Introduction

Myelodysplastic syndromes (MDS), a heterogeneous group of clonal haematopoietic stem cell disorders, are among the most frequent myeloid neoplasias in elderly patients. Patients with MDS present with uncharacteristic symptoms like fatigue, anaemia, infections and bleeding caused by peripheral cytopenia. The morphological classification is based on the percentage of blasts in the bone marrow, the type and degree of dysplasia and the presence of ring sideroblasts [1]. The incidence is five per 100,000 persons per year in the general population, but increases to 20 to 50 per 100,000 persons per year after 60 years of age [2].

Chromosomal abnormalities provide important prognostic information. Patients with isolated del(5q) MDS often present with macrocytic anaemia, a normal or elevated platelet count, spheronuclear and hypolobulated megakaryocytes, a medullary blast count <5% and female predominance [1, 3–5], sometimes referred to as 5q-syndrome, are believed to have a favourable prognosis. Life expectancy of older patients without transfusion requirement does not seem to be different from the general population [6]. The risk of progression to acute myeloid leukaemia is estimated to be low, at least for patients who do not have additional risk factors such as an elevated blast count, transfusion dependence, or additional chromosome aberrations. However, data from prospective studies on the risk of leukaemic transformation in patients with MDS and deletion 5q are lacking.

Lenalidomide, a thalidomide analogue with immunomodulatory and anti-angiogenic properties, induces transfusion independence and cytogenetic response in a high proportion of patients with MDS and 5q deletion [7–10]. Based on the results of an international phase II study (MDS-003) [7], lenalidomide has been approved by the US Food and Drug Administration in 2005. So far, there is limited information on the long-term outcome of MDS patients treated with lenalidomide. We report on detailed follow-up data of 42 patients treated with lenalidomide.

Design and methods

Patients

Our cohort included all patients treated with lenalidomide at the only European centre of the MDS-003 (Lenalidomide-MDS-003) study (42 patients St. Johannes Hospital, Duisburg, Germany). All patients with transfusion-dependent anaemia due to low- or intermediate-1-risk MDS (according to the International Prognostic Scoring System (IPSS)) associated with a 5q deletion with or without additional cytogenetic abnormalities were eligible. Thirty-six of the patients were enrolled in the MDS-003 study between November 2003 and May 2006 [7]. In the period between the closure of the MDS-003 study and the start of a subsequent study, six additional patients with MDS were treated with lenalidomide (starting dose, 10 mg daily) in accordance with the study protocol of the MDS-003 study [7]. Written informed consent was provided according to the Declaration of Helsinki. Lenalidomide therapy is still ongoing in responding patients. Dose adjustments were performed in the majority of patients due to side effects such as neutropenia. Patient characteristics are summarised in Table 1 and Electronic supplementary material 1 [7].

Table 1 Comparison of clinical and haematologic characteristics of patients from List et al. (148 patients) and our cohort (42 patients)
Full size table

Median follow-up of the patients after entering the study was 3.3 years, ranging from 2.3 to 5 years. Time between diagnosis and study entry was 0 to 15 years, median 2.1 years. Before study entry, 20 patients were treated with various agents, e.g. erythropoietin, all-trans-retinoic acid or anti-thymocyte globulin (Electronic supplementary material 2).

Standardised cytogenetic investigations

Cytogenetic investigations were performed centrally according to standard procedures as described [11]. Whenever possible, 25 metaphases were analysed. Description of chromosome aberrations and clone definition followed the recommendations of the International System for Cytogenetic Nomenclature [12]. A structural chromosome aberration or a gain of a whole chromosome found in at least two metaphases, or a loss of a whole chromosome found in at least three metaphases, were defined as clonal aberrations. Single cell aberrations were also considered clonal if they were confirmed by fluorescence in situ hybridization (FISH) or if they were already known from previous investigations. FISH for deletion 5q was included in each investigation. Depending on the aberrations detected during karyotyping, further probes were applied.

A complex karyotype was defined as ≥3 independent aberrations within one clone. Clonal evolution was defined as the acquisition of additional aberrations in clones with a 5q deletion. Independent clones did not contain a 5q deletion. All newly detected clonal aberrations were investigated retrospectively by interphase FISH, in order to clarify whether they were already present at earlier time points.

Cytogenetic and erythroid response

Complete cytogenetic response (CCyR) was defined as disappearance of the 5q deletion or any other chromosome aberration. Partial cytogenetic response (PCyR) was defined as a reduction of aberrant cells of more than 50% compared to the previous cytogenetic investigation [7]. Patients with persistent clones or with a reduction of less than 50% were defined as having no CyR. Furthermore, we distinguished between continuous CyR and transient CyR. Patients with transient CyR initially showed either CCyR or PCyR, but lost their cytogenetic response during follow-up (Electronic supplementary material 3). Accordingly, patients with transient erythroid response reached transfusion independence, but lost it during follow-up.

Morphologic evaluation

Morphologic evaluation was performed centrally at St. Johannes Hospital Duisburg and Hannover Medical School. Patients were diagnosed according to FAB and WHO classification [13, 14]. There was a high concordance of diagnosis between both institutions. Two cases were jointly reviewed and a final diagnosis agreed upon.

Statistical analyses

Time to AML was defined as time from study entry to the date of diagnosis of AML or to the date of last follow-up. Cumulative incidence functions of progression to AML were constructed using the method of Kalbfleisch and Prentice and compared with Gray’s test [15–17]. The Cox proportional hazards model was used for multivariate analysis. Differences in the distribution of individual parameters among patient subsets were analysed using Fisher’s exact or χ2 test for categorised variables and the Mann–Whitney U test for continuous variables. Statistical analyses were conducted using SAS version 9.1.

Results

Long-term follow-up analyses (median observation time: 3.3 years; range: 2.3–5 years) were performed in 42 patients with transfusion-dependent MDS and 5q deletion treated with lenalidomide. The median age was 64 years (range 32–84 years), and there was a female predominance (25 women, 17 men). This cohort had more patients with advanced MDS according to FAB than the whole patient set in the previously reported MDS-003 study (Table 1), which might account for the lower cytogenetic response rate for this population. Twelve of the 42 patients (29%) reached continuous (median 52 months, range 8–60 months), and another 12 patients (29%) reached transient transfusion independence. Twenty patients (48%) achieved a cytogenetic response, but 12 patients lost this during follow-up. Notably, in 12 analyses, the 5q- clone was detected only by karyotyping and not by FISH. Eight patients are in continuous cytogenetic remission (Table 2).

Table 2 Patients characteristics at study entry and progression
Full size table

Four patients progressed from low-risk MDS to higher risk MDS (refractory anaemia with excess of blasts according to FAB-classification [13]) within a median of 12.5 months after initiation of treatment with lenalidomide (range 12–16 months; Table 2). Among the 14 patients with low-risk MDS and an isolated deletion in 5q without severe bone marrow hypoplasia, three patients (patients 26, 29, and 38) underwent leukaemic transformation after having been treated with lenalidomide for a median of 10.5 months (range 9–40 months; Table 2). In total, 15 of the 42 patients (36%) progressed to acute myeloid leukaemia, while undergoing treatment with lenalidomide for a median of 12 months (range 5–50 months). Eight of them had not received any pre-treatment (Electronic supplementary material 2). Time from diagnosis to development of AML was 51 months (median, range 15–90 months). Except for one patient, all patients who progressed to AML died within a few months (Table 2). After start of treatment, the cumulative incidence of AML was 29% after 3 years and 42% after 5 years (Fig. 1).

Fig. 1
figure 1

Comparison of cumulative incidence of acute myeloid leukaemia after first diagnosis of myelodysplastic syndrome (a) and after study entry and start of lenalidomide therapy (b)

Full size image

We analysed the influence of gender, age, blast count, MDS subtype according to the WHO classification, IPSS risk score and additional chromosome aberrations to identify predictive markers for progression to AML (Fig. 2). There were no significant differences in the cumulative incidences of AML 3 and 5 years after treatment with lenalidomide with regard to gender, age, blast count, WHO subtype, IPSS risk score or additional chromosome aberrations at study entry. However, because of small numbers in some of those subgroups, the statistical analysis may not reveal a relationship.

Fig. 2
figure 2

Depicted are the cumulative incidences of acute myeloid leukaemia since study entry with lenalidomide therapy regarding gender (a), age (b), amount of blasts (c), WHO diagnosis (d), International Prognostic Scoring System score (e) and additional cytogenetic aberrations besides deletion 5q (f). Lack of markers at study entry allowed no prediction of an increased risk of leukaemic transformation

Full size image

When analysing erythroid and cytogenetic response, significant differences in the risk of progression to AML became evident. Patients with continuous or transient erythroid response carried a significantly lower risk of progression to AML than patients without an erythroid response (p = 0.001; Fig. 3a, b). Three out of 20 patients with cytogenetic response and 12 of 22 patients without cytogenetic response underwent progression to AML (p = 0.009). Notably, none of the eight patients (19%) who achieved a continuous cytogenetic response underwent leukaemic transformation. Three and 5 years after study entry, the cumulative incidence of AML for patients with a cytogenetic response was 10% and 21%, respectively, and for patients without cytogenetic response, it was 46% and 60%. Patients with a transient CyR had a risk of progression of 17% after 3 years of lenalidomide therapy. The risk rose to 51% after 5 years once the cytogenetic response was lost. Thus, there was a significantly different risk of progression to AML for patients with continuous, transient and lack of cytogenetic response after 3 and 5 years of lenalidomide treatment (p = 0.018; Fig. 3c, d). Three patients with transient cytogenetic response transformed into AML; loss of cytogenetic response and transformation into AML occurred almost simultaneously. In a Cox regression analysis considering transfusion independence and cytogenetic response as covariables, the risk ratio was 0.24 (95% confidence interval 0.073–0.798, p = 0.02) and 0.334 (0.09–1.238, p = 0.1011), respectively. Thus, the transfusion independence had a stronger impact on the risk of leukaemic transformation.

Fig. 3
figure 3

Depicted are the cumulative incidences of acute myeloid leukaemia regarding erythroid response (a, b) and cytogenetic response (c, d) during follow-up. Both factors are significant predictors for leukaemic transformation

Full size image

Most patients (13 of 15, 87%) with progression to AML acquired chromosome aberrations in addition to 5q deletion during follow-up, and nine of those (60%) evolved to a complex karyotype (Fig. 4; Electronic supplementary material 4). The most frequent additional aberrations were deletion 17p, trisomy 21, trisomy 8 and isodicentric chromosome X. Two patients, who developed a complex karyotype, underwent early stem cell transplantation and did not progress to AML. Moreover, two patients without clonal evolution and without cytogenetic response progressed to AML at 5 and 6 months after study entry, respectively. For 13 patients with clonal evolution, time to progression to AML was 12 months (median, range 6 to 50 months), after treatment with lenalidomide had been initiated. Apart from two patients, none of the patients with clonal evolution or with leukaemic transformation had reached continuous transfusion independence.

Fig. 4
figure 4

Multicolour fluorescence in situ hybridization (mFISH) of patient 39 who developed an acute myeloid leukaemia 12 months after study entry and treatment with lenalidomide. Besides the deletion 5q, in the sense of clonal evolution, further aberrations evolved resulting in a complex karyotype

Full size image

To clarify the question whether small clones with complex karyotypes were already present at study entry, FISH probes to detect additional chromosome aberrations such as trisomy 8, trisomy 21 or loss of 17p were applied. In no case, however, did FISH identify a small clone with the respective chromosome aberrations at an earlier investigation or at study entry, although aberrations of the complex clones, e.g. loss of the tumour suppressor gene TP53 due to loss of 17p, were confirmed at the time of clonal evolution.

Discussion

Lenalidomide, an amino-substituted thalidomide analogue with immunomodulatory and anti-angiogenic properties, leads to transfusion independence in over two thirds of patients with MDS and 5q deletion [7]. This is in agreement with the results of this study, in which about 60% of the patients achieved an erythroid response. Clearly, this leads to a major improvement in the quality of life of MDS patients with transfusion-dependent anaemia.

An important finding of this study is that none of the patients with a continuous erythroid and cytogenetic response transformed to acute myeloid leukaemia. In contrast, patients who did not achieve a cytogenetic response or lost this response during the course of the disease had an increased risk of leukaemic transformation. During progression to AML, most of the clones with deletion 5q acquired additional chromosome aberrations and developed into complex clones. Thus, genetic instability and clonal evolution seem to be the driving forces of leukaemic transformation in MDS patients treated with lenalidomide. It is known that complex clones are associated with a very poor prognosis both in MDS and AML [18, 19]. In this study, most patients who had developed complex clones also died within a few months.

It is difficult to evaluate whether patients treated with lenalidomide have an increased risk of leukaemic transformation. So far, no prospective clinical data from fully randomised trials are available, and there are only limited retrospective data. According to the IPSS [20] and newer risk classification systems [6], MDS with an isolated 5q deletion is associated with a good prognosis, although transfusion dependence is an independent unfavourable prognostic factor. In a retrospective study from the pre-lenalidomide era [21], the leukaemic transformation rate was 8% for patients with less than 5% blasts, but rose dramatically if more than 5% blasts were present. In contrast, the risk of progression to AML after lenalidomide treatment did not differ for patients with more or less than 5% blasts. Also, other potential risk factors like gender, age, blast count, MDS subtype according to WHO classification, IPSS score and cytogenetic aberrations in addition to 5q deletion did not help to predict leukaemic transformation. Although this might partly be due to low patient numbers in the patient population analysed, the negative effect of these risk factors could also be blunted by treatment with lenalidomide.

The only significant risk factors predicting progression to AML were the lack of continuous erythroid and continuous cytogenetic response. Remarkably, none of the patients with continuous cytogenetic response developed AML. Furthermore, patients with transient erythroid and transient cytogenetic response had a significantly lower risk of transformation after 3 and 5 years of lenalidomide treatment than patients without erythroid or cytogenetic response (p = 0.001 and p = 0.009). This finding is in line with a previous report on 168 patients participating in four clinical trials, in which the estimated 10-year risk for leukaemia progression was 15% for cytogenetic responders and 67% for cytogenetic non-responders or non-evaluable patients [22]. However, as shown in this and previous studies [7, 23], patients with erythroid or cytogenetic response can also develop complex clones and progress to AML. Since the majority of patients with erythroid or cytogenetic response lost the response during follow-up, careful clinical, morphological and cytogenetic follow-up is needed. Cytogenetic investigation should always include full karyotype analyses, since in many patients loss of cytogenetic response was indicated by only one metaphase with 5q deletion, whereas FISH analysis provided a negative result. Thus, FISH using a probe for the commonly deleted region in 5q [24] does not seem to be sufficient for follow-up analysis. Our results underline the importance of a close and careful cytogenetic follow-up even after reaching a CCyR, since the clone with 5q deletion may reappear, increase in size and acquire additional chromosome aberrations, indicating progression of the disease.

The long arm of chromosome 5 is the most frequently deleted region in MDS, being present in 20% to 30% of the patients [19]. Deletions of 5q are also the most frequent aberrations in complex karyotypes in secondary MDS/AML after chemo- and radiotherapy. The complex karyotypes in this series contained characteristic aberrations like deletions of 7q, deletions of 17p or trisomy 21. Inactivation of p53, the guardian of the genome with important functions in DNA repair and apoptosis induction, may be a critical early event during clonal evolution of 5q- clones, triggering genetic instability and the acquisition of secondary chromosome aberrations [25]. Two complex clones in this series contained a deletion of 17p with loss of the TP53 gene.

Importantly, the time between MDS diagnosis and treatment with lenalidomide in the patient cohort presented in this manuscript was 2.1 years, ranging from 0–15 years. Also, most of the patients had previous treatments with other agents that might have led to selection of genetic subclones. Unfortunately, FISH analyses did not seem to be the optimum method to identify these small pre-existing clones at diagnosis, probably due to a low detection limit of 2% to 11%, depending on the probe applied [26]. There are some limitations to the study presented here. The patient numbers are too small for robust subgroup analyses, so that some prognostic factors like the IPSS risk at lenalidomide treatment start might not have become statistically significant. Also, pre-lenalidomide therapies and the long median time between MDS diagnosis and lenalidomide exposure might have impacted on the disease course.

It will be of utmost importance to identify genetic lesions inducing resistance to lenalidomide and possibly also predisposing to leukaemic transformation. Gene expression profiling studies have identified genes like SPARC (osteonectin) that are up-regulated during lenalidomide treatment or identified an erythroid differentiation signature predicting response to lenalidomide [27, 28]. Thus, in the near future, new tools may be available to identify patients who have an increased risk not to respond to lenalidomide up-front before treatment and who may be at increased risk for transformation to AML. Until these new tools are available, the results of this study warrant regular and careful follow-up investigation of all patients treated with lenalidomide.