High CD44 expression is not a prognosis marker in patients with high-risk neuroblastoma.

We have read with interest the article by Vega et al. about the impact of CD44 expression on survival of neuroblastoma patients [1]. We had major concerns about the survival analyses showed in this manuscript.

By doing the analyses using the same public dataset (GSE45547) [2] and the R2 analysis platform (http://r2.amc.nl), we uncovered that: (1) the survival curves presented in this paper have been obtained taking into account all neuroblastoma stages, and not only stage 4 patients; (2) for panel d (i), the correct raw p-value is 0.048 and not 0.018; (3) the authors did not used the true top 10% cases with the highest CD44 expression nor a minimal group of 8 patients as they claim. For all stages and MYCN amplified cases, only one cut-off in the number of patients per group provides a significant raw p-value (R2 analysis) and this cut-off has likely been selected by the authors to draw their conclusions; (4) the indicated p-values are raw p-values but not Bonferroni corrected values as indicated. Corrected p-values are all not significant. Using the indicated criteria, we got no significant Bonferroni corrected p-values for the three categories of patients (Fig. 1a, b and c). Strikingly, for all stages, when considering the top 10% cases with the lowest CD44 expression a significant Bonferroni corrected p-value was observed, indicating that low CD44 expression may be associated with poor survival (Fig. 1d).

Fig. 1

(a, b, c and d) Survival curves for all patients with high or low CD44 expression in the indicated populations. In a and c the top 10% cases with the highest CD44 expression were considered, in all stages and in all stages without MYCN amplification, respectively. In b, the 8 cases with the highest CD44 expression were considered among all stages with MYCN amplification. In d, the top 10% cases with the lowest CD44 expression were considered in all stages. (e, f and g) Survival curves for the subset of stage 4 patients with high or low CD44 expression in the indicated populations. In e and g, the top 10% cases with the highest CD44 expression were considered; In f, the 8 cases with the highest CD44 expression were considered among stage 4 patients with MYCN amplification.

Survival analysis performed specifically in stage 4 patients using the same criteria shows that CD44 expression is not of prognosis significance for overall survival (Fig. 1e, 1f and 1g).

In conclusion, our analyses clearly indicate that CD44 expression is not a biomarker of prognostic value for high-risk stage 4 neuroblastoma patients.

Declaration of Competing Interest

None