Pediatr Blood Cancer 2015;62:1368–1373
Minimal Residual Disease Detection in Autologous Stem Cell Grafts From Patients With High Risk Neuroblastoma Esther M. van Wezel, MD,1,2 Janine Stutterheim, MD, PhD,1,2 Florentien Vree, MD,2 Lily Zappeij-Kannegieter,1 Boris Decarolis, MD,3 Barbara Hero, MD, PhD,3 Frank Berthold, MD, PhD,3 Roswitha Schumacher-Kuckelkorn,3 Thorsten Simon, MD, PhD,3 Marta Fiocco, PhD,4 Carlijn Voermans, PhD,1 Max M. van Noesel, MD, PhD,5 Huib N. Caron, MD, PhD,2 C. Ellen van der Schoot, MD, PhD,1 and Godelieve A. M. Tytgat, MD, PhD,2* † for the GPOH MRD Study Group Background. The clinical significance of minimal residual disease (MRD) detected by real-time quantitative PCR (qPCR) in autologous stem cell grafts in high risk neuroblastoma is still controversial. In this retrospective multicenter study, autologous stem cell grafts of a large cohort were studied using a panel of RNA markers. Procedure. From 104 patients with high risk neuroblastoma, who received autologous stem cell transplantation as first line treatment, 66 peripheral blood stem cells (PBSC) and 38 CD34þ selected grafts were retrospectively collected at 2 Dutch and 12 German centers between 1997 and 2010. To investigate graft contamination qPCR was performed by using 5 neuroblastoma specific markers (PHOX2B, TH, DDC, CHRNA3, and DBH). Results. In PBSC 6/66 (9%) and in CD34þ selected grafts 3/38 (8%) samples were contaminated. Graft
contamination was not associated with an unfavorable outcome (5-years OS, 66% vs. 50.5%; P ¼ 0.6 and 5-years EFS, 22% vs. 35%, P ¼ 0.7). In multivariate Cox analysis BM MRD at time of harvest was significantly associated with survival (P ¼ 0.008 OS and P ¼ 0.002 EFS), but graft contamination was still not associated with an unfavorable outcome (P ¼ 0.9 OS and P ¼ 1 EFS). Conclusions. Graft contamination is very infrequent in this retrospective cohort of patients with no or minimal BM disease prior to stem cell collection and does not influence outcome in univariate and multivariate analysis. The presence of MRD at time of harvest is a strong outcome predictor. However, these results will have to be verified in a large prospective study. Pediatr Blood Cancer 2015;62:1368–1373. # 2015 Wiley Periodicals, Inc.
Key words: autologous stem cell transplantation; high risk; MRD; neuroblastoma; PBSC
INTRODUCTION Neuroblastoma is an extracranial solid tumor of childhood, with a broad spectrum of clinical behavior [1,2]. About 50% of patients present with bone marrow (BM) metastases at diagnosis. Despite intensive multimodal treatment, including high dose chemotherapy and autologous stem cell rescue, patients with high-risk neuroblastoma have a poor prognosis and relapse is a frequent occurrence [3–6]. It remains controversial whether reinfusion of autologous grafts contaminated with tumor cells can cause relapse after autologous stem cell transplantation. Several small studies on detection of neuroblastoma in autologous grafts show conflicting results [7–12]. Handgretinger et al. [7] even found that reinfusion of contaminated grafts, as detected by GD2-immunocytology, was associated with a higher probability of survival, possibly because of a protective antitumor immune response after autologous stem cell transplantation. However, with the use of retroviral marking Rill et al. [13] have shown that autologous grafts from patients with neuroblastoma contain neuroblastoma cells and can contribute to relapse. It has been described that there is a decreased risk for tumor cell contamination in peripheral blood stem cell (PBSC) harvest compared to BM [14,15]. To decrease tumor contamination of the grafts CD34 positive selection can be performed on BM or PBSC collections [8,16]. However, disadvantages are costs of the procedure and a possible reduction in the total number of CD34þ cells [17]. Real-time quantitative PCR (qPCR) is a very sensitive technique to detect small numbers of tumor cells in blood, BM or PBSC harvests [18,19]. Currently, qPCR based minimal residual disease (MRD) detection is based on neuroblastoma specific RNA markers. Several small retrospective studies have investigated the association between tumor cell contamination based on qPCR and outcome [7– 11]. The most commonly used markers in these studies were tyrosine hydroxylase (TH) and GD2 synthase (GD2S). However, C
2015 Wiley Periodicals, Inc. DOI 10.1002/pbc.25507 Published online 4 May 2015 in Wiley Online Library (wileyonlinelibrary.com).
these markers are expressed in normal hematological cells (even in CD34þ selected cells), which might lead to false positive results and subsequently an overestimation of graft contamination [18]. Previously we have identified PHOX2B as a neuroblastoma specific marker [20] and we have developed a panel of markers, that is more sensitive for the detection of MRD than the use of PHOX2B alone, Additional Supporting Information may be found in the online version of this article at the publisher’s web-site. 1
Department of Experimental Immunohematology, Sanquin Research, Amsterdam, The Netherlands and Landsteiner Laboratory of the AMC, University of Amsterdam, Amsterdam, The Netherlands; 2Department of Pediatric Oncology, Emma Children’s Hospital, Academical Medical Center, University of Amsterdam, Amsterdam, The Netherlands; 3Children’s Hospital, University of Cologne, Pediatric Hematology and Oncology, Cologne, Germany; 4Department of Biostatistics, Leiden University Medical Center and Dutch Childhood Oncology Group, The Hague, Netherlands; 5Department of Pediatric Oncology, Sophia Children’s Hospital, Erasmus Medical Center, Rotterdam, The Netherlands Grant sponsor: Koningin Wilhelmina Fonds, The Dutch Cancer Society Esther M. van Wezel and Janine Stutterheim contributed equally to this work. Conflict of interest: Nothing to declare. † GPOH MRD Study Group: Ulrike Koehl, Johannes H Schulte, Felix Niggli, Michael C Fruhwald, Charlotte M Niemeyer, Udo Bode, Freimat H Schilling, Christian Schultz, Norbert Graf, Michaela Nathrath and Irene Schmid.
Correspondence to: G. A. Tytgat, Department of Pediatric Oncology, Academical Medical Center, University of Amsterdam, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. E-mail: [email protected]
Received 20 October 2014; Accepted 18 February 2015
MRD in PBSCs From Neuroblastoma Patients whereas the specificity was not lost because of stringent thresholds for positivity [18]. Kreissman et al. have recently shown in a large prospective study that purging with five monoclonal antibodies targeting neuroblastoma cell surface antigens, does not improve outcome. However, the detection of neuroblastoma mRNA (PHOX2B or TH in 245 patients) in PBSC samples is associated with an unfavorable outcome, possibly because of incomplete purging or the presence of residual tumor in the patient. Multivariate analysis (adjusting for BM status at time of harvest) was not performed [21]. We hypothesized that residual BM disease is correlated with PBSC contamination and that this residual BM disease is indeed associated with outcome. Therefore the aim of our study was to investigate whether infusion of a contaminated graft and/or the presence of residual BM disease influences outcome in a large retrospective cohort in a multivariate setting.
METHODS Patients and Treatment In this retrospective study 104 autologous stem cell grafts (PBSCs or CD34þ selected grafts) were analyzed. Samples were obtained from 104 Dutch and German patients treated with autologous stem cell transplantation as first line treatment. Highrisk patients were defined according to the international neuroblastoma staging system (INSS) as stage 4 over 1 year of age or all stages with MYCN amplification. A total of 26 grafts from Dutch patients and 78 grafts from German patients, treated in two consecutive protocols NB97 (n ¼ 37) [22] and NB04 (n ¼ 41) [23], were collected between 1997 and 2010 (Table I). In NB97 and NB04 induction chemotherapy consisted of alternating N5 and N6 cycles (Supplementary Table SI). The timing for surgery varied during induction chemotherapy. Stem cell collection was recommended as soon as the bone marrow was clear of residual neuroblastoma cells (no tumor cells or MRD on four site cytological and immunological analyses with antiGD2) (qPCR results were not used clinically) and was done after 2–4 cycles of chemotherapy. All patients were treated with melphalan, carboplatin and etoposide as consolidation chemotherapy prior to autologous stem cell rescue. After recovery of hematopoiesis patients received oral retinoic-acid. Patients treated according to the NB97 protocol before 2002 received immunotherapy instead of retinoic acid (n ¼ 13) and immunotherapy was given for 1 year (Supplementary Table SI) [24]. From a second cohort of 21 Dutch patients only BM grafts were available. These BM grafts were also tested and analyzed. All patients were treated according to the VECI protocol (Supplementary Table SII) and samples were collected between 1991 and 2000 [25]. Written informed consent for data collection and for the use of stored remains of autologous stem cell harvests for research purposes was obtained.
RNA Extraction, Reverse Transcription, and Real-Time Quantitative PCR RNA was extracted from 1 to 2 ml of autologous stem cell grafts by using the Trizol method according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA). cDNA was synthesized Pediatr Blood Cancer DOI 10.1002/pbc
1369
TABLE I. Patient Characteristics Patient characteristics Age (months) Median Minimum Maximum
Minimal Residual Disease Detection in Autologous Stem Cell Grafts From Patients With High Risk Neuroblastoma Esther M. van Wezel, MD,1,2 Janine Stutterheim, MD, PhD,1,2 Florentien Vree, MD,2 Lily Zappeij-Kannegieter,1 Boris Decarolis, MD,3 Barbara Hero, MD, PhD,3 Frank Berthold, MD, PhD,3 Roswitha Schumacher-Kuckelkorn,3 Thorsten Simon, MD, PhD,3 Marta Fiocco, PhD,4 Carlijn Voermans, PhD,1 Max M. van Noesel, MD, PhD,5 Huib N. Caron, MD, PhD,2 C. Ellen van der Schoot, MD, PhD,1 and Godelieve A. M. Tytgat, MD, PhD,2* † for the GPOH MRD Study Group Background. The clinical significance of minimal residual disease (MRD) detected by real-time quantitative PCR (qPCR) in autologous stem cell grafts in high risk neuroblastoma is still controversial. In this retrospective multicenter study, autologous stem cell grafts of a large cohort were studied using a panel of RNA markers. Procedure. From 104 patients with high risk neuroblastoma, who received autologous stem cell transplantation as first line treatment, 66 peripheral blood stem cells (PBSC) and 38 CD34þ selected grafts were retrospectively collected at 2 Dutch and 12 German centers between 1997 and 2010. To investigate graft contamination qPCR was performed by using 5 neuroblastoma specific markers (PHOX2B, TH, DDC, CHRNA3, and DBH). Results. In PBSC 6/66 (9%) and in CD34þ selected grafts 3/38 (8%) samples were contaminated. Graft
contamination was not associated with an unfavorable outcome (5-years OS, 66% vs. 50.5%; P ¼ 0.6 and 5-years EFS, 22% vs. 35%, P ¼ 0.7). In multivariate Cox analysis BM MRD at time of harvest was significantly associated with survival (P ¼ 0.008 OS and P ¼ 0.002 EFS), but graft contamination was still not associated with an unfavorable outcome (P ¼ 0.9 OS and P ¼ 1 EFS). Conclusions. Graft contamination is very infrequent in this retrospective cohort of patients with no or minimal BM disease prior to stem cell collection and does not influence outcome in univariate and multivariate analysis. The presence of MRD at time of harvest is a strong outcome predictor. However, these results will have to be verified in a large prospective study. Pediatr Blood Cancer 2015;62:1368–1373. # 2015 Wiley Periodicals, Inc.
Key words: autologous stem cell transplantation; high risk; MRD; neuroblastoma; PBSC
INTRODUCTION Neuroblastoma is an extracranial solid tumor of childhood, with a broad spectrum of clinical behavior [1,2]. About 50% of patients present with bone marrow (BM) metastases at diagnosis. Despite intensive multimodal treatment, including high dose chemotherapy and autologous stem cell rescue, patients with high-risk neuroblastoma have a poor prognosis and relapse is a frequent occurrence [3–6]. It remains controversial whether reinfusion of autologous grafts contaminated with tumor cells can cause relapse after autologous stem cell transplantation. Several small studies on detection of neuroblastoma in autologous grafts show conflicting results [7–12]. Handgretinger et al. [7] even found that reinfusion of contaminated grafts, as detected by GD2-immunocytology, was associated with a higher probability of survival, possibly because of a protective antitumor immune response after autologous stem cell transplantation. However, with the use of retroviral marking Rill et al. [13] have shown that autologous grafts from patients with neuroblastoma contain neuroblastoma cells and can contribute to relapse. It has been described that there is a decreased risk for tumor cell contamination in peripheral blood stem cell (PBSC) harvest compared to BM [14,15]. To decrease tumor contamination of the grafts CD34 positive selection can be performed on BM or PBSC collections [8,16]. However, disadvantages are costs of the procedure and a possible reduction in the total number of CD34þ cells [17]. Real-time quantitative PCR (qPCR) is a very sensitive technique to detect small numbers of tumor cells in blood, BM or PBSC harvests [18,19]. Currently, qPCR based minimal residual disease (MRD) detection is based on neuroblastoma specific RNA markers. Several small retrospective studies have investigated the association between tumor cell contamination based on qPCR and outcome [7– 11]. The most commonly used markers in these studies were tyrosine hydroxylase (TH) and GD2 synthase (GD2S). However, C
2015 Wiley Periodicals, Inc. DOI 10.1002/pbc.25507 Published online 4 May 2015 in Wiley Online Library (wileyonlinelibrary.com).
these markers are expressed in normal hematological cells (even in CD34þ selected cells), which might lead to false positive results and subsequently an overestimation of graft contamination [18]. Previously we have identified PHOX2B as a neuroblastoma specific marker [20] and we have developed a panel of markers, that is more sensitive for the detection of MRD than the use of PHOX2B alone, Additional Supporting Information may be found in the online version of this article at the publisher’s web-site. 1
Department of Experimental Immunohematology, Sanquin Research, Amsterdam, The Netherlands and Landsteiner Laboratory of the AMC, University of Amsterdam, Amsterdam, The Netherlands; 2Department of Pediatric Oncology, Emma Children’s Hospital, Academical Medical Center, University of Amsterdam, Amsterdam, The Netherlands; 3Children’s Hospital, University of Cologne, Pediatric Hematology and Oncology, Cologne, Germany; 4Department of Biostatistics, Leiden University Medical Center and Dutch Childhood Oncology Group, The Hague, Netherlands; 5Department of Pediatric Oncology, Sophia Children’s Hospital, Erasmus Medical Center, Rotterdam, The Netherlands Grant sponsor: Koningin Wilhelmina Fonds, The Dutch Cancer Society Esther M. van Wezel and Janine Stutterheim contributed equally to this work. Conflict of interest: Nothing to declare. † GPOH MRD Study Group: Ulrike Koehl, Johannes H Schulte, Felix Niggli, Michael C Fruhwald, Charlotte M Niemeyer, Udo Bode, Freimat H Schilling, Christian Schultz, Norbert Graf, Michaela Nathrath and Irene Schmid.
Correspondence to: G. A. Tytgat, Department of Pediatric Oncology, Academical Medical Center, University of Amsterdam, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. E-mail: [email protected]
Received 20 October 2014; Accepted 18 February 2015
MRD in PBSCs From Neuroblastoma Patients whereas the specificity was not lost because of stringent thresholds for positivity [18]. Kreissman et al. have recently shown in a large prospective study that purging with five monoclonal antibodies targeting neuroblastoma cell surface antigens, does not improve outcome. However, the detection of neuroblastoma mRNA (PHOX2B or TH in 245 patients) in PBSC samples is associated with an unfavorable outcome, possibly because of incomplete purging or the presence of residual tumor in the patient. Multivariate analysis (adjusting for BM status at time of harvest) was not performed [21]. We hypothesized that residual BM disease is correlated with PBSC contamination and that this residual BM disease is indeed associated with outcome. Therefore the aim of our study was to investigate whether infusion of a contaminated graft and/or the presence of residual BM disease influences outcome in a large retrospective cohort in a multivariate setting.
METHODS Patients and Treatment In this retrospective study 104 autologous stem cell grafts (PBSCs or CD34þ selected grafts) were analyzed. Samples were obtained from 104 Dutch and German patients treated with autologous stem cell transplantation as first line treatment. Highrisk patients were defined according to the international neuroblastoma staging system (INSS) as stage 4 over 1 year of age or all stages with MYCN amplification. A total of 26 grafts from Dutch patients and 78 grafts from German patients, treated in two consecutive protocols NB97 (n ¼ 37) [22] and NB04 (n ¼ 41) [23], were collected between 1997 and 2010 (Table I). In NB97 and NB04 induction chemotherapy consisted of alternating N5 and N6 cycles (Supplementary Table SI). The timing for surgery varied during induction chemotherapy. Stem cell collection was recommended as soon as the bone marrow was clear of residual neuroblastoma cells (no tumor cells or MRD on four site cytological and immunological analyses with antiGD2) (qPCR results were not used clinically) and was done after 2–4 cycles of chemotherapy. All patients were treated with melphalan, carboplatin and etoposide as consolidation chemotherapy prior to autologous stem cell rescue. After recovery of hematopoiesis patients received oral retinoic-acid. Patients treated according to the NB97 protocol before 2002 received immunotherapy instead of retinoic acid (n ¼ 13) and immunotherapy was given for 1 year (Supplementary Table SI) [24]. From a second cohort of 21 Dutch patients only BM grafts were available. These BM grafts were also tested and analyzed. All patients were treated according to the VECI protocol (Supplementary Table SII) and samples were collected between 1991 and 2000 [25]. Written informed consent for data collection and for the use of stored remains of autologous stem cell harvests for research purposes was obtained.
RNA Extraction, Reverse Transcription, and Real-Time Quantitative PCR RNA was extracted from 1 to 2 ml of autologous stem cell grafts by using the Trizol method according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA). cDNA was synthesized Pediatr Blood Cancer DOI 10.1002/pbc
1369
TABLE I. Patient Characteristics Patient characteristics Age (months) Median Minimum Maximum
