To assess the role of treatment-related risk factors for AML/MDS development in patients with MPNs, we took advantage of a large (N = 11,039) national MPN cohort.23
By identifying patients who experienced transformation to AML/MDS and their matched controls from this cohort, a nested case-control study was performed. The SIR of AML transformation was 24.7 to 63.8, depending on MPN subtype. This is probably a conservative estimate because reporting of transformation in certain patient categories, such as the elderly, occasionally may have been neglected. The increased transformation risks, calculated as ORs, associated with the highest exposure to HU, P32
, and alkylators were only 1.3, 4.6, and 3.4, respectively (Table 2). Importantly, 25% of patients with transformation to AML/MDS were never exposed to HU, P32
, or alkylating agents. In addition, only 32% of patients with transformed disease were exposed to cumulative doses of P32
and/or alkylating agents shown here to be leukemogenic. We therefore conclude that the risk of transformation is mainly associated with the disease itself, whereas exposure to cytoreductive agents is of less importance.
Our results support the hypothesis that nontreatment-related factors play a major role in AML/MDS development after MPNs. In a recent study based on 18 patients with MPNs who experienced transformation to leukemia, AML1/RUNX1
mutations were detected in five patients at transformation; one patient was never exposed to cytoreductive therapy.24
Furthermore, when AML1/RUNX1
mutants were transduced into stem cells of patients with MPNs in chronic phase, it resulted in proliferation of immature myeloid cells, enhanced self-renewal capacity, and proliferation of primitive progenitors.24
In another study involving 24,577 first-degree relatives of the 11,039 patients with MPNs described in the present study, we found a 1.8-fold (95% CI, 0.9 to 3.8; P
= .09) excess risk of AML development in relatives of patients with ET.23
Another main finding is the lack of a significant association between HU exposure and AML/MDS risk. The potential leukemogenic effect of HU has remained a controversial issue for many years.5,25,26
AML transformation has been reported in up to 22% of patients after therapy with HU alone16,27
and 30% of patients after therapy with HU in combination with busulphan.16
Other studies, including the large prospective ECLAP (European Collaboration on Low-dose Aspirin in Polycythemia Vera) study, have found no association between HU treatment and leukemic transformation in patients with ET and PV.12,25,28–31
In the randomized trial comparing HU and pipobroman (piperazine derivative that acts via alkylating mechanism) in patients with PV, leukemic transformation was significantly more common after pipobroman therapy compared with HU therapy, with cumulative incidences of 22% versus 56% at 20 years.27
Importantly, in this study, half of the patients who developed AML/MDS in the HU group received more than one cytoreductive drug, whereas in the pipobroman group, 85% of AML/MDS cases received pipobroman only.14,32
The discrepant results published regarding HU exposure and risk of AML/MDS are probably related to differences in patient characteristics (including MPN subtypes, MPN therapy, follow-up, and study size). On the basis of our findings, we cannot totally exclude a leukemogenic effect of HU. However, this potential risk is limited compared with the risk associated with the disease itself. Although sickle-cell anemia is not a myeloproliferative disorder, we believe the fact that AML/MDS development is a rare event even after many years of HU treatment in patients with sickle-cell anemia provides additional support for this notion.33,34
In vitro studies have also suggested that the mutagenic and carcinogenic potential of in vivo HU therapy is low.35
As expected, patients exposed to P32
or alkylating agents carried an increased risk of AML/MDS transformation, though only at cumulative doses greater than 1,000 MBq or 1 g, respectively. No dose dependency was observed in the lower dose intervals. Although based on small patient numbers, this may indicate the existence of a threshold exposure of P32
and alkylators for AML/MDS transformation in MPNs.36
However, patients may be at higher risk of transformation not because of treatment-related factors but rather because of longer disease span and more aggressive disease biology, causing exposure to higher doses and/or multiple drugs. Interestingly, most of the patients who developed AML/MDS after no treatment or HU treatment only did so within 5 years of MPN diagnosis. This may corroborate the notion that HU is nonleukomogenic because the majority of patients administered P32
and/or alkylators experienced transformation at a later time point.
SIRs for AML transformation among men and women did not differ when all MPN diagnoses were included in the analysis. We observed a lower SIR for women with ET, whereas the SIR for women with PV was slightly higher than that for men. None of these differences were statistically significant; thus we could not verify the previously reported higher incidence of transformation in women.1,25
In addition, we did not observe any association between leukocytosis and risk of AML transformation, which has been suggested by some investigators37
but not others.38
As reported previously, the risk of transformation was highest among patients with PMF.1,2
The median survival of patients after AML transformation was 3 months, confirming the dismal prognosis associated with this event.1,6
Our study has several strengths, including its large size and the application of high-quality data from Sweden in a stable population with access to standardized universal medical health care. The use of the nationwide register-based case-control design ruled out recall bias, ensured a population-based setting, and provided generalizability of our findings. Patient cases and controls were diagnosed with MPNs over a period of 47 years, with almost 60% diagnosed before 1990. A long observation time is clearly advantageous when studying diseases with long and indolent courses and late-appearing events of interest.
Although we used the largest population-based MPN database (N = 11,039) to date, we were limited by numbers for the nested case-control study. More specifically, when we matched patient cases and controls, we applied conservative matching criteria (MPN subtype, year of birth [± 5 years], sex, date of MPN diagnosis [± 1 year]). Consequently, 65 patient cases were excluded because to lack of matched controls. We obtained clinical data on 44 of these patient cases and found clinical and treatment features for the excluded patient cases to be similar to patients included in the study (Appendix Table A1, online only). In a sensitivity analysis in which we relaxed the matching criteria to include only duration of disease and MPN subtype, these 44 patients could be included in an analysis of a larger number of patients (206 patient cases and 366 controls). The results were similar to those obtained in the main analysis (Appendix Table A2, online only). Thus, we feel confident in the correctness and robustness of our results.
Only 2.6% of the 11,039 patients with MPNs in the cohort experienced transformation to AML/MDS. Given the long observation time, this proportion is lower than that previously reported,1–3,7–9,27
although AML transformation was observed in 1.4% and 2% of patients receiving busulphan and P32
, respectively, at a median follow-up time of 8 years in a randomized trial by the European Organisation for Research and Treatment of Cancer.36
We speculate that this may be the result of selection bias in prior clinical studies, under-reporting in our population-based study, or most probably a combination of both factors. Such a discrepancy would, however, not likely affect the results of this nested case-control study because of its design and the large number of patients included.
The diagnostic criteria for MPNs changed during the study period; many patients with ET in this study would today be diagnosed with PMF. In a Swedish follow-up study of 60 patients treated with anagrelide between 1998 and 2002, the diagnostic bone marrow samples were reevaluated blindly according to WHO criteria. Twenty-one of 42 patients with ET were identified as truly having ET.39
Thus, the AML/MDS SIRs reported here among patients with ET may be overestimations. However, this is not likely to affect the results regarding treatment-related risk.
In summary, we conclude that the inherent propensity of MPNs to AML/MDS transformation is substantial, as judged from the fact that a quarter of the patients who developed AML/MDS were never exposed to cytoreductive therapy. In addition, HU exposure, even at high doses, is not associated with a significantly increased risk of transformation to AML/MDS. These findings have important implications regarding treatment strategies in MPNs, especially in younger patients requiring decades of active treatment.