 6 did not receive RT; 2 were ≤ 2 years of age and did not receive RT as allowed by protocol guidelines, 3 did not receive RT against protocol guidelines, and 1 was taken off protocol therapy before scheduled RT. There were 2 minor RT protocol deviations and 1 major RT protocol deviation in 2 patients with group IIA RMS who were treated with RT.

Table 3.

Patient Characteristics for the Eligible Patients Enrolled onto Subset 2 of ARST0331

Characteristic	Subset 2 (n=66)
Age (years)
< 5	46 (70%)
5-9	8 (12%)
10-14	5 (8%)
15-21	6 (9%)
> 21	1 (1%)
Sex
Female	44 (67%)
Male	22 (33%)
Histology *
Botryoid	21 (32%)
Embryonal	38 (58%)
Spindle Cell	7 (11%)
Stage*
1	45 (68%)
2	0 (0%)
3	21 (32%)
Group *
I	7 (11%)
IIA	13 (20%)
IIB	1 (1%)
IIC	0 (0%)
III	45 (68%)
Primary Site*
Head and Neck	12 (18%)
PM/PM Ext	1 (1%)
GU, non B/P
Female GU	21 (32%)
Paratestis	3 (5%)
Bladder	5 (8%)
Extremity	2 (3%)
Hepatobiliary	9 (14%)
Peritoneum	2 (3%)
Diaphragm	1 (1%)
Retroperitoneum/Pelvis	4 (6%)
Trunk	5 (8%)
Other	1 (1%)
Tumor Size*
≤ 5 cm	27 (41%)
> 5 cm	39 (59%)
Tumor Invasion*
T1	44 (67%)
T2	22 (33%)
Nodal Status *
N0	53 (80%)
N1	8 (12%)
Nx	5 (8%)

Abbreviations: GU – genitourinary, non B/P = non-bladder/prostate, PM – parameningeal, PM Ext – parameningeal extension, T1 = non-invasive, T2 = invasive, N0 = regional lymph nodes not clinically involved, N1 = regional lymph nodes clinically involved by neoplasm, Nx = regional lymph node status unknown

^*The composite variables are the value from central review, if available; otherwise they are values reported by the institution.

Twenty-two of 45 (49%) patients with group III tumors underwent delayed surgical procedures, including 17 girls with genital tract RMS (15 vagina, 1 uterus, 1 cervix). There were no deviations from the protocol-directed surgical approach to delayed resection in patients with group III tumors identified. Sixteen of the 22 patients (73%) were tumor free or with microscopic tumor only after the delayed surgical procedure and received a reduced RT dose (36 Gy) or no RT (Table 2); margins were unknown for 3. Only 6 of the 21 girls with group III genital tract RMS were treated with RT.

Outcome

With a median follow-up of 3.5 years, there were 20 failures compared with 10.53 expected failures under the null hypothesis if the FFS rate for patients in subset 2 of ARST03331 was the same as that for similar patients treated on D9602. The estimated 3-year FFS and OS rates for subset 2 patients on ARST0331 were 70% (95% CI: 57%, 80%) and 92% (95% CI: 83%, 97%), respectively (Figure 1). Fifty-five of the 78 subgroup B patients on D9602 met subset 2 stage and group requirements. The corresponding 3-year FFS and OS rates for these similar patients on D9602 were 83% (95% CI: 70%, 91%) and 90% (95% CI: 79%, 96%), respectively. The estimated 3-year FFS rate for girls with subset 2 genital tract tumors (n=21) on ARST0331 was 57% (95% CI: 33%, 75%) compared with 77% (95% CI: 61%, 87%) for all other subset 2 patients (n=45) (p=0.02) (Figure 2a). The estimated 3-year OS rate for girls with subset 2 genital tract tumors (n=21) was 95% (95% CI: 69%, 99%) compared with 91% (95% CI: 78%, 97%) for all other subset 2 patients (p=0.21) (Figure 2b). The estimated 5-year OS rate for subset 2 patients on ARST0331 was 87% (95% CI: 77%, 97%), and the 5-year OS rate for subgroup B patients on D9602 who met subset 2 stage and group requirements was 88% (95% CI: 83%, 97%).

Figure 1.

FFS and OS rates for patients in subset 2 on ARST0331

Figure 2.

FFS and OS rates for patients in subset 2 on ARST0331 by primary site

Recurrences

There have been 20 recurrences (10 in patients with primary tumors of the female genital tract and 10 in patients with primary tumors arising in other primary sites) and 8 disease-related deaths (1 in a patient with a primary tumor of the female genital tract and 7 in patient with primary tumors arising in other primary sites). There were 14 local recurrences, 2 locoregional recurrences, 1 local + distant recurrence, 1 regional recurrence and 2 distant recurrences. Local failure occurred in 9 of 21 (43%) girls with genital tract tumors and in 8 of 45 (18%) of all other subset 2 patients. There were 8 recurrences among 15 girls with genital tract tumors who did not receive RT and 1 recurrence among 6 girls with genital tract tumors who received RT. There were 4 recurrences (2 local, 1 locoregional and 1 local + distant) among 13 patients with stage 3, group IIA tumors. Two of the 4 that recurred did not receive RT, one of whom was ≤ 2 years of age.

Toxicity

There was 1 patient with hepatobiliary RMS who developed sinusoidal obstruction syndrome after RT, which was classified as moderately severe using criteria that we have previously published (4). We did not observe any unexpected grade 4 toxicities or toxic deaths.

Discussion

To limit the toxicity for subset 2 low-risk RMS patients, we reduced the total cyclophosphamide dose on ARST0331 to 4.8 g/m², a dose that would be expected to preserve fertility in most patients (6, 7). Using this approach, the 3-year FFS survival rate was 70% (95% CI; 57%, 80%) compared with a 3-year FFS for similar patients of 83% (95% CI: 70%, 91%) when treated with a more intensive VAC regimen (total cumulative cyclophosphamide 28.6 g/m²) on D9602. We consider the overall FFS for all subset 2 low-risk RMS patients on ARST0331 to be suboptimal, especially for girls with genital tract RMS.

The 3-year FFS was significantly worse for girls with genital tract tumors than for patients with tumors in other primary sites, suggesting that the local control guidelines that eliminated RT for group III vaginal RMS in combination with lower total cyclophosphamide contributed to the suboptimal outcome. The 2-year cumulative incidence of local recurrence was 43% among girls with group III vaginal RMS following protocol guidelines before September 7, 2009 as reported previously (10). We subsequently amended ARST0331 to require RT for patients with group III vaginal RMS, following the same paradigm that was used for other sites. Only 5 patients with group III vaginal RMS enrolled onto the study after the amendment. Therefore, our experience is too limited to reach a conclusion regarding the efficacy of RT with or without delayed resection in the setting of a reduced total cyclophosphamide dose of 4.8 g/m² for vaginal RMS.

Although FFS was suboptimal for patients with subset 2 low-risk RMS on ARST0331, 3-year OS was comparable between subset 2 patients on ARST0331 (92%; [95% CI: 83%, 97%]) and D9602 (90% [95% CI: 79%, 96%]). The 5-year OS was also comparable between subset 2 patients on ARST0331 (87% [95% CI: 78%, 97%]) and D9602 (88% [95% CI: 83%, 97%]). In addition, the 3-year OS was similar on ARST0331 between girls with genital tract RMS (95% [95% CI: 69%, 99%]) and patients with tumors at other primary sites (91% [95% CI: 78%, 97%]) (p=0.21). Debate continues among cooperative groups in Europe and North America concerning the optimal endpoint for low-risk RMS studies (13). The COG Soft Tissue Sarcoma Committee has used FFS as the primary endpoint for therapy-reduction low-risk RMS studies with the objective to prevent recurrence and the need for salvage therapy, which is usually more toxic and may not be successful as often as initial therapy (14). We did not collect data regarding salvage therapy that was used following recurrence on ARST0331. In contrast, a goal of the International Society of Pediatric Oncology Malignant Mesenchymal Tumor 89 (MMT 89) study was to reduce local therapy and its toxicity for patients with localized RMS and used OS as the primary endpoint (15). Their design accepted the possibility of poorer FFS and the need for salvage therapy for those who relapse. For example, patients with orbit RMS on MMT 84 and MMT 89, which eliminated RT if complete response was achieved with chemotherapy, had a 10-year event-free survival rate of 58% and a 10-year OS rate of 85%, whereas patients with orbit RMS on IRS-II, -III, and –IV, which used primary RT for group II/III tumors, had a 10-year event-free survival rate of 86% and a 10-year OS rate of 88% (16). Therefore, a subset of patients with orbit RMS was cured without local therapy and its toxicity on the MMT studies. However, using this approach another subset of patients underwent additional treatment for relapse. The impact of these philosophical differences in study design on quality of survival for low-risk patients will only be addressed through studies directed at these populations after years of follow-up. We do not believe the suboptimal FFS and burden of subsequent salvage therapy justifies ARST0331 therapy for subset 2 patients.

A local failure rate of 17 of 66 (26%) for subset 2 patients on ARST0331 is unacceptably high, including 43% for girls with genital tract tumors and 18% for all other subset 2 patients. Given the high local failure rates for subset 2 patients on ARST0331, our results support a role for cyclophosphamide in local control for these patients. Indeed, our experience with group IIA or III N0 vaginal RMS patients on D9602 and ARST0331 who did not receive RT also shows a higher local recurrence rate when less cyclophosphamide is given (10).

Both D9602 and ARST0331 tested a reduced RT dose (36 Gy) compared with IRS-IV (41.4 Gy) for patients with group IIA tumors. There were only 13 subset 2 patients with group IIA tumors on ARST0331. We cannot assess the efficacy of 36 Gy for subset 2 patients with group IIA tumors because of the following factors: the small number of stage 3, group IIA patients; the high proportion of stage 3, group IIA patients, who did not receive RT; and the suboptimal total cyclophosphamide dose.

We believe that quality of outcome could be assessed for low-risk RMS patients by taking FFS, OS and toxicity into account. We achieved a favorable balance for subset 1 patients (stage1/2, group I/II or stage 1, group III orbit ERMS) using ARST0331 therapy, including 22 weeks of VAC (total cumulative cyclophosphamide dose 4.8 g/m²) followed by VA and RT for patients with group II/III tumors. We do not believe we have achieved a favorable balance for subset 2 patients given their suboptimal FFS on ARST0331, using a reduced total cyclophosphamide dose of 4.8 g/m². Elimination of RT in girls with group III vaginal tumors following a complete response with chemotherapy, with or without delayed resection, in combination with reduced cyclophosphamide, likely contributed to the suboptimal outcome. IRS-IV or D9602 VAC (total cumulative cyclophosphamide 26.4-28.6 g/m²) achieved a FFS ≥ 83% for subset 2 patients, but potential toxicity is concerning. Future COG trials will include subset 2 low-risk RMS patients in the intermediate-risk group, which on the most recent trial received 8.4 or 16.8 g/m² of cyclophosphamide (17). For girls with group III genital tract tumors, further investigation would be necessary to establish whether a total cyclophosphamide dose between 4.8 g/m² and 26.4 g/m² optimizes FFS without RT. In addition, further investigation would be necessary to determine whether adjustments to the RT regimen could improve FFS using the ARST0331 chemotherapy backbone for patients with subset 2 low-risk RMS.