MYCN protein stability is a better prognostic indicator in neuroblastoma

A cohort of 53 neuroblastoma patients was selected as the main study object. They received curative surgery at Shanghai Children’s Medical Center (SCMC), Shanghai, China, between January 2010 and September 2019. 28 tumor samples of this cohort were MYCN amplification tested by FISH (MYCN FISH⁺), which was the maximum count of MYCN FISH⁺ prechemotherapy samples suitable for the IHC test during this time. As a control, 25 patients with MYCN FISH⁻ tumors were chosen according to their clinical consequences: 1) 8 patients died from tumors, 2) 17 patients had a favorable long-term prognosis. Follow-up within this cohort was completed on December 31, 2019. To ensure prognostic accuracy for individuals, only 41 patients (including 16 with FISH⁺ tumors and 25 with FISH⁻ tumors) of this cohort diagnosed in or before 2016 were included when referred to the follow-up time. More detailed clinical information was listed in Table 1 and Table S1.

Another two cohorts of 71 and 127 patients were identified as the validation cohorts for FISH and IHC results, respectively. Diagnostic tumor samples from the cohort of 71 patients were tested by whole exome sequencing (WES) and FISH at the same time, and those from the cohort of 127 patients were tested by MYCN IHC (MYCN antibody: # 51705, Cell Signaling Technology) and FISH at the same time. Because their other clinical information was not involved in this study, we would not further enumerate them.

All 53 samples were evaluated MYCN amplification status by FISH using 2 μm formalin-fixed, paraffin-embedded (FFPE) sections. Laboratory-developed probes targeting MYCN gene (2p24) were used. Tissue sections were washed with SSC buffer and mounted in 4′, 6-diamidino-2-phenylindole for nuclear counterstaining. The results were analyzed and interpreted following the probe manufacturer’s instructions. MYCN FISH⁺ at region 2p24 showed red signals (Fig. 1a). If the copy numbers of MYCN were ≥ 5 per haploid genome, related patients were classified into the “MYCN FISH⁺” group.

Tumor specimens were fixed in 10% formalin and embedded in paraffin as soon as they were obtained from patients. Pathologists chose specimens with the highest tumor content by H&E staining to navigate tumor pathological heterogenicity. MYCN IHC was performed on the same specimen in which MYCN FISH was performed or on a different specimen of the same tumor if that was unavailable. The performance of two anti-MYCN antibodies (#84406 s and # 51705, Cell Signaling Technology) in IHC was compared. Subcutaneous tumors of MYCN amplified SK-N-BE(2) cell line were used as positive controls, and subcutaneous tumors of MYCN non-amplified SY-5Y cell line as negative controls. Scores for staining intensity were graded on a scale of 0–3 (0 = negative, 1 = weak, 2 = moderate, 3 = strong), while the positive proportion was graded on a scale of 0–4 (0 for 0%, 1 for < 25%, 2 for 25–50%, 3 for 50–75%, 4 for 75–100%). The IHC score was calculated independently by two pathologists blinded to the FISH results, based on the formula that final score = staining intensity * positive proportion. Samples with scores of 0 were defined as “IHC = 0”. Samples with scores of 1–9 were classified into the “low expression” group. Beyond that, samples belonged to the “high expression” group.

Total RNA was extracted and purified using the Qiagen RNeasy Mini kit (Valencia, CA, USA) according to the manufacturer’s instructions. The quality of RNA was assessed by a bioanalyzer before sequencing [20]. RNA libraries for RNA-seq were based on TruSeq Stranded Total RNA Gold library by Novaseq S4 PE150 (Illumina). Regrettably, only 14/53 samples detected by MYCN IHC had frozen tumor tissues and had been fully sequenced.

WB was performed as previously described [21] against the following antibodies: rabbit anti-MYCN antibody (1:1000) (#84406 s, Cell Signaling Technology), mouse anti-β-actin antibody (1:10000) (HF1024, HuaAn).

The accession numbers of RNA-seq data and clinical information reported in this paper were SEQC Dataset (GSE62564) from the GEO website and TARGET Dataset deposited at the cancer genome atlas website.

As MYCN amplification closely correlates with the neoplastic prognosis of neuroblastoma [4], identifying MYCN amplification status is greatly important for related patients. The ratios of MYCN amplification among the SEQC dataset (n = 493), TARGET dataset (n = 243) and SCMC dataset (n = 568) were compared. Among the results, ratios in the SEQC dataset (n = 18.7%) and TAGET dataset (n = 27.6%) were within the universally acknowledged positive rate [22], while the ratio in the SCMC dataset (n = 13.4%) was far below (Fig. 1b). WES was performed to test the accuracy of SCMC FISH results. The results showed that the positive rates were almost equal between two methods (Fig. 1c), which meant that the FISH results in our hospital were reliable.

Then, patients diagnosed at SCMC during 2013–2014 were further analyzed. The prognosis of patients with MYCN FISH⁺ tumors was poor (p < 0.05, Kaplan-Meier survival analysis) (Fig. 1d). By categorizing patients with MYCN FISH⁻ tumors according to INSS stage, we found that the adverse event rate significantly increased when patients progressed to stage 4 (Fig. 1e). However, for patients with MYCN FISH⁺ tumors, their prognosis was poor once they developed to stage 3 (Fig. 1f). These results suggested that stage 1/2 and 4S patients usually had a promising future after rational treatment. If stage 3 patients with MYCN FISH⁺ tumors or stage 4 patients were older than 18 months at first diagnosis, their prognosis could hardly be sanguine even under the most aggressive treatment. Similar trends were found in database analysis (Supplemental Fig. 1). Overall, the MYCN gene test is most valuable for INSS stage 3 patients in predicting prognosis.

Whether FISH or WES, both focus on MYCN gene status, while protein is the real one performing biological functions. However, there is no reliable antibody that could be clinically useful in IHC. Pathologists at SCMC tried various MYCN antibodies for a long time and finally chose one (MYCN antibody: #51705, Cell Signaling Technology) used in IHC. Between 2010 and 2015, 127 tumor samples were detected MYCN amplification status simultaneously by FISH and IHC (MYCN antibody: #51705, Cell Signaling Technology) at SCMC. Compared with FISH data, IHC results only had 43.1% concordance (50/116) in the MYCN FISH⁻ group and 36.4% concordance (4/11) in the MYCN FISH⁺ group (Table S2), which meant its specificity and sensitivity were substandard. Finding a reliable antibody that could rapidly and accurately detect MYCN protein expression was urgent. An antibody (#84406 s, Cell Signaling Technology), never used in the IHC test before, was chosen.

53 prechemotherapy samples were obtained, and 28 were MYCN FISH⁺ (Table 1). IHC detection revealed that their staining intensity and positive proportion of malignant cells showed a remarkable difference (Fig. 2a). With FISH results as standard, MYCN IHC could accurately detect more than 80% of cases regardless of whether the MYCN gene was amplified (Table 2, Fig. 2b). For those cases with inconsistent results, prognostic information was involved to verify which method was more reliable. The results showed that IHC scores increased with INSS stage (Fig. 2c, d). In the MYCN FISH⁺ group, 7/11 patients with high MYCN protein expression (IHC score ≥ 9) had a poor prognosis, whereas 5/5 with no or low (IHC score < 9) recovered well. In the MYCN FISH⁻ group, MYCN protein was not detected in 17/17 event-free patients’ tumors, but it was positive for 5/8 patients who died of neuroblastoma (Table 3-4, Fig. 2e). These data suggest that this MYCN antibody (#84406 s) has reasonable specificity and sensitivity.

More analyses were performed to detect the value of MYCN IHC in clinical prediction. The results showed that IHC could further distinguish patients with different clinical outcomes in the MYCN FISH⁺ and FISH⁻ groups (Fig. 3a-b). If specifying patients only by FISH results, the MYCN FISH⁺ and FISH⁻ groups’ EFS rates were 56.3 and 68.0%, respectively. There was no significant difference (p > 0.05, Kaplan-Meier survival analysis). Grouped according to IHC results did better, and their difference was statistically significant (p < 0.05, Kaplan-Meier survival analysis). Combining IHC and FISH reached the best predicting effect among the three methods (Fig. 3c). As shown in Fig. 3d and Table 4, patients with MYCN FISH⁻ but protein-expressing tumors always had a poor prognosis. In contrast, patients with MYCN FISH⁺ but low protein expression tumors always had a favorable prognosis (Fig. 3d, Table 4). In summary, our results suggest that IHC could make up for FISH in predicting prognosis.

FISH focused on DNA copy numbers, while IHC centered on protein expression. We wondered what exactly affected MYCN protein expression. Western blotting was performed to test MYCN protein expression (Fig. 4a). As predicted, their protein level varied from DNA copy numbers. For instance, the #3 tissue detected no protein expression but high DNA copy numbers. Based on RNA-seq data, we found that their MYCN mRNA expression was more consistent with the FISH results (Fig. 4b). These data demonstrated that aberrant protein expression might not stem from DNA copy numbers but be related to RNA translation or protein stability. Previous studies have shown that FBW7 [23], RUNX3 [24], and MCPIP1 [25] play a fundamental role in MYCN ubiquitination and degradation, and all these genes showed declining trends in the high MYCN protein expression group (Fig. 4c) (ns: no significant, unpaired T-test). PLK-1 [26], CDK-1 [27], PIK3CA [28], ATF4 [23], TEAD4 [29] and ALYREF [30] could enhance MYCN protein stability and sustain MYCN expression in neuroblastoma. The expression of these genes showed a rising trend in the high MYCN protein expression group (Fig. 4d) (* p < 0.05, ns: no significant, unpaired T-test). More factors that could affect MYCN protein expression were analyzed. The results showed that their expression was more consistent with the MYCN protein level detected by IHC (Fig. S2). Overall, some genes could influence MYCN protein expression through post-transcriptional or post-translational modification. RNA-seq revealed that their expression changed among different samples, which might be the reason why the results of MYCN IHC could be different from those of FISH. Conjoint analysis of IHC and FISH could test MYCN protein stability, a key prognostic factor.