Frequency and Prognostic Impact of ALK Amplifications and Mutations in the European Neuroblastoma Study Group (SIOPEN) High-Risk Neuroblastoma Trial (HR-NBL1)

PURPOSE In neuroblastoma (NB), the ALK receptor tyrosine kinase can be constitutively activated through activating point mutations or genomic amplification. We studied ALK genetic alterations in high-risk (HR) patients on the HR-NBL1/SIOPEN trial to determine their frequency, correlation with clinical parameters, and prognostic impact. MATERIALS AND METHODS Diagnostic tumor samples were available from 1,092 HR-NBL1/SIOPEN patients to determine ALK amplification status (n = 330), ALK mutational profile (n = 191), or both (n = 571). RESULTS Genomic ALK amplification (ALKa) was detected in 4.5% of cases (41 out of 901), all except one with MYCN amplification (MNA). ALKa was associated with a significantly poorer overall survival (OS) (5-year OS: ALKa [n = 41] 28% [95% CI, 15 to 42]; no-ALKa [n = 860] 51% [95% CI, 47 to 54], [P < .001]), particularly in cases with metastatic disease. ALK mutations (ALKm) were detected at a clonal level (> 20% mutated allele fraction) in 10% of cases (76 out of 762) and at a subclonal level (mutated allele fraction 0.1%-20%) in 3.9% of patients (30 out of 762), with a strong correlation between the presence of ALKm and MNA (P < .001). Among 571 cases with known ALKa and ALKm status, a statistically significant difference in OS was observed between cases with ALKa or clonal ALKm versus subclonal ALKm or no ALK alterations (5-year OS: ALKa [n = 19], 26% [95% CI, 10 to 47], clonal ALKm [n = 65] 33% [95% CI, 21 to 44], subclonal ALKm (n = 22) 48% [95% CI, 26 to 67], and no alteration [n = 465], 51% [95% CI, 46 to 55], respectively; P = .001). Importantly, in a multivariate model, involvement of more than one metastatic compartment (hazard ratio [HR], 2.87; P < .001), ALKa (HR, 2.38; P = .004), and clonal ALKm (HR, 1.77; P = .001) were independent predictors of poor outcome. CONCLUSION Genetic alterations of ALK (clonal mutations and amplifications) in HR-NB are independent predictors of poorer survival. These data provide a rationale for integration of ALK inhibitors in upfront treatment of HR-NB with ALK alterations.


INTRODUCTION
Neuroblastoma (NB), the most frequent solid, extracranial malignancy in children, exhibits wide clinical and genetic heterogeneity. High-risk neuroblastoma (HR-NB), defined as metastatic disease over the age of 12 months or MYCN-amplified (MNA) disease at any age, remains associated with long-term survival rates of only 50%. 1 Current treatment approaches consist of intensive induction chemotherapy, surgical resection of the primary tumor, consolidation with high-dose chemotherapy (HDC), and autologous stem-cell rescue, and for minimal residual disease, isotretinoin in combination with human or mouse chimeric anti-GD 2 antibody, ch14.18. [2][3][4][5][6][7][8] In NB, several recurrent genetic alterations have been described. MNA is a strong biomarker associated with rapid tumor growth. 9 Other copy-number alterations occur over more extensive chromosome regions, with segmental chromosome alterations being associated with a poor outcome. 10 Recurrent mutations have been described in the RAS-MAPK pathway, chromatin remodeling genes (ATRX and ARID1A), and TERT rearrangements. [11][12][13][14] Activating anaplastic lymphoma kinase (ALK) mutations are the most frequent mutations in NB, occurring in both familial and sporadic cases, with somatically acquired ALK mutations (ALKm) observed in 6%-12% of sporadic NBs in all risk groups. [15][16][17][18] These ALK activating mutations are localized most frequently within the kinase domain at hotspots identified at the F1174, R1275, and F1245 positions, with mutations occurring both at clonal (. 20% mutated allele fraction [MAF]) or subclonal levels (, 20% MAF). [19][20][21][22][23] ALK can also be activated by genomic focal amplification, described in 1%-2% of NBs, almost exclusively with MNA, 17,24 or, more rarely, following structural rearrangements. 25 Genetic alterations of ALK are associated with poorer survival in the overall NB population. 24,26 However, their prognostic role in HR-NB has been less well studied. 10,17,24 Altogether, ALK alterations are an important molecular target, given the role of ALK as a driver oncogene in NB and its actionability with small molecule therapies. [27][28][29] To determine the frequency of ALK alterations (mutations and amplifications), their correlation with clinical characteristics, and their prognostic impact in HR-NB, we analyzed a large series of 1,092 diagnostic NB samples from patients on the HR-NBL1/SIOPEN trial.

Patients and Samples
Patients were treated within the HR-NBL1/SIOPEN Protocol (ClinicalTrials.gov: NCT01704716, EudraCT: 2006-001489-17; Protocol [online only]), an international, randomized, multiarm, open-label, phase III trial. [2][3][4][5]30,31 Patients with International Neuroblastoma Staging System stages 2, 3, 4, or 4S with MNA, or International Neuroblastoma Staging System stage 4 without MNA $ 12 months of age at diagnosis were eligible for the trial up to 20 years of age. Within the trial, several randomized treatment arms were conducted over different periods (Appendix Fig A1, online only). Induction random assignments included the following: R0-random assignment of prophylactic granulocyte colony-stimulating factor during rapid COJEC induction 31 ; R3-comparison of two induction regimens, rapid-COJEC versus modified N7. 32 HDC was evaluated in the R1 random assignment: busulfan or melphalan versus carboplatin or etoposide or melphalan. 3 Anti-GD 2 immunotherapy random assignments during maintenance phase were explored in R2 (2009-2013) and R4 (2014-2017), both comparing dinutuximab beta with oral isotretinoin to dinutuximab beta and subcutaneous interleukin-2 with oral isotretinoin, but with altered schedules. 5,30 In the interim, dinutuximab beta with oral isotretinoin was the recommended standard.
Patients were enrolled on the HR-NBL1/SIOPEN trial after approval by national regulatory authorities and by national, and institutional, ethical committees or review boards in participating countries. Parents or guardians and patients according to age provided written informed consent for treatment, data collection, and analysis.
The ALK analysis cohort consisted of patients for whom a contributive tumor sample obtained at diagnosis was available in a SIOPEN reference laboratory 33 for additional molecular analysis with available follow-up data (Fig 1).
MYCN status and tumor genomic copy-number profiles were determined in SIOPEN reference laboratories as described previously. 10,[33][34][35][36] Samples were required to contain at least 20% tumor cells on pathologic examination.
The ALK amplification (ALKa) status was evaluated using either fluorescence in situ hybridization and/or multiplex ligation polymerase chain reaction-dependent amplification, array comparative genomic hybridization (aCGH), and/or array single-nucleotide polymorphism according to established guidelines. 10,33,34,37 ALK gene amplification was defined as more than fourfold increase of ALK signals in relation to numbers of chromosome 2 by fluorescence in situ hybridization, or as more than 10 copies of the gene estimated by multiplex ligation-dependent amplification, aCGH, or array single-nucleotide polymorphism.  20,22 MAF $ 20% were defined as clonal events and MAF , 20% as subclonal events, as reported previously. 20,22 No correction for tumor cell content was undertaken when reporting MAF. Mutations identified by Sanger sequencing were considered clonal. All detected mutations were validated by a second independent experiment: for clonal events, TDS data were validated by Sanger sequencing, and for subclonal events, NGS or TDS was validated in an independent second experiment.
Standard bioinformatics were used to detect mutations in NGS experiments as previously reported. Mutations in TDS experiments were determined as described previously. 20,22 In brief, to highlight mutations, in each NB sample, the frequencies of each base at each position of the analyzed regions were compared with those observed in all other samples and controls. This approach enabled the identification of mutations with a statistically significant increase in percentage of a variant base, compared with background noise.

Statistical Analysis
Event-free survival (EFS) was calculated from diagnosis to the first relapse, progressive disease, secondary malignancy, or death from any cause, or until last patient contact. Overall survival (OS) was calculated from diagnosis to death from any cause, or until the last patient contact. EFS and OS were estimated using the Kaplan-Meier method and compared using the logrank test, and if indicated with pseudo-value regression for 5-year OS. [39][40][41] EFS and OS are presented as 5-year point estimates together with 95% CIs using log-log transformation. 41 To adjust for established risk-factors (age at diagnosis, stage, number of metastatic compartments, and MYCN amplification), a Cox proportional hazards regression model was used.  Correlations between patient and disease characteristics and ALK genetic alterations were explored using chi-square tests.
To allow for sufficient follow-up time, only patients enrolled until December 31, 2019, were considered. The data cutoff for the final analysis was October 3, 2020. We calculated median follow-up using the inverse Kaplan-Meier estimate. Statistical analysis was performed using SAS (version 9.4).

RESULTS
Of 3,334 patients enrolled on the HR-NBL1/SIOPEN trial between November 24, 2002, and December 31, 2019, 1,092 patients were included in the ALK analysis cohort (Fig 1; Appendix Table A1, online only). Patients were accrued from 132 SIOPEN member institutions or hospitals in 19 countries (Appendix Table A2, online only). Among these 1,092 patients, 81% (889 out of 1,092) were . 18 months of age at diagnosis, 47% (521 out of 1,092) showed MNA, and 88% (966 out of 1,092) had stage 4 disease, with no statistically significant difference in EFS or OS between the ALK analysis cohort and the overall HR-NBL1 cohort (Appendix Fig A2, online only). 42 The median follow-up period was 6.8 years (0.1-17.4 years).

ALK Alterations
Within the ALK cohort, the ALKm status was analyzed in 762 patients, the ALKa status in 901 cases, with both ALKm and ALKa studied in 571 patients (Fig 1, Table 1).
ALK alterations were detected in 146 out of 1,092 patients with ALKa occurring in 4.5% (41 out of 901 cases) and ALKm in 13.9% (106 out of 762 cases). Only one case showed ALKa and a concomitant ALK R1275Q mutation with an MAF of 93%, suggesting that the mutated allele is contained in the amplicon (Appendix Fig A3, online only).

ALK Amplification and Correlation With Risk Factors
High-level genomic amplification of the ALK gene was found in 4.5% (41 out of 901) of cases (Fig 2A, Table 1). All but one also had MNA. ALKa significantly correlated with MNA (P , .001), non-stage 4 disease (P , .001), and age at diagnosis , 18 months (P 5 .005). No correlation between the presence of ALKa and response at the end of induction treatment was observed.
There were no statistically significant correlations between ALKm and stage, age at diagnosis, or localization of the primary tumor (adrenal, abdominal, or other) (Table 1). However, a significant correlation was observed between the presence of an ALKm and MNA (P , .001), with an enrichment of ALKm F1174 in MNA tumors (P 5 .0005).   This was also observed when analyzing only stage 4 tumors.
No correlation between ALKm and response at the end of induction treatment was observed.
No statistically significant difference in outcome was observed between patients harboring any ALKm versus none ( Fig 3B, Table 2).
Patients with metastatic disease (stage 4 or 4S MNA) and a clonal ALKm showed a trend toward poorer OS. However, in patients with localized disease, the presence of ALKm did not confer poorer survival (Table 2).

Overall Prognostic Impact of ALK Genetic Alterations
To determine the overall prognostic impact of ALK genetic alterations, we focused on the subgroup of 571 patients with both known ALKa and ALKm status. In this subgroup of patients, a statistically significant poorer OS was observed in patients whose tumors harbored any ALK alteration ( Table 2).
Among the subgroup of patients with known ALK status, we sought to determine the prognostic impact of ALK alterations according to the different treatment arms of HR-NBL1. Indeed, in the HR-NBL01/SIOPEN trial, the introduction of busulfan and melphalan as standard for HDC, and anti-GD 2 maintenance therapy as a new standard since 2010, has led to significantly improved survival (Appendix Fig A5F, online only). [3][4][5] Importantly, when considering patients treated according to the SIOPEN standard with busulfan and melphalan HDC and maintenance immunotherapy, the presence of an ALK alteration (ALKa or clonal ALKm) remained associated with a poorer 5-year OS of 48% (95% CI, 28 to 65), versus no ALK alteration 67% (95% CI, 56 to 75); P 5 .03 (Fig 3F, Appendix Table A3, online only), with a trend also observed when taking into account all ALKm (clonal and subclonal, P 5 .059).
Based on univariate risk factor exploration of the whole ALK analysis cohort (Appendix Fig A5), we developed a Cox model for multivariate analysis including clinical and biologic parameters previously shown to be of prognostic impact (n 5 571 patients  (Table 3).

DISCUSSION
In HR-NB, the identification of prognostic biomarkers is crucial for the development of new treatment approaches.
Recent studies have shown that MNA is not associated with poorer outcome among the overall cohort of patients with HR-NB, but the presence of genomic amplifications other than MYCN might constitute a poor outcome biomarker. 43 We now show in this large ALK analysis cohort that the presence of ALKa or clonal ALKm resulted in significantly worse outcome.
Given the oncogenic driver role of ALK activation, and the prognostic impact of ALKa or clonal ALKm, the introduction of frontline ALK-targeted treatment is now strongly supported by the current study. Although early phase clinical trials of firstand second-generation ALK inhibitors showed modest efficacy of the first-generation inhibitor crizotinib in NB with F1174 hotspot mutations being resistant, 44 third-generation ALK inhibitors such as lorlatinib exhibit improved efficacy alone and when combined with chemotherapy. 28,[44][45][46] Crizotinib is currently being administered with chemotherapy in a phase III upfront trial for patients with HR-NB with ALK alterations (ClinicalTrials.gov: NCT03126916).
Improvements in HR-NB patient survival have been achieved with intensification of HDC and immunotherapy with dinutuximab (ch14.18/Sp02 and ch14.18/CH0), [3][4][5]7 and our results highlight the potential of ALK inhibition as an attractive upfront precision-medicine strategy in patients with ALK alterations to further improve survival. Importantly, in patients reaching the maintenance treatment phase   with dinutuximab beta in the HR-NBL1/SIOPEN trial, the presence of an ALK alteration was still associated with poorer survival, thus strongly suggesting that integration of ALK-targeted therapy is warranted throughout all treatment phases of modern-era HR-NB therapy.
ALKa was observed in 4% of NB cases, accounting for approximately 1 out of 3 of ALK-activated NB cases. To date, co-occurrence of ALK hotspot mutations and genomic amplification has rarely been reported in NB. 17 In this extensive cohort of patients, one case harboring both ALKa and an R1275 ALKm was identified. This indicates that these alterations are not fully mutually exclusive, although co-occurrence is extremely rare.
ALKm were found in 13.9% of cases at the studied exonic regions harboring known ALK mutational hotspots. 17,24 This is higher than previously reported frequencies of ALKm in HR-NB of approximately 10%, most likely as previous reports using Sanger sequencing or standard-resolution NGS approaches. 24,26 Sanger sensitivity is limited to the detection of MAF . 15%-20%, but in NB, ALK mutations with lower MAFs have been reported. 14,[19][20][21] Ultradeep sequencing used in this analysis has a sensitivity limit of MAF of 0.1%. 19,20 This approaches the theoretical limit of detection based on the genomic DNA input of 50 ng for one experiment, equivalent to 5,000 diploid genomes. This study demonstrates that use of higher-resolution techniques enables a higher detection rate of ALKm. The MAF distribution indicated a majority of clonal events (76 out of 106 cases). Importantly, clonal ALKm were associated with poorer outcome and were of independent prognostic significance, but subclonal events were not. Subclonal events, defined in this study by MAF , 20%, comprised 28% (30 out of 106) of all ALKm, with a very low MAF (, 5%) observed in 19 cases.
However, when considering ALKm, the OS remains poor in all patient subgroups (5-year OS , 62%). Furthermore, although of different prognostic impact in this study, the biomarker (ALK mutation) might not be of distinct predictive impact, and even in patients with subclonal ALK mutations, ALK inhibitor treatment might be effective in the targeted cell population. Thus, future upfront trials should consider ALKtargeted treatment based on clinically applicable reliable detection limits (for instance MAF 5% for NGS techniques) rather than the MAF defining prognostic subgroups.
As tumor samples harbored at least 20% tumor cells by pathologic examination, with additional confirmation provided by a dynamic aCGH or SNPa profile in the majority of cases, the observed low MAF is likely to correspond to intratumoral heterogeneity. In NB, intratumor heterogeneity has been reported for MNA and segmental chromosome alterations. [47][48][49] The coexistence of ALK nonmutated and mutated cells within a single tumor suggests that these different subclones might coexist in an advantageous equilibrium, which might crucially affect the dynamics of cancer progression. 50,51 Correlation with pathologic findings, single-cell RNA or DNA experiments, and in situ approaches will elucidate how ALK-mutated cells are distributed throughout an NB. A higher frequency of ALKm at NB relapse has been demonstrated, suggesting clonal evolution of a minor ALK-mutated subclone to a dominant ALK mutated clone at relapse, but these cases might not represent clinically unfavorable cases initially. 23,52,53 Further studies focusing on serial blood samples for ctDNA studies will further elucidate clonal evolution, also under targeted therapy. 54 In HR-NB, mutations in the p53 or RAS-MAPK pathways, including ALK, together with telomere maintenance caused by induction of telomerase or ALT (alternative lengthening of telomere) are thought to increase tumor aggressiveness, resulting in even poorer survival among patients with HR-NB. 55,56 As MYCN leads to upregulation of TERT expression, MNA associated with any ALK alteration might lead to inferior outcome. Cases with ALKa show both ALK pathway activation and activation of telomere maintenance through MNA, with a suggested additive effect of these genetic events. The very poor survival of ALKa patients is concordant with this observation. However, survival of patients whose tumors harbored ALKm and MNA was not different from those without MNA, suggesting that ALKm cases constitute a more heterogeneous group with regards to the mechanistic tumor classification. 55 ALKa and ALK clonal mutation were both independent predictors of poor outcome in our multivariate Cox model. Notably, the end-of-induction response rate was not associated with ALK genetic alterations, suggesting that ALK-altered tumor cells are unlikely to be primarily chemotherapy resistant.
In summary, our data contribute to the rationale for future clinical trials introducing ALK-targeted treatment in the frontline setting together with chemotherapy and immunotherapy, and the distinct prognostic impact of different ALK alterations (ALKa and ALKm) needs to be considered.              . 20 Sanger NOTE. Among these patients, ALK amplifications were detected in eight cases, and clonal ALK mutations were detected in 21 cases. In addition, six cases with subclonal mutations are also listed. Abbreviations: ALK-A, ALK-amplified; ALK-NA, ALK not amplified; BUMEL, busulfan and melphalan; CEM, carboplatin, etoposide, and melphalan; COJEC, chemotherapy regimen, details in Figure A1;