To the editor,

Myelodysplastic syndrome (MDS) is a clonal disorder of hematopoietic stem cell with maturation defect resulting in ineffective hematopoiesis. A recent mouse study showed that dysfunction of telomere drives MDS [1], which indicates that short telomeres might contribute pathogenesis of MDS. When telomeres become extremely short and sufficient loading of the shelterin complex is not possible, cells undergo senescence and induce cell cycle arrest [2].

Several studies have shown that patients with MDS have a shorter telomere length (TL) than healthy controls. However, these studies used conventional telomere restriction fragment (TRF) [3] and interphase quantitative fluorescence in situ hybridization (Q-FISH) [4] method, which provide average TL or relative TL. Several methods to measure the absolute length of short telomere such as STELA (Single telomere length analysis) [5], Universal-STELA [6], and TeSLA (Telomere shortest length assay) [7] have been developed. The most recent method to measure short TL, TeSLA was designed to detect TL up to 18 kb of whole chromosomes [7]. Furthermore, Telomeres below 1.6 kb cannot be visualized by other methods except TeSLA [7].

Here, we investigated TL profile including the proportion of cells with critically short TL in bone marrow (BM) nucleated cells from 52 MDS patients by modified TeSLA with the internally labeled biotin probe. To assess critically short TL as a potential prognostic factor, the association of short telomeres with prognostic risk and clinical outcomes was evaulated (Supplementary materials and methods).

In this study, a total of 52 patients were included. Further characteristics of patients are described in Supplementary Table 2. A median 177 telomere bands (range 75 – 262) were measured for each single patient’s sample. The average TL was 3.18 kb (range, 2.2 – 4.34), which was remarkably shorted than those of healthy populations (5 – 15 kb) previously reported [8, 9]. The shortest TL was 0.787 kb (Supplementary table 3). Our results clearly demonstrated the profiles of shortest telomeres in MDS considering the fact that TL below 1.6 kb was not visualized by other methods except TeSLA [7]. Further details of TL are described in supplementary Table 3. The average TL was not different among MDS (3.32 kb in MDS-SLD, 3.10 kb in MDS-RS-SLD, 3.30 kb in MDS-RS-MLD, 3.18 kb in MDS-MLD, 3.16 kb in MDS-EB1, 3.02 kb in MDS-EB2, 3.43 kb in MDS-U, p = 0.71). In contrast, the shortest TL and the ratio of shortest telomere below 1.0 kb (ShTL1.0) showed trends towards positive correlations to MDS subtypes. ShTL1.0 was increased in higher risk MDS group (2.39% in lower risk MDS group and 5.20% in higher risk MDS group, p = 0.0227). The scores of IPSS-R and ShTL1.0 was linearly correlated (spearman’s rho = 0.35 and p = 0.0196) (Fig. 1A). ShTL1.0 showed a positive correlation with blast percentage in BM (spearman’s rho = 0.32 and p = 0.0224) (Fig. 1B). Although ShTL1.0 s according to the presence of thrombocytopenia showed a trend toward a significant difference, we could not find differences of ShTL1.0 values according to the presence of cytopenia or dysplasia (Supplementary Fig. 3). We further investigated if coefficient of variation (CV) of TL was correlated with IPSS-R or BM blasts. The CV of the TL was not significantly correlated with either (Supplementary Fig. 4). MDS patients with a shortest TL ≥ 0.787 kb at the time of diagnosis showed better OS and a trend toward better PFS compared to patients with a shortest TL < 0.787 kb in univariate analyses (HR = 0.23 and 0.44, p = 0.011 and 0.064, respectively) (Fig. 2). Multivariate analyses showed that the shortest TL was an independent prognostic factor for PFS and OS both (Supplementary tables 4 and 5). Interestingly, when we analyzed the impact of age on OS according to the cut-off of 60 years, survival difference was consistent in a subgroup of elderly patients (≥ 60 years, p = 0.028), showed trends towards differences according to the shortest TL for PFS in young patients (< 60 years, p = 0.099) and OS in elderly patients (≥ 60 years, p = 0.084) (Supplementary Fig. 5).

Fig. 1
figure 1

A Correlation plot of the ratio of the shortest telomere length below 1.0 kb (%, y axis) and IPSS-R score (x axis). Spearman’s rho = 0.35 and p = 0.0196. B correlation plot of the ratio of the shortest telomere length below 1.0 kb (%, y axis) and BM blast % (x axis). Spearman’s rho = 0.32 and p = 0.0224

Full size image
Fig. 2
figure 2

A Progression-free survival (PFS) in MDS patients with shortest TL < 0.787 kb vs. with shortest TL ≥ 0.787 kb (median PFS = 16.3 months vs. not reached [NR], p = 0.0577) and B Overall survival in MDS patients with shortest TL < 0.787 kb vs. with shortest TL ≥ 0.787 kb (median OS = 22.43 months, vs. NR, p = 0.0056)

Full size image

Our results suggest that cells with critically short telomeres expand during progression of MDS. The proportion of cells with critically short TL in MDS was a robust indicator to predict poor OS. Our study has a few limitations from its retrospective nature and a small number of patients enrolled into this study. Further, analyses of germline mutations of TERT [10] was not performed in this study. However, our findings on the importance of the presence and abundance of extremely short telomere in the prognosis of MDS can be incorporated into the current prognostic scoring system of the IPSS-R. A large prospective study is needed to establish a new prognostic scoring system in which TL parameters are incorporated as prognostic factors.