Pediatric rhabdomyosarcoma with bone marrow metastasis

Abstract

Background:

Rhabdomyosarcoma (RMS) is the most common soft tissue sarcoma of adolescence and childhood. Although most patients with localized RMS are cured, outcome of those with metastatic disease remains unsatisfactory. RMS with bone marrow (BM) metastasis accounts for approximately 6% of all cases with RMS and has a 3-year event-free survival of 14%. Our study aims to describe our institution’s experience of patients with metastatic RMS with BM involvement.

Methods:

This was a single-institution retrospective study from Memorial Sloan Kettering Kids, a tertiary pediatric oncology center. Patients with RMS who were diagnosed with BM metastasis between 1998 and 2018 were identified from pathology reports.

Results:

For patients with RMS and BM positivity at diagnosis (N = 27), the median survival was 1.5 years. The 1-, 2-, and 3-year overall survival (OS) were 81%, 32%, and 20%, respectively. There is one long-term (defined as >4 year) survivor who is still alive 14.9 years after diagnosis despite two metastatic recurrences. An Oberlin status of 4 that included BM metastasis portended a 3-year OS of 0%.

Conclusions:

Although most patients will respond to initial therapy, BM metastasis at the time of diagnosis lends a near-fatal diagnosis in pediatric patients with RMS. Novel therapies are desperately needed to consolidate their initial remission.

1 |. INTRODUCTION

Rhabdomyosarcoma (RMS) is the most common soft tissue sarcoma of adolescence and childhood.¹ The two most common histological subtypes of RMS are alveolar rhabdomyosarcoma (ARMS) and embryonal rhabdomyosarcoma (ERMS). ARMS exhibits typical cytogenetic characteristics, including t(2;13)(q35;q14) or t(1;13)(p36;q14), which result in PAX3-FOXO1(FKHR) or PAX7-FOXO1(FKHR) fusion genes, respectively. PAX3-FOXO1 or PAX7-FOXO1 has been detected in 55% and 22% of ARMS cases, respectively. Cases of histologically identified ARMS lacking a fusion appear to be genetically and clinically indistinguishable from cases of ERMS.^2,3

The typical sites of presentation of RMS include the structures of the head and neck, followed by the genitourinary (GU) tract and the extremities.^1,4 In nonmetastatic RMS, the estimated 5-year overall survival (OS) rate is around 65% for all subtypes; more specifically, there is reported 65–90% 5-year OS in the embryonal type and 50–65% in the alveolar type.^4,5 Metastasis of RMS portends the worst prognosis with a 3-year OS of 34%.⁶ RMS with bone marrow (BM) metastasis accounts for approximately 6% of all cases with RMS,⁷ and cases with BM involvement have a 14% 3-year event-free survival (EFS) compared to 34% without BM metastasis (P < .0001).⁸

As a consequence of the dearth of published cases of RMS with BM metastasis, we accumulated a single-institution case series in an attempt to further stratify this cohort of patients into different risk and/or prognostication categories. Our study aims to perform a comprehensive review of the literature and present our institution’s cases with the hopes of further describing the subset of patients with RMS with BM involvement.

Our primary objective was to review the charts of RMS patients with BM metastasis treated under the Pediatric Oncology Service at Memorial Sloan Kettering (MSK) Kids from 1998 to 2018 and collect data on their treatment regimen, EFS and OS. Secondary objectives of this study included comparing the outcomes of our patients with historical case reports and determining a means of classifying patients according to histologic type or presentation.

2. METHODS

This was a single-institution retrospective study from MSK Kids, a tertiary pediatric oncology center. Patients with RMS who had BM metastasis as identified on BM pathology reports at diagnosis between January 1, 1998 and July 1, 2018 were included. Patients were excluded if they were not treated at least in part on the MSK Kids Pediatric Service. The medical records of the children in the case series were reviewed retrospectively. Demographics, clinical presentation, imaging, pathology, management, and outcome were recorded. The study was approved by the Institutional Review Board (IRB) at MSK Cancer Center.

2.1 |. Statistics

Data were collected from the electronic medical records at MSK Kids. Patient demographics were tabulated using mean, median, and mode for continuous variables and percentages for categorical variables. The statistical methods used to evaluate the study objectives included Kaplan-Meier to estimate survival and log-rank to compare OS and EFS. OS was calculated as the date of diagnosis to the date of death or last follow-up. EFS was calculated from the date of the start of treatment to the date of the first event, defined as tumor progression (PR), relapse, or death as a result of any cause.

3 |. RESULTS

3.1 |. Demographics of patients with bone marrow disease at diagnosis (N= 27)

A total of 27 patients were identified and included in the analyses of patients with RMS with BM disease at diagnosis. The median age at the time of diagnosis was 15.8 years (range 3–33); average was 16.5 years (Table 1). Most patients had a diagnosis of ARMS (93%) with a majority harboring the PAX3-FOXO1 fusion. The most common primary site of disease was extremity (52%) and a majority of patients had ≥3 sites of metastasis (including BM) at diagnosis (85%). There were five patients with only one other site of metastasis other than BM. In three patients with ARMS with regional and distant lymph node (LN) involvement, BM metastasis was seen in the absence of concurrent osseous metastases; two patients had primary sites of extremity and one was GU (vulvar). Additionally, two patients with extremity ARMS had bone and BM as sole sites of metastasis. The Oberlin score was a point system with a maximal score of 4⁶ and a majority of patients (52%) had a score of 4. Twenty-three of 27 patients had noninvasive F-18-fluorodeoxyglucose positron-emission tomography (FDG-PET) as part of their diagnostic workup. Initial treatment regimens at outside hospitals for four patients included AEWS1031, MAID protocol, COP (for presumed Burkitt lymphoma), and AAML1031 [for presumed acute myeloid leukemia (AML)]. All patients were subsequently treated according to the existing Children’s Oncology Group (COG) trials (generally ARST0531⁹) or MSK single-institutional pilot studies (IRB 03–099¹⁰) at the time of transfer to MSK Kids.

TABLE 1.

Demographics of patients with rhabdomyosarcoma and bone marrow metastasis at diagnosis (N = 27)

	N	Percentage
Gender
Female	14	52
Male	13	48
Age
Age <10 years	6	22
Age ≥10 years	21	78
Subtype
Alveolar	25	93
PAX3-FOXO1 fusion	22
PAX7-FOXO1 fusion	1
Negative for FOXO1 fusion	1
Unknown fusion status	1
Embryonal	2	7
Primary site of disease
Extremity/gluteal	14	52
Trunk/mediastinum/paravertebral	4	15
Abdomen/retroperitoneum	2	7
Genitourinary (vulvar/paratesticular)	2	7
Parameningeal	2	7
Perineal/pelvis	2	7
Prostate	1	4
Metastatic sites
Bone marrow + bone	2	7
Bone marrow + lymph node	2	7
≥3 sites including bone marrow	23	85
Oberlin score
2	2	7
3	11	41
4	14	52

Note: Alveolar fusion status indicates fusion presence in bone marrow and/or tumor at diagnosis. The Oberlin score was a point system with a maximal score of 4.⁶

3.2 |. Clinical course of patients with bone marrow disease at diagnosis (N= 27)

For patients with BM positivity at diagnosis (N = 27), the 1-, 2-, and 3-year OS were 81%, 32%, and 20%, respectively (Figure 1), with a median survival of 1.5 years. There is one long-term (defined as > 4 year) survivor who is still alive 14.9 years after diagnosis despite two metastatic recurrences; he is the only patient with PAX7-FOXO1 fusion in our study. One patient died 6.2 years after diagnosis due to recurrent RMS and liver failure. Two other patients are currently alive with disease (1 year 6 months and 3 years 7 months after diagnosis), while all other patients (N = 23) died <4 years from the time of initial diagnosis. There was one treatment-related death (TRD) from secondary AML developing following a prior central nervous system metastatic recurrence; all other deaths were attributed to progressive RMS.

FIGURE 1.

Overall survival of patients with rhabdomyosarcoma and bone marrow metastasis at diagnosis (N = 27)

All but one patient had clearance of detectable disease from the BM within 3 months of starting treatment; this one patient with a positive BM at the cycle 4 (week 12) time point (of the MSK 03–099 protocol) did not have a BM performed after cycle 2 (week 6). Seventy-four percent (14/19) of patients had a complete remission (CR) of their BM at approximately week 6 (status post two cycles of chemotherapy); the other eight patients did not have data available for BM biopsy at that time point. There was no statistically significant difference in OS between patients who had positive BM at either the cycle 2 or cycle 4 time point (6/19) compared to patients who had negative BM (N = 20) at those time points (Figure 2). Three-year OS for negative BM (at both cycles 2 and 4) was 16% and for positive BM (at either cycles 2 and/or 4) was 33% (P = .91).

FIGURE 2.

Kaplan-Meier of patients with rhabdomyosarcoma and bone marrow metastasis at diagnosis who had a bone marrow that was positive at either the cycle 2 (week 6) or cycle 4 (week 12) time point (N = 6), or negative at both the aforementioned time points (N = 20) (P = .91). One of the N = 27 patients did not have bone marrow data available for either time point

The 1- and 2-year OS for patients who initially went into remission (per radiology and BM pathology), then had a later relapse, were 92% and 33%, respectively. All of the patients had a relapse or progression, and average time to the “first event” from diagnosis was 14 months (range 1–39 months). The 1- and 3-year EFS were 57% and 8%, respectively. For patients who went into remission, median time to relapse from diagnosis was 1.4 years (range 8–46 months). The 1- and 2-year OS for patients who had stable/progressive disease were 73% and 30%, respectively. There was no statistically significant difference (P = 0.11) in OS for patients who relapsed (median survival 1.8 years, N = 12) versus patients who had stable/progressive disease (median survival 1.5 years, N = 15) (Figure 3).

FIGURE 3.

Kaplan-Meier of patients with rhabdomyosarcoma and bone marrow metastasis who had radiologic and/or bone marrow pathologic stable disease or progression (N = 15) compared to those who had a complete remission (based on radiology and bone marrow pathology) (N = 12) (P = .11)

Abbreviations: CR, complete remission; SD, stable disease; PR, progression

To determine whether BM as a site of progression or relapse conferred a worse prognosis, we stratified patients into three groups. The first group was patients who achieved CR, followed by a relapse that did not include the BM (N = 9). The second group consisted of patients who had progressive (or stable) disease that did not include BM (N = 13) (ie, patients who had cleared their BM disease based on BM pathology reports). The third group (N = 5) included patients who achieved a CR followed by a relapse that included the BM (N = 3), in addition to patients who had progressive (or stable) disease that included persistent BM positivity (N = 2) (ie, patients who never cleared their BM disease). There was no statistically significant difference in the OS between these three groups (P = .57). For the N = 3 patients whose site of relapse included the BM, the relapse was in the setting of other metastatic sites of disease.

In stratifying by age at diagnosis, there was no significant difference in OS between those who were age <10 (N = 6) and ≥10 (N = 21) at diagnosis (P = .0508). There were no patients <1 year old.

In stratifying by Oberlin status,⁶ there was a significant difference between Oberlin score of 2 or 3 compared to 4 (P = .0089), as well as a score of 3 compared to 4 (P = .0041) (Figure 4). The 3-year OS for the Oberlin scores of 2 or 3 together (N = 13), 3 alone (N = 11), and 4 alone (N = 14) were 43%, 52%, and 0%, respectively.

FIGURE 4.

Kaplan-Meier of patients with rhabdomyosarcoma and bone marrow metastasis who had an Oberlin score of 3 (N = 11) versus 4 (N = 14) (P = .0041)

3.3 |. Clinical course of patients with bone marrow disease at diagnosis

We examined the BM biopsy, aspirate, or cytogenetics to look for concordance between the results of BM evaluation. There were a total of 119 BM individual time point samples procured throughout all stages of treatment. In some instances, patients were treated at outside hospital for a portion of their treatment, thus it was not clear from medical records if “bone marrow positive” indicated only a positive biopsy, as the aspirate and/or cytogenetics did not always accompany the positive BM biopsy pathology report. Some BM samples had insufficient sample to be sent for molecular analysis, thus only a BM biopsy and/or aspirate were analyzed.

Of the 115 BM time points for patients with ARMS, there were 85 times when biopsy, aspirate, and cytogenetics were available, 26 times when only two of these studies were available, and 4 times when only one of these studies was available. One patient developed treatment-related AML, however, BM studies that were positive for AML were negative for simultaneous RMS.

Eighty percent (20/25) of patients with ARMS harbored a fusion gene in the BM at diagnosis. For the five patients with unknown BM cytogenetics at diagnosis, three had positive fusions in their tumor at diagnosis and the other two patients had unknown tumor fusion status. One of these two unknown patients with tumor/BM at diagnosis had a later repeat BM cytogenetics that were negative for FOXO1 fusion. Within these 115 BM samples from patients with ARMS, there were 15 times (13%) when there was discordance between the results, such as negative biopsy/aspirate with positive cytogenetics.

4 |. DISCUSSION

In this study, we describe 27 patients with newly diagnosed RMS with BM involvement. Patients with RMS and BM metastasis at diagnosis were typically ≥10 years old with ARMS harboring a PAX3-FOXO1 mutation. The primary tumors were in unfavorable locations, most commonly arising in the extremity and generally presented in the setting of ≥3 sites of metastasis. Additionally, the tumors were typically small in size and occasionally did not have overt clinical symptoms of their primary sites. The poor prognosis of RMS with BM metastasis reflects tumors that aggressively metastasize early in their course.

Almost all (97%) of patients cleared their BM within 3 months of starting treatment. All but two (93%) of the N = 27 patients died of disease within 4 years of initial diagnosis; only one patient is a long-term (defined as >4 year) survivor despite two subsequent metastatic recurrences. BM involvement was not usually the sole site of progression or recurrence, but typically occurred in the context of widespread, multifocal metastatic disease. Within our study, an Oberlin status of 4 that included BM metastasis portended a 3-year OS of 0%. Given their grim outcome despite near-uniform initial complete responses, these patients are in desperate need of novel therapies focusing on consolidating remission.

For our two patients who had OS >4 years, they achieved CR by the end of their initial treatment (MSK 03–099), then had first relapse 2.1 and 3.3 years after diagnosis. For the first aforementioned patient (who died of disease), the treatment course included (in chronologic order) vinorelbine/oral cyclophosphamide/bevacizumab, vincristine/dactomycin, then topotecan/cyclophosphamide. Our second patient (who is our sole long-term survivor) sequentially received vinorelbine/oral cyclophosphamide/bevacizumab, vincristine/topotecan/doxorubicin, IL7-supported autologous tumor vaccine, and pazopanib. Both patients received adjunctive tumor resections and/or radiation therapy when appropriate.

Our sole long-term survivor presented at the age of 3 with a gluteal mass and BM, bone, and LN metastasis (Oberlin 3). It is very curious that he is the only patient who harbored a PAX7-FOXO1 mutation of ARMS. A prior COG analysis has shown that PAX7 metastatic disease has an improved OS compared to PAX3, though none of the PAX7 patients had BM as the site of metastasis.² There are only three other cases in the literature of patients who have PAX7-FOXO1 mutation with BM metastasis.^11–13 It would be interesting to study whether there is a difference in survival or biological factors of PAX7-positive compared to PAX3-positive patients with RMS and BM metastasis.

For our MSK pilot study IRB 03–099, restaging marrows were required after two (week 6) and four (week 12) cycles and then every 3 months for patients with initial BM involvement. This was, in part, to permit a determination of the response rate to two cycles of the novel two-drug combination of irinotecan and carboplatin; however, in the absence of positive-predictive value of early time to BM negativity, frequent and ongoing BM sampling would seem difficult to justify. We recommend bilateral BM biopsies, aspirates, and cytogenetic (fusion status) studies to be sent on all patients with RMS at the time of BM assessment, as the presence of BM metastasis greatly alters prognosis, and our study found a few instances where the three studies had discordant results, likely due to sampling.

The sensitivity for BM biopsy for RMS is unknown. It is not clear whether increasing the number of BM biopsy/aspirate sampled sites from 2 to 4, as is done with neuroblastoma, would increase the sensitivity. Since all of our patients had BM metastasis (whether at diagnosis or relapse) in the setting of at least one other site, we do not think that additional sampling is necessary. Additionally, this study initially searched pathology reports for all patients with BM involvement at any point in diagnosis. There was only one additional patient (not included in the study) who had BM involvement at a later time point (that was not present with her metastatic disease at initial diagnosis). It would be impossible to extrapolate whether BM involvement at diagnosis was “missed” in that patient and what the true sensitivity of the BM test is.

FDG-PET scans have a high ability to detect LN, bone, and BM involvement in patients with RMS.^14–16 Notably, in earlier studies of RMS, bone scan was the standard for finding osseous and other metastases. In a review of 88 patients, FDG-PET performed better than bilateral BM biopsy/aspirate for detection of bone disease at initial diagnosis of RMS.¹⁷ Therefore, currently FDG-PET is used at diagnosis, instead of bone scan, due to its greater sensitivity. Within our series, initial staging in 23 of 27 patients was based on evaluation of disease that included FDG-PET. Overall, the role of FDG-PET in identifying BM metastasis in solid tumors, including RMS, remains unclear.¹⁸

The published estimated 3-year OS for patients with RMS with metastasis is 34%.⁶ Within our study, for patients with BM positivity at diagnosis (N = 27), 3-year OS was 20%, with a median survival of 1.5 years. It is unclear if these two numbers are statistically significantly different. Several studies have investigated factors that alter the prognostic significance of metastatic disease. The site of primary tumor for metastatic disease is important, as nonbladder nonprostate GU tumors had more favorable outcome¹⁹; however, it is difficult to extrapolate from our N = 2. Within our small N = 27, we did not note any statistically significant findings in regards to age <10 or ≥10, so a larger cohort may be necessary to determine alterations in that prognostic factor. Increased number of metastatic sites is a well-established individual poor prognostic indicator in patients with RMS.^3,20,21 However, the metastatic involvement of solely the BM (in N = 12 patients) was reported to have no effect on survival in the IRS-IV study.²¹ This finding was never repeated and it may reflect an era when bone scans, instead of FDG-PET, were utilized to detect metastatic disease. There may be very rare patients with RMS who present with only BM metastasis, as we detected zero within our population over a 21-year period.

In the Oberlin paper, the published 3-year EFS of 14% was identical for both patients with three or more metastatic sites (N = 145), as well as patients with BM metastasis (N = 292).⁶ Based on these numbers, not every patient with BM metastasis also had three or more metastatic sites. Additionally, the OS for the BM metastasis and non-BM metastasis groups was not published. Since we noted a trend toward poorer OS in our BM metastasis population, which contradicted the aforementioned paper²¹ and since OS of this subgroup was not reported in the Oberlin paper,⁶ we felt that the OS and patient characteristics of RMS with BM metastasis warranted reporting. Overall, a higher Oberlin score has been shown to confer a worse EFS,^6,21–23 and our N = 14 cohort of RMS with BM metastasis and Oberlin score of 4 (compared to a score of 2 or 3, or to a score of 3 alone) had a 3-year OS of 0%, which may be the most significant prognostic indicator within metastatic RMS. This OS of patients with RMS with BM metastasis plus an Oberlin score of 4 represented a steep decline that was not published or widely known.

The study is limited in that it is a retrospective review, in which patient treatment may have been biased by the patient’s disease status and other factors, which confound outcomes. As mentioned above, our review included patients who initially received alternative treatments to what could be considered standard of care for stage IV RMS, as we believe these could similarly represent real-world delays in treatment from delay in diagnosis. None of the patients who received COP, AEWS1031, MAID protocol, or AAML1031 had a TRD, and none of them are long-term survivors.

There are several current treatment options for patients with recurrent RMS. The most recent COG study for the first recurrence or progression (ARST0921) was a randomized phase II study to assess the feasibility and activity of bevacizumab or temsirolimus in combination with vinorelbine/cyclophosphamide, which showed superior 6-month progression-free survival in the temsirolimus arm.²⁴ In a single-institution setting, nine patients received topotecan plus vincristine and doxorubicin; this regimen was able to produce objective responses in six of six patients who were able to be evaluated for response after two cycles.²⁵ A dendritic cell autologous vaccine trial was designed at the NCI/POB/NIH, and of the 30 patients who initiated immunotherapy, 1 of the 6 surviving patients at the time of publication had ARMS with BM metastasis.²⁶ Single-agent therapies under clinical investigation can be offered at the time of progression or relapse; these include pazopanib, exatecan mesylate, trabectedin, lorvotuzumab mertansine R1507, and other small molecule oral kinase inhibitors (sorafenib, regorafenib, and pazopanib).^27–31 Offering relapsed biopsy material to be sent for xenograft development is also important, as several targeted mutations have been discovered from laboratory assessment of tumors, though they have not yet translated into clinical trials.³²

The general course for patients with RMS with BM metastasis is a rapid and dramatic radiographic and BM pathologic remission, followed by 100% risk of recurrence or progression in approximately 14 months. Without chemotherapy, but with aggressive supportive care, patients would likely survive for somewhere between 3 and 9 months after recurrence. With chemotherapy, it is likely that disease would be controlled for at least 12–16 months before it worsens or recurs. For any early, on-therapy multimetastatic recurrence of disease, there is almost certainty that patients will die of their disease, likely within a year and with almost certainty within 4 years. Our review of patients confirms that BM disease confers a near-fatal prognosis, and BM disease with an Oberlin score of 4 may trump other established prognostic factors. Since 100% of patients in our study had a relapse or progression, a prospective trial is needed with novel therapies to consolidate remission prior to the first recurrence. RMS patients with BM metastasis may also be one of the subsets of patients who could benefit from 2 or more years of maintenance therapy.

Abbreviations

AML

acute myeloid leukemia

ARMS

alveolar rhabdomyosarcoma

BM

bone marrow

COG

Children’s Oncology Group

CR

complete remission

EFS

event-free survival

ERMS

embryonal rhabdomyosarcoma

FDG-PET

F-18-fluorodeoxyglucose positron-emission tomography

GU

genitourinary

IRB

Institutional Review Board

LN

lymph node

MSK

Memorial Sloan Kettering

OS

overall survival

PR

progression

RMS

rhabdomyosarcoma

TRD

treatment-related death