Sensitive detection of minimal residual disease and immunotherapy targets by multi-modal bone marrow analysis in high-risk neuroblastoma - a multi-center study

Abstract

Background: Bone marrow dissemination of tumor cells, common in various cancers, including neuroblastoma, is associated with poor outcome, necessitating sensitive detection methods for bone marrow minimal residual disease (MRD) and offer detection of biomarkers for therapy stratification. Current standard-of-care diagnostics, involving cytomorphological and histological assessment of bone marrow aspirates and trephine biopsies, lack sensitivity, leading to undetected MRD in many patients, and do not allow molecular biomarker assessment.

Methods: This study evaluates advanced multi-modal high-sensitivity MRD detection techniques in 509 bone marrow specimens from 108 high-risk neuroblastoma patients across two centers. We employed automatic immunofluorescence plus interphase fluorescence in situ hybridization (AIPF) and reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) panels to quantify disseminated tumor cells (DTCs), disialoganglioside 2 (GD2) and CD56/Neural cell adhesion molecule (NCAM) levels, and adrenergic (ADRN) and mesenchymal (MES)-phenotype mRNA markers.

Results: This multi-modal analysis significantly improved MRD detection compared to standard-of-care methods; 395 samples yielded results for RT-qPCR-ADRN, AIPF and CM/histology and 223 showed concordant results (64 positive, 159 negative). 114 samples did not produce results as either no cytospins were prepared (n = 96) or results were inconclusive (all techniques n = 18). AIPF and RT-qPCR complemented each other in detecting MRD and characterizing ADRN- and MES-phenotypes and GD2 immunotherapy target. RT-qPCR-ADRN alone frequently detected low tumor cell burden. High DTC infiltration at diagnosis showed bilateral bone marrow disease, whereas MRD settings often involved only one side. RT-qPCR-MES, despite lower sensitivity, identified 37 additional cases and showed delayed clearance of MES markers post-chemotherapy, with increases prior to relapse.

Conclusions: Our findings demonstrate the feasibility of integrating high-sensitivity techniques with standard-of-care assessments in an international multicenter setting. Advanced multi-modal MRD detection, monitoring phenotype switches and assessing immunotherapy targets are crucial for improving patient outcomes in neuroblastoma and other cancers.

Keywords: Automated immunofluorescence; Bone marrow; Liquid biopsies; Metastasis; Minimal residual disease; Neuroblastoma; RT-qPCR.

Fig. 1

Study design, cohort and benchmarking against standard of care bone marrow assessment. A Outline of study cohort and analytical workflows for automated immunofluorescence plus fluorescence in situ hybridization (AIPF), RT-qPCR and cytomorphology/histology. B Consort diagram depicting sample inclusion and exclusion. C Contribution of AIPF, RT-qPCR (adrenergic (ADRN)-mRNA markers) and cytomorphology (CM)/histology. Venn diagram shows samples positive for at least one technology. Each circle represents positive results of one technique. N = 395 samples analyzed by all three techniques. N = 236 samples positive by ≥ 1 technique. D Representative microscopy image for GD2, CD56/NCAM and DAPI for a bone marrow specimen that was identified as positive only by AIPF, and negative by RT-qPCR and CM/histology. Scale bar represents 20 µm. E Level of tumor cell infiltration according to AIPF (y-axis; give n as % DTCs detected by AIPF) in specimens with single or combined positivity for AIPF, RT-qPCR and CM/histology (x-axis; + positive,—negative). Box plots represent 10–90 percentiles, line shows median. ** = 0.0023; *** = 0.0003. F Level of tumor cell infiltration according to RT-qPCR-ADRN (y-axis; given as % relative to neuroblastoma cell line IMR32) in specimen with single or combined positivity for AIPF, RT-qPCR-ADRN and CM/histology (x-axis; + positive,—negative). Box plots represent 10–90 percentiles, line shows median. * = 0.0135; ** = 0.0007; *** < 0.0001

Fig. 2

High tumor cell infiltration at diagnosis is associated with bilateral bone marrow involvement and shift towards unilateral upon therapy-induced reduction of tumor cells. A AIPF and RT-qPCR positivity on one side only (unilateral) and on both sides (bilateral) in samples where both sides were analyzed for AIPF and RT-qPCR-ADRN (n = 93 AIPF samples positive; n = 201 RT-qPCR-ADRN samples positive). B Level of tumor cell infiltration by RT-qPCR-ADRN (left y-axis) and AIPF (right y-axis) according to bilateral (RT-qPCR-ADRN n = 132; AIPF n = 53) and unilateral positive (RT-qPCR-ADRN n = 69; AIPF n = 36) result at all timepoints. ** = 0.0045; ns = not significant

Fig. 3

Adrenergic RT-qPCR is more sensitive to detect MRD and AIPF detects heterogeneity in immunotherapy targets GD2 and CD56/NCAM. A Level of tumor cell infiltration according to RT-qPCR-ADRN (left y-axis; given as % relative neuroblastoma cell line IMR32) and AIPF (right y-axis; given as % DTCs detected by AIPF) per timepoint during therapy and progression or relapse (event). Box plots represent 10–90 percentiles, line shows median. Heatmap represents number of negative and positive cases. B Representative case (patient 166) showing initial response to therapy and relapse. Bone marrow liquid biopsy assessed by RT-qPCR-ADRN (left y-axis; given as % relative to neuroblastoma cell line IM32) and AIPF (right y-axis; given as % DTCs) per timepoint. DE = diagnosis, RE1 = mid-induction chemotherapy, RE2 = end of induction therapy, RE3 = surgery, RE4 = before stem cell transplantation, RE5 = before immunotherapy, RES6 = mid-immunotherapy, RE7 = at end of immunotherapy. C Scatter plot showing level of infiltration according to AIPF (y-axis; given as % DTCs) and RT-qPCR-ADRN (x-axis; given as % relative to neuroblastoma cell line IMR32) in bone marrow from all timepoints analyzed by both techniques (n = 395; Spearman correlation = 0.69, 95% CI 0.57–0.78; p < 0.001 of samples positive for both techniques (n = 104)). D Adrenergic mRNA-marker (co-)expression by RT-qPCR on samples with positive result all timepoints (n = 215). E Automated immunofluorescence plus iFISH (AIPF) of patient 166 showing heterogeneous levels of GD2 and CD56/NCAM at event versus homogeneous GD2 at diagnosis (DE). Tumor cell identity was confirmed by iFISH for MYCN amplification and NME as reference. F Representative images of tumor marker protein expression by imaging mass cytometry (IMC) in patient 166 illustrating loss of GD2 at event (relapse). G Imaging mass cytometry (IMC) of patient 166. Tumor marker protein expression (y-axis) in tumor cells at diagnosis (DE) and event (relapse)

Fig. 4

Mesenchymal markers identify MRD in adrenergic negative and AIPF negative bone marrow liquid biopsies prior to relapse. A Contribution of mesenchymal mRNA RT-qPCR-markers (RT-qPCR-MES),RT-qPCR-ADRN, AIPF and cytomorphology (CM)/histology. Venn diagram shows samples positive for at least one technology. Each circle represents positive results of one technique. N = 395 samples analyzed by all techniques; N = 273 samples positive by ≥ 1 technique. B MES-positive samples by RT-qPCR-MES. Level of infiltration according to AIPF (y-axis; given as % DTCs) and RT-qPCR-ARDN (x-axis; given as % relative to neuroblastoma cell lines IMR32) (n = 107; Spearman correlation = 0.59, 95% CI 0.33–0.77; p < 0.001 of samples positive for both techniques (n = 38). C Level of infiltration by AIPF (y-axis; given as % DTCs) versus MES-positivity (x-axis). Box plots represent 10–90 percentiles, line shows median. Heatmap represents number of negative and positive cases. * = 0.0377. D Level of infiltration by RT-qPCR-ADRN (y-axis; given as % relative to neuroblastoma cell line IMR32) versus MES-positivity (x-axis). Box plots represent 10–90 percentiles, line shows median. Heatmap represents number of negative and positive cases. ** = 0.0029. E Representative case (patient 166) showing MES marker positivity prior to relapse. At diagnosis adrenal tumor with MYCN amplification and multiple bone metastases and bone marrow infiltration; patient suffered a relapse one month after completion of SIOPEN/HR-NBL1 treatment and died 20 months later from disease. Bone marrow samples assessed by RT-qPCR-ADRN and RT-qPCR-MES (normalized mRNA expression left y-axis) and AIPF (given as % DTCs, right y-axis) per timepoint (x-axis). DE = diagnosis, RE1 = mid-induction chemotherapy, RE2 = end of induction therapy, RE3 = surgery, RE4 = before stem cell transplantation, RE5 = before immunotherapy, RES6 = mid-immunotherapy, RE7 = at end of immunotherapy