The study had ethical review board approval (REC reference 14/WM/1253 “Establishing primary cultures and cell lines from pediatric cancers”) and samples were made available following informed consent. Patients were eligible if there was a known or suspected diagnosis of neuroblastoma. All samples were included based on known or suspected diagnosis of neuroblastoma without selection on clinical criteria. In some cases with insufficient material for culture, the transport fluid used for sample transfer (0.9% saline) was placed directly into culture.

Tru-cut needle biopsies or tissue from tumor resections from neuroblastoma patients were transferred directly from operating theatre to the hospital histopathology laboratory. Following sterile cut up and routine diagnostic processing including freezing of material for research, surplus material was evaluated by a consultant pathologist for tissue viability and made available for culture. Tissue was manually disaggregated in a 10cm tissue culture dish using a sterile scalpel. Tumors that dissociated readily in this way were placed immediately into stem cell media in 25cm ² flasks, 24 well plates or 12 well plates depending on the available tissue. Since it is not possible to perform an accurate cell count on the partially dissociated tumor, choice of culture container was based on estimation of size that would yield approximately 100% confluence, were the sample fully disaggregated. Where spare solid material was unavailable from a biopsy, up to 1ml of saline used to transport the sample to the histopathology laboratory was taken and added to 4ml of stem cell media in a 25cm ² flask. When more than 1ml of transit fluid was provided, it was centrifuged at 300G for 5 minutes and the pellet resuspended in stem cell media. Samples that were tougher to disaggregate were digested with Accutase (Thermofisher 00-4555-56) for up to 1 hour, or until obvious disaggregation was observed, and divided between wells of a multiwell plate based on the estimation of confluence as above.

Disaggregated cells and tumor fragments were placed into standard serum free neurosphere culture media (Stem Cell Media, SCM) composed of DMEM/F12 (Sigma-Aldrich D8437),° 1% B27 (Gibco 17504044) with 20ng/ml EGF (Sigma Aldrich #E9644) and 20ng/ml FGF (Fibroblast Growth Factor, Peprotech 100 – 18B) in the presence of penicillin and streptomycin antibiotics. Cells were propagated at 37°C in 5% CO ₂. Cultures were inspected twice a week for cell density and presence of neurospheres, and split 1:2 to 1:5 depending on cell density and speed of growth. Cultures were frozen in serum free DMEM/F12 media with 10% DMSO at approximately 10 ⁷ cells per ml.

A total of 14 randomly selected representative primary cultures were analyzed by flow cytometry. To produce a single cell suspension necessary for this, cells growing as a monolayer were detached with Accutase (Thermofisher) according to the manufacturer’s instruction, while neurospheres were disaggregated with Accutase and the mechanical force of gentle pipetting. Surviving cells, re-suspended into single cell suspension, were stained in FACS tubes using saturating amounts (1–5ul of stock) of antibody in 100ul PBS using the following panel of directly conjugated monoclonal antibodies all from Biolegend: GD2-PE (357304, RRID: AB_2561885), CD56-APC/Cy7 (318332, RRID: AB_10896424), CD45-BV711 (304049, RRID: AB_2563465) controlled for nonspecific staining using isotype-matched labelled polyclonal antibodies. The stained primary culture cells were incubated on ice with antibody mix for 30 minutes and then washed with PBS and centrifuged at 300G for 5 minutes. Data was collected with the BD LSRII flow cytometer (BD Biosciences) and data analysis used FlowJo software (v8.8.3).

The gating strategy excluded dead cells by a live/dead stain using DAPI. To exclude leukocytes that may be present from blood in the original sample, the resulting live population was gated on CD45 negative and CD56 positive cells, the latter a standard marker for neuroblastoma. This gated cell population of CD45-ve/CD56+ve cells was then examined for expression of GD2, also ubiquitously expressed on neuroblastoma.

Chi-squared tests were used to determine significance of association of categorical variables; for example successful versus non successful expansion correlated with pre versus post chemotherapy. Data were tabulated and tests of significance were performed using Microsoft Excel 2016 and GraphPad Prism v6.0

Between October 2014 to end of 2016, tumor biopsies or resections of 67 consecutive neuroblastoma samples from 52 patients treated at Great Ormond Street Hospital London were systematically evaluated for their ability to establish primary neurosphere lines. Of these, 39 were primary needle biopsies and 28 were surgical resections of which 24 were post chemotherapy resections and 4 were primary (treatment-naïve) resections.

The success in terms of establishment of cultures was determined by the ability of cultures to produce discrete visible neurospheres. All cultures derived from neuroblastoma samples were observed to grow as a mixture of phenotypes including neurospheres, single suspension cells, and as monolayer, although often one of the three growth patterns was predominant. Cases could be dichotomized into those that formed spheres within the first 14 days of establishment and then went on to form long term successful cultures, and those that never formed spheres and were classified as “unsuccessful cultures”. The overall success rate, as thus defined, was 55%. It appeared that successful establishment of neurospheres could be made from chemotherapy-naïve primary tumors and post chemotherapy surgical samples, as well as primary needle biopsies. Successful establishment of lines is summarized in Table 1. Of 43 chemotherapy-naive samples (39 biopsies and 4 surgical excisions), 29 (67%) were successfully established, whereas for the post-chemotherapy surgical samples success rate was 8 out of 24 (33%), which is significantly inferior (Chi Sq p=<0.008) (see underlying data ⁸ ). This suggests that the protocol, if applied to optimally procured tissue, would have a success rate over 65%.

Due to the impossibility of counting cells both at the start of culturing and following establishment of spheres, it is not possible to plot growth curves. However, the range of time from initial seeding to establishment of a confluent 75cm ² flask is approximately 2 to 10 weeks showing the marked heterogeneity of growth rates.

In order to determine that these were neuroblastoma cells, 14 representative samples were analyzed by flow cytometry. Cells from primary cultures were assessed for expression of the pan leucocyte marker CD45, which neuroblastoma cells do not express, and for ubiquitously expressed NBL markers CD56 and GD2. Of the samples tested, 79% exhibited this characteristic neuroblastoma staining pattern. We found that the neurosphere cultures showed a very bright and homogeneous staining for both NBL markers ( Figure 1)

We were interested in whether it is possible to establish neuroblastoma lines when using very small amounts of tissue. In cases with insufficient tissue to culture, 1ml of transportation fluid was put directly into 4ml stem cell media (SCM) in a 25cm ² tissue culture flask, testing the hypothesis that small amounts of residual tumor cells within the transport media would be able to establish. Where transport fluid provided was more than 1ml, this was centrifuged and the pellet resuspended in SCM. Cultures established from transit fluid had a comparable success rate to those established from solid tissue samples (9 out of 15= 55% versus 25 out of 39 =64% for biopsies: Chi Sq p=0.78).

To perform a further unbiased comparison, eight patients had neurospheres cultured from the same biopsy with separate culturing of the solid tissue and surgical transit fluid. In three there was no successful growth, in three there was concordance of growth from the two sources, and in two there was success from solid tissue but not transit fluid; there were no cases with success of transit fluid but failure from solid tissue. Therefore transit fluid appears a less reliable source of material but a useful adjunct in the context of scarce material.

In addition to growth as neurospheres, primary cultures established in SCM were observed to grow as adherent monolayers, single cell suspensions, or mixtures of all three growth patterns. Moreover, there is marked heterogeneity in the size of neuropheres both within and between samples. It is not possible to quantify this heterogeneity, but all the NBL SCM lines contained some elements of adherent, single cell suspension and neurospheres. SCM derived from gliomas can be grown as adherent cultures by coating culture flasks with laminin and so we were interested in whether neuroblastoma primary cultures could also adopt this phenotype. We did not attempt initial establishment in laminin in any cases. In our hands, when established or establishing SCM lines were transferred to laminin, the degree of attachment increased markedly, with outgrowth of adherent spheres and associated adherent cells growing out as a monolayer. In some cases where growth in suspension appeared slow, the change to laminin appeared to accelerate growth. Because of the impossibility of counting cells in neuropheres without majorly disrupting them, it was not possibly to quantify these aspects of altered growth in laminin. On return of laminin cultured cells to non-coated flasks, they reverted to predominant suspension growth pattern.

Patient characteristics and ability to establish neurospheres is shown in Table 2. The number of male samples were 37 out of 67. The ability to establish neurospheres did not vary depending on gender (59% vs 50%). Biopsies which demonstrated only undifferentiated neuroblastoma had a higher but non-significant success rate in establishing neurospheres than those which showed any evidence of differentiation or post chemotherapy changes (60% vs 45%). Age, stage, MYCN amplification and the presence of segmental chromosomal aberrations were all non significant in terms of neurosphere success rate ( Table 2).

Six patients had two samples collected at different time points during treatment. For four of them there was no concordance for success between the timepoints whereas two patients (one success and one failure) the different timepoints were concordant. Therefore we failed to find any evidence that patient-specific factors govern success rate.

Open Science Framework: Primary neuroblastoma cultures database. https://doi.org/10.17605/OSF.IO/T2RFB ⁸

This project contains the following underlying data:

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).