Chronic myeloid leukemia and risk of second malignancy in two eras of treatment

Tyrosine kinase inhibitor (TKI) therapy has dramatically altered the natural history of chronic myeloid leukemia (CML), and newly diagnosed patients who meet treatment milestones now have a life expectancy nearing that of the general population [1]. As a result, patients and providers must devote a similar intensity toward prevention or treatment of additional comorbid chronic illnesses that track with aging and longevity. While an interesting clinical trial identified a proportion of patients who were able to retain complete molecular remission for 1 year after discontinuing imatinib [2], the current standard of care is to continue TKI therapy indefinitely. Knowledge of the safety profile of imatinib is limited to just over a decade, and younger patients may be exposed to TKI therapy for much longer. For this reason, the relevance of long-term/late-emerging toxicities is obvious, but the probability with which they might occur is unknown. Ranking high among these concerns is the potential association between imatinib and secondary cancers, which is shared by Duman et al. in this issue of Leukemia and Lymphoma [3].

Earlier studies have evaluated the association between CML and secondary cancers in the “pre-imatinib era.” Investigators from the National Cancer Institute analyzed subsequent primary cancers in 8005 patients with CML diagnosed between 1973 and 2000, and found a 16% increased risk (observed/expected ratio [O/E] 1.16, confidence interval [CI] 1.04–1.29), with a cumulative incidence of new malignancy of 4.8% at 15 years [4]. The increased risks of lung and oral cavity cancer, and non-Hodgkin lymphoma (NHL), were reported in both sexes, although women had a higher risk of colon cancer. Using population-based data from the Swedish national cancer registry, 2753 patients with CML diagnosed between 1970 and 1995 were found to have a higher incidence of subsequent primary cancer, compared to those Swedish citizens of similar age and sex (standardized incidence ratio [SIR] 1.82, CI 1.53–2.14) [5]. Increased risks of stomach, skin and urogenital cancers were identified, but not for oral and lung cancer, or NHL [5]. Neither study intended to establish causality, nor did they implicate specific CML treatments of this era (hydroxurea, busulfan and interferon) with second cancer risk, yet they provide a useful reference rate.

In the “pre-imatinib era,” transplant was a far more common treatment approach for CML, and studies have similarly evaluated the risk of second cancers in transplant recipients with CML. A multi-institutional study of 28 874 allogeneic transplant recipients identified 189 second cancers (most commonly, oral cavity, liver, central nervous system, thyroid, bone, soft tissue and melanoma), at a rate that was twice that of the general population (O/E 2.1, CI 1.8–2.5), with a risk that increased over follow-up time [6]. Patients with CML constituted 26% of this cohort, but the frequency and specific cancer type in patients with CML were not reported. In this study, risk factors for invasive cancer were related to age and radiation exposure, and in some, chronic graft-versus-host disease (GVHD). A second, large study comprising 2576 patients with CML, typically pretreated with hydroxyurea or interferon before transplant in chronic phase, identified 44 solid cancers in this group, with a higher risk of cancers of the lip and esophagus; in comparison to the general population, the O/E ratio was 1.31 (CI 0.95–1.76) [7]. At 10 years post-transplant, the cumulative incidence of solid cancer was 2.4%. Smoking prior to transplant was more common in those with second cancers, and chronic GVHD may have influenced cancer risk. Taken together, an increased risk of second cancers after CML has been reported in the “pre-imatinib era,” when a broad array of treatment modalities was employed.

Imatinib became implicated in this secondary cancer risk after investigators in France observed secondary neoplasms in their own cohort of 189 patients with CML [8]. An in-depth analysis would uncover six cases treated with imatinib over a 5-year period; each of these patients had received interferon previously. While the total number of cancer cases was similar to the French population at large, the frequency of prostate cancer was felt to be four times higher. Novartis responded with an analysis of 110 second malignancies from their own global database of 9518 patients treated with imatinib, but the SIRs for all cancers, including prostate, were similar (0.87, 95% CI 0.69–1.08) to cancer rates in the general population [9]. Details of the patients in this cohort were not reported, including their particular disease (CML or other), treatments prior to imatinib, and other competing risk factors for secondary cancer. The mean time-at-risk for patients in this database was just over 1 year, and to bridge a gap in data, investigators from M. D. Anderson reported their own experience, with a retrospective evaluation of 1342 patients with CML treated with imatinib [10]. Sixty-three patients with CML were diagnosed with second malignancies; 34 received imatinib only, while 27 were previously treated with interferon. Overall, the SIR of second cancers was lower than expected (0.6, 95% CI 0.44–0.81), but an analysis of cancer subtype implied an increased risk of melanoma, kidney and endocrine cancers. Finally, the Imatinib Long-Term Side Effects (ILTE) study, an independent multicenter study with data from 27 centers worldwide, reported 30 second cancers (prostate and breast most common) among 832 patients, treated for approximately 6 years; this frequency was comparable to the expected incidence based on age and sex distribution from worldwide cancer registries [1].

Duman et al. discuss the majority of these studies, with the exception of the ILTE, and add seven cases of secondary cancers in patients with CML to the literature (one additional case was reported in a patient with a gastrointestinal stromal tumor). The duration of imatinib use ranged from 14 months to 7 years, and three of these patients had received prior therapy, particularly hydroxyurea and interferon. Three patients were heavy smokers, and in each case, developed cancers that have been epidemiologically associated with smoking (gastric cancer, small cell lung cancer and non-Hodgkin lymphoma), confounding an association between imatinib and secondary neoplasia. The largest limitation of this report is acknowledged by the authors: these patients were not part of a cohort, and the frequency of secondary cancer in their population with CML is unknown. Also, the secondary cancer cases were not compared to the population at large with such measures as the O/E ratio or SIR. This report cannot strengthen or weaken any association between CML, imatinib and secondary cancer, yet the importance of continued reporting of long-term toxicities cannot be denied.

When placing the second cancer risk in the context of the population at large in the two “CML eras,” the evidence is stronger for this association in the “pre-imatinib era” (Table I). This is not to suggest that imatinib is protective, but that an association between this agent and second cancer risk in CML is yet to be formally established. Clearly, the issue of secondary cancer in CML is complex, and if there is any true increase in the incidence, it may be driven by a culmination of various factors, including contributions from cumulative treatment, possibly when radiation, hydroxyurea or busulfan are utilized; immunodeficiency, lifestyle choices, particularly smoking, aging, genetic predisposition and aggressive surveillance might also account for the increased incidence recognized in “pre-imatinib era” studies. As long-term follow-up with nilotinib and dasatinib become available, these agents will be similarly scrutinized with regard to second cancer risk, and other long-term toxicities. Because of improving survival and longevity, age-appropriate cancer screening should play a significant role in the care of the patient with CML, but the current literature cannot implicate imatinib as a culprit in second cancer risk.

Table I. Second cancers in patients with CML prior to and during the “imatinib era.”

Author/reference, no. of patients with CML	Era	No. of second cancers	Cancer types emphasized	Observed to expected ratio/SIR (95% CI)
Ishibe et al. [4], 8005	Pre-imatinib: 1973–2000	334	Oral cavity, lung, non-Hodgkin lymphoma	1.16 (1.04–1.29)
Rebora et al. [5], 2753	Pre-imatinib: 1970–1995	145	Stomach, skin, urogenital tract	1.82 (1.52–2.14)
Rizzo et al. [6], 7594 of 28 874	Pre-imatinib: transplant, 1964–1996	189*	Oral cavity, liver, CNS, thyroid, bone and soft tissue, melanoma	2.1 (1.8–2.5)
Majhail et al. [7], 2576 of 4318	Pre-imatinib and imatinib: transplant, 1986–2005	44 of 66	Lip, esophagus	1.31 (0.95–1.76)
Roy et al. [8], 189	Imatinib	6	Prostate	Observed^†: 478.5 per 100 000 person-years; expected^†: 35.0–110.6 per 100 000 men
Pilot et al. [9], 9518	Imatinib	110	Not specified	0.87 (0.69–1.08)
Verma et al. [10], 1342	Imatinib	63	Melanoma, kidney and endocrine	0.6 (0.44–0.81)
Gambacorti-Passerini et al. [1], 832	Imatinib	30	Prostate, breast	Observed^‡: 30 cases; expected^‡: 28.6
Duman et al. [3], denominator unknown	Imatinib	7	Non-Hodgkin lymphoma, gastrointestinal, lung, skin	Not reported

^†These observations pertain to prostate cancer alone; total cancer cases in accordance with French population.

^‡Authors report an expected incidence of 28.6 cases based on age/sex distribution from worldwide cancer registries.

CML, chronic myeloid leukemia; CNS, central nervous system; SIR, standardized incidence ratio; CI, confidence interval.

*Number in entire cohort (second cancers after CML not specified).

Potential conflict of interest

Disclosure forms provided by the authors are available with the full text of this article at www.informahealthcare.com/lal.