Risk factors for MDS and acute leukemia following total therapy 2 and 3 for multiple myeloma

Lenalidomide has been associated with an increased risk of secondary malignancies, including myelodysplastic syndrome (MDS) and acute leukemia (AL), when applied as maintenance after high-dose or concomitantly with standard-dose melphlan for newly diagnosed multiple myeloma (MM).^1-4  We previously identified risk factors for MDS-associated metaphase cytogenetic abnormalities (MDS-CA).⁵  The composition of recent total therapy (TT) trials (applying thalidomide in the experimental arm of TT2,⁶  thalidomide plus bortezomib in TT3a,⁷  and subsequently lenalidomide plus bortezomib in TT3b⁸ ) offered the unique opportunity to investigate the development of MDS-CA and clinical MDS in relationship to the use of thalidomide and lenalidomide.

Details of these trials and patient outcomes have been reported previously.^6-8  Briefly, the experimental arm of TT2 applied thalidomide with induction, consolidation, and maintenance until relapse. TT3a and TT3b used bortezomib and thalidomide during induction and consolidation but differed in maintenance. TT3b used lenalidomide rather than thalidomide. Median follow-up was 118 months in TT2 (n = 668), 78 months in TT3a (n = 303), and 48 months in TT3b (n = 177). All patients signed an informed consent in keeping with institutional, federal, and Helsinki Declaration guidelines. Protocols were approved by the University of Arkansas for Medical Sciences institutional review board.

Bone marrow examinations were performed at baseline and for restaging before each protocol phase, approximately every 3 months in year 1 of maintenance, semiannually in years 2 and 3 of maintenance, and annually thereafter. Metaphase karyotyping was an integral part of each bone marrow examination for the detection of cytogenetic abnormalities typical of MM (MM-CA) or MDS (MDS-CA).⁵  All reported MDS-CA were reviewed and confirmed by 1 of the authors (J.S., B.B., S.Z.U.) The term “persistent MDS-CA” referred to the presence of such abnormalities on 3 successive occasions. In total 123 patients developed MDS-CA, including 41 who displayed the persistent type; del20 was the most common abnormality (supplemental Table 1). Because we examined bone marrow aspirates and biopsies for MDS stigmata in the context of cytopenia, it became obvious that such alterations were often transient and related to MM therapy. We therefore defined “clinical MDS/AL” as a condition requiring MDS- or AL-directed therapies; these applied to 36 patients and 1 patient with overt MDS who only received palliative supportive care. No differences in MDS-CA subtype distribution were noted when overall MDS-CA, persistent MDS-CA, and MDS-CA encountered in clinical disease were compared. The regimens used to treat clinical MDS/AL are summarized in supplemental Table 2. In short, 16 patients were managed without hematopoeitic progenitor cell support, 18 received autograft-supported high-dose therapy, and 2 received an allotransplant including 1 with a preceding autotransplant.

Among 1148 patients enrolled, 1088 had baseline and follow-up cytogenetic data; 8 patients who exhibited MDS-CA at baseline were excluded. For the remaining 1080 patients, we report on the risk factors linked to MDS-CA (both overall and persistent) and clinical MDS/AL. Three landmarks (baseline, first transplant, and start of maintenance) were considered in order to distinguish the contributions of different protocol phases to the MDS problem. Details of protocol compliance, such as completion of transplant, consolidation, and maintenance components, including cumulative dosing of maintenance components as previously described, were also examined.⁹  The analysis is based on a cutoff date of May 4, 2012. Univariate and multivariate Cox proportional hazard regression models were used to identify associations of baseline and treatment-emergent covariates with onset of MDS-CA or clinical MDS/AL.¹⁰  Cumulative incidence plots were compared using the log-rank test.¹¹ 

MDS-CA was diagnosed in 11% of patients, including 4% with persistent MDS-CA; clinical disease developed in 3% of patients. Cumulative incidences of MDS-CA and clinical MDS/AL are shown in Figure 1, from initiation of protocol therapy (Figure 1A), following first transplant (Figure 1B), and from start of maintenance therapy (Figure 1C). Regardless of the landmark considered, MDS-CA (including the persistent subtype) increased with the transition from TT2 control to the thalidomide arm to TT3a and TT3b. TT3a and TT3b, in general, had greater incidence of clinical MDS than both arms of TT2 (not always at P < .05 level). MDS-CA preceded clinical disease by at least 3 months in two-thirds of patients and by 6 months in approximately one-half of patients (supplemental Figure 1A-B, which also displays time of MM relapse and death). The median survival of patients from onset of MDS-CA and clinical MDS/AL was 3.7 and 1.1 years, respectively (supplemental Figure 2). Transplant management of clinical disease was not superior to chemotherapy, probably because of small sample size. The development of both MDS-CA and clinical disease adversely affected overall survival and progressive free survival from initiation of protocol therapy; other risk factors included gene expression profiling (GEP)–defined high-risk presence of MM-CA, high lactate dehydrogenase and creatinine levels, and immunoglobulin A isotype (supplemental Table 3).

Next we investigated baseline and treatment-emergent risk features linked to MDS-CA and clinical MDS/AL (Table 1). After adjusting for other significant variables, TT3b had strong links to the development of MDS-CA for all 3 landmarks examined. Univariately, clinical disease was linked to TT3a and TT3b (especially for baseline and transplant landmarks). Thus, risk factors for MDS-CA and clinical MDS/AL were similar with regard to treatment variables. When outcomes from baseline and first transplant intervention were considered, the absence of thalidomide in the control arm of TT2 was associated with a significant risk reduction of both MDS-CA and clinical disease. The observation of lower risk of MDS with the TT2 control arm and of higher risk with bothTT3a and TT3b suggests that both thalidomide and lenalidomide contribute to MDS. Regardless of landmark, additional independent contributors to MDS-CA included age ≥65 years, male gender, and elevated levels of β-2-microglobulin (B2M) >5.5 mg/L. The linkage of these variables to clinical MDS/AL was weak. A trend for higher cumulative doses of dexamethasone to protect against MDS-CA is unexplained. Lower CD34 quantities (<4 million/kg) applied with first transplants tended to raise the probability of MDS-CA, while lower CD34 dosing (<5 million/kg) was an independent contributor to MDS-CA and clinical disease (see Table 1, supplemental Figure 3). MM relapse was an independent risk factor for MDS-CA and univariately tended to increase risk of clinical MDS/AL (from baseline landmark; see Table 1).

Not strictly confirming results from 3 randomized phase III trials,^1-3,12  lenalidomide maintenance (in TT3b) increases the probability of MDS-CA with a trend toward contributing to clinical disease. The lower incidence of MDS-CA and clinical disease in the TT2 control arm suggests that thalidomide also enhances the risk of MDS. Frequent cytogenetic monitoring for MDS is therefore in order, especially since MDS-CA often precedes clinical disease by many months. Mechanistically, both thalidomide and especially lenalidomide enhance Dickkopf 1 expression,¹³  thereby inhibiting Wnt signaling, which is critical to preservation of hematopoietic self-renewal.¹⁴  The higher incidence of MDS-CA in males, translating into a strong trend for higher risk of clinical MDS/AL, is unexplained. At least 1 study reported on inferior survival of males with MDS.¹⁵  Dexamethasone, which is a well-known inhibitor of interleukin-6 and other cytokines,¹⁶  may exert some protection against MDS-CA via inhibition of joint growth factors for both MM and MDS. Although high levels of B2M have been associated with inferior prognosis in primary MDS,¹⁷  a linkage of the tumor burden marker B2M to MDS-CA (all 3 landmarks) with a trend effect for clinical disease (from maintenance) has not been reported. The notion of MM relapse as an independent contributor to MDS-CA supports the notion of MM’s contribution to the MDS problem encountered. It is conceivable that an altered bone marrow microenvironment contributes to both MM and MDS via commonly altered signaling pathways, which can be suppressed by dexamethasone. Such altered stroma may also be generated by cryptic MDS (ie, MDS-CA), favoring MM cell survival and progression and thereby explaining patients’ shorter survival time. The higher risk of MDS-CA and clinical disease in those who receive lower doses of CD34 has been previously reported⁵  and may be related to more stem cell divisions required to reestablish hematopoietic homeostasis. Such stem cell stress is associated with telomere shortening, which is linked to myeloid neoplasia.¹⁸ 

In summary, we identified treatment (both TT3, TT2 with thalidomide; lower CD34 dose with both transplants), MM (B2M, absence of CR), and host variables (age, gender) that are linked to MDS-CA and, to a lesser degree, to clinical disease development. The lack of consolidation therapy as a contributor to MDS may be confounded by its use in all 4 TT regimens. As most patients who developed clinical disease had MDS-CA previously, the discontinuation of immune modulators or institution of preemptive MDS therapy deserve formal investigation. The MDS protective effect of sufficient CD34 dosing at time of first autologous transplantation requires large collections upfront.¹⁹  One study of lymphoma identified such MDS risk by GEP of hematopoietic progenitor cell products,²⁰  which we are currently evaluating in our MM population as well. We have also performed GEP analyses of whole bone marrow biopsies¹²  procured prior to therapy to determine whether a MDS-prone state can be anticipated in an older patient population at risk of both MM and MDS.

Peggy Brenner, Office of Grants and Scientific Publications, University of Arkansas for Medical Sciences, edited the manuscript.

This work was supported by a grant from the National Institutes of Health, National Cancer Institute (CA 55813).

Contribution: S.Z.U., A.R., and B.B. designed the study and wrote the manuscript; S.Z.U., A.R., B.B., J.S., J.E., S.Y., R.S., A.H., J.C., D.J., N.C., C.B., and N.P. analyzed data; S.Z.U., B.B., F.v.R., and S.W. made patients available for the study; M.C.F., A.O.A., J.M., C.J.H. and Z.S. helped review the manuscript and provided feedback for improvement; and all authors reviewed the manuscript.

Conflict-of-interest disclosure: S.Z.U. is a consultant to Celgene, Millennium, and Onyx. He has received research funding from Onyx and Celgene and speaking honoraria from Celgene. B.B. has received research funding from Celgene and Novartis. He is a consultant to Celgene and Genzyme and has received speaking honoraria from Celgene and Millennium. B.B. is a coinventor on patents and patent applications related to use of GEP in cancer medicine. The remaining authors declare no competing financial interests.

Correspondence: Saad Usmani, Myeloma Institute for Research & Therapy, University of Arkansas for Medical Sciences, 4301 W. Markham St, Slot 816, Little Rock, AR 72205-7199; e-mail: susmani@uams.edu.