Chronic Myeloid Leukemia Following Treatment for Primary Neoplasms or Other Medical Conditions

A Report of 21 Cases and Review of the Literature

Lian-He Yang, MD, PhD; Pu Su, MD, PhD; Catherine Luedke, MD; Chuanyi Mark Lu, MD; Abner Louissaint Jr, MD, PhD; Chad M. McCall, MD, PhD; Sarah Rapisardo, PhD; Bethany Vallangeon, MD; Endi Wang, MD, PhD


Am J Clin Pathol. 2018;150(3):246-258. 

In This Article


Therapy for malignancies has been shown to increase the risk of developing secondary neoplasms. In particular, the risk for myeloid neoplasms is more than 10-fold higher after chemotherapy and/or radiotherapy, with an accumulated incidence ranging from 1% to 10% depending on types of treatment.[61,63,95–97] While the vast majority of therapy-related neoplasms are MDS or AML,[65] other types of leukemia have been reported, including therapy-related CML in occasional cases.[4–64,66–69] The relative risk for CML in patients who have been exposed to chemotherapy and/or radiotherapy varies with different studies, ranging from 0 to 4 in the literature.[61,63] Among all therapy-related leukemia, therapy-related CML constitutes a relatively small fraction, but the rate tends to increase with certain types of primary neoplasms. According to Aguiar's analysis,[61] CML associated with prior chemotherapy and/or radiotherapy constituted 13.3% of the overall cases of therapy-related leukemia, with breast or ovarian cancer as the most frequent primary malignancies. On the other hand, Kaldor et al[98] reported two cases of CML from a total of 163 leukemia patients who had been treated for Hodgkin lymphoma, resulting in a rate (1.2%) much lower than that reported by Aguiar.[61] A recent study by a Japanese group reported 11 cases of therapy-related CML among 308 patients with CML, a rate of 3.6%.[64] In their meta-analysis, Waller et al[63]reported that the predominant primary malignancy in patients with therapy-related CML was uterine and cervical cancer, followed by ovarian cancer and Hodgkin lymphoma. In our series, 10 patients had received treatment either for lymphoma or breast carcinoma, comprising 52.6% of our patients (19) treated for primary malignancies. This finding was in keeping with the previous studies. Our analysis of the reported cases demonstrates similar findings, particularly a high percentage of lymphoid neoplasms, non-CML leukemia, and breast carcinoma as primary malignancies before the development of CML (Table 2 and Table 3). In contrast, prostatic carcinoma and lung carcinoma, the two neoplasms with high incidence, have been infrequently reported as primary malignancies preceding CML. This tendency toward certain types of primary malignancies for developing secondary CML, as seen in therapy-related MDS/AML,[95,97] is likely explained by application of specific types of treatment modalities rather than prevalence of the primary neoplasm.[61,63] Alternatively, the disproportionate distribution of primary neoplasms may be due to the curability and longer survival of certain neoplasms over others.[95,97] Age at the diagnosis of CML varies with different studies, with several reports showing data similar to that in the de novo counterpart.[61,63,99] This is particularly noted in our analysis of the reported cases (Table 3), the majority of which were published more than 15 years ago. On the other hand, more recent studies demonstrate an older age (about 10 years older in comparison with nonstratified CML) at diagnosis in therapy-related CML,[64,100,101] as seen in our series. The older age in this particular patient population likely reflects improved treatments for the primary neoplasm/condition and better survival, resulting in a selection bias for the patients' age in therapy-related CML.

If treatment for the primary neoplasm/condition promotes the development of CML, can a specific regimen be blamed for the cause? Modern therapy for malignancies advocates a personalized approach, and the treatment protocols are often individualized with diversified regimens throughout the disease course. Therefore, it is difficult to identify specific regimens or agents responsible for the emerging Philadelphia-positive clone and eventual development of CML. In our series, nine of 18 (50%) patients with primary malignancy received both chemotherapy and radiotherapy, while five received chemotherapy alone and four received radiotherapy alone. These percentages were similar in the reported cases of therapy-related CML according to our literature review and analysis (Table 2 and Table 3). Ionizing radiation has been implicated in the development of leukemia, including CML, in patients who survived the atomic bomb in Japan. Radiotherapy has been demonstrated to have a three-fold increase in the morbidity of leukemia, with 18% being CML.[63] The incidence of leukemia seems to correlate with dose and duration of radiotherapy, and leukemogenesis has been explained by radiation-induced chromosomal breakage. However, the latency from treatment to the diagnosis of CML seems to vary with the types of primary neoplasm, suggesting that the incidence may be determined by the actual dose of radiation delivered to the bone marrow.[63]

It has been well documented that chemotherapeutic regimens such as alkylating agents and topoisomerase II inhibitors induce chromosomal aberrations causing leukemia. While the risk for AML is increased over 10-fold for patients exposed to these chemotherapeutic regimens,[63,95] these patients demonstrate minimal increased risk for CML (approximately two-fold or less).[61,63] Whether or not chemotherapy in combination with radiotherapy increases the risk of leukemia is controversial, with the majority of studies demonstrating an increased relative risk for leukemia when compared with either chemotherapy or radiotherapy alone.[63] In therapy-related MDS/AML, the cases associated with alkylating agents and/or ionizing radiation usually occur 1 to 5 years after exposure, and those related to topoisomerase II inhibitors tend to have a longer latency, occurring 5 to 10 years after exposure.[65] In contrast, therapy-related CML shows a latency (median interval of about 5 years; Table 3) between that of AML/MDS induced by alkylating agents and/or radiotherapy and that associated with topoisomerase II inhibitors,[61,63] suggesting differing pathogenesis or overlapping features of leukemogenesis.

In addition, patients who receive chemotherapy or radiotherapy could have their immunity impaired or bone marrow microenvironment injured. This may lead to a predisposition for secondary neoplasms via diminished immune surveillance or damaged "soil" in favor of a mutant hematopoietic clone. Two (10%) of our patients received organ transplantation before development of CML, with a latency of 16 and 36 months, respectively. An increased risk for leukemia has been reported in organ transplant recipients, including posttransplant CML.[102] Although the absolute occurrence is low, the proportion of posttransplant CML to total posttransplant leukemias seems higher than the proportion of de novo CML to total de novo leukemias. While it is possible that some transplant medications may play a role as genomic mutagens, organ-specific rate of posttransplant leukemia suggests that the degree of iatrogenic immunosuppression directly correlates with the incidence of posttransplant leukemia.[103] In theory, diminished immune surveillance may increase the chance for a preleukemia clonal hematopoietic stem cell to escape checkpoint restriction and gain a growth advantage after receiving additional genetic or epigenetic hits. This hypothesis is in keeping with the recent finding of preexisting clonal hematopoiesis of indeterminate potential in blood samples of patients who received autologous stem cell transplants and subsequently developed therapy-related AML/MDS.[96,104] The latter finding provides some evidence that certain groups of patients may be more susceptible to leukemia than others via therapy-induced enrichment of a mutant clone.[97] Alternatively, a hereditary predisposition due to polymorphisms in genes that affect drug metabolism or DNA repair mechanisms might explain the difference in susceptibility for therapy-induced leukemia.[105]

One (5%) of the patients in our series developed CML after simple excision of the primary tumor without exposure to either chemotherapy or radiotherapy. Similar cases have been reported in the literature. In addition, CML has been seen concurrent with or subsequent to the diagnosis of chronic lymphocytic leukemia (CLL) and other indolent B-cell neoplasms without pertinent treatment.[74] Diminished immunity has been observed in CLL and other indolent B-cell neoplasms and is hypothesized as the cause for the increased risk for CML in this patient population. Alternatively, this tendency to the development of multiple neoplasms in individual patients may be explained by intrinsic predisposition due to certain genomic defects that remain to be characterized in the future. Nonetheless, CML associated with untreated primary neoplasms constitutes a small fraction of secondary CML. According to a study by Specchia et al,[74] over 95% of patients with secondary CML were reported to have exposure to chemotherapy and/or radiotherapy.

In therapy-related AML/MDS, distinct clinical presentations, pathologic features, and cytogenetic profiles have been observed, compared with their de novo counterparts.[65] In contrast, therapy-related CML seems to have features indistinguishable from de novo CML. We would anticipate more cytogenetic aberrations in therapy-related CML in addition to the Philadelphia chromosome if its mechanism were the same as that of therapy-related AML/MDS. However, 88.9% of therapy-related CML in our series (Table 1) and 83.5% of the reported cases harbor an isolated Philadelphia chromosome, and fewer than 20% have additional genomic aberrations (Table 3). This cytogenetic profile is similar to that observed in de novo CML.[106] In addition, cytogenetic changes associated with chemotherapeutic regimens and/or ionizing irradiation, such as −7, −7q, −5, −5q, MLL rearrangement, and complex abnormalities, are either not identified or not reported in a significant number of the cases (Table 3). These findings seem to support an alternative pathogenesis for the development of therapy-related CML, one different from that for therapy-related AML/MDS. While the exact pathway remains unclear, the recent development of high throughput sequencing analysis may provide a tool for the genomic analysis of therapy-related CML and allow for a comparative study with therapy-related AML/MDS and de novo CML, hopefully shedding more light on its genomic profile and pathogenesis.[96,97,104]

In contrast to therapy-related AML/MDS,[65] therapy-related CML demonstrates clinical outcomes similar to that of de novo CML,[64] in line with its pathologic features and cytogenetic profile.[61,63] The patients in our series seemed to respond to imatinib or its derivatives, as in the recently reported cases.[64] Disease-specific survival was 100% in our series, though the follow-up was relatively short for a significant number of our patients. Analysis of the reported cases demonstrates considerably shorter survival than what was observed in our contemporary CML study (Table 3 and Figure 1).[107] When the cases were stratified by tyrosine kinase inhibitor (TKI) or non-TKI therapy, a significantly longer survival was seen in those cases with TKI treatment (Figure 1), suggesting an effect of TKI therapy on survival. A shorter survival in those cases with TKI therapy in comparison with our series and other contemporary studies may be explained by a lag in reporting, delayed use of TKIs in patients diagnosed before 2001, or possibly inclusion of deaths due to causes other than CML. In conclusion, our analysis suggests a good response to contemporary treatment modalities in therapy-related CML patients with a response rate equivalent to that observed in de novo CML, while primary diseases pose a major threat to patients with a prior history of malignancy or other critical conditions.

Figure 1.

Kaplan-Meier survival analysis. Yellow line: cause-specific survival analysis for the cases in our series. Two patients who died of primary neoplasms rather than chronic myeloid leukemia were censored at the time of death. Blue line: survival analysis for the reported cases treated with tyrosine kinase inhibitors (TKI). Green line: survival analysis for the reported cases treated with conventional non-TKI regimens.