Second Malignancies after Chemotherapy

Daniel J. Becker

Thomas S. Uldrick

Alfred I. Neugut

The era of modern chemotherapy traces back to the 1940s when it was discovered that the nitrogen mustards suppressed both the lymphoid and the myeloid cell lines. By the 1960s, MOPP (mechlorethamine, vincristine, procarbazine, and prednisone) was first used to treat patients with Hodgkin lymphoma effectively. Today, multidrug regimens are routinely available to treat practically all forms of cancer. These therapies, however, can increase long-term complications, including the risk of developing a new cancer. Typical early reports were published in 1969 and 1970 with descriptions of bladder cancer after the use of chlornaphazine for polycythemia vera and the occurrence of acute myeloid leukemia after alkylating agents for multiple myeloma.¹^,² Epidemiologic research with controlled observational studies over the ensuing decades has confirmed this consistent association of secondary leukemias with alkylating agents and drugs that target DNA-topoisomerase II, such as the epipodophyllotoxins.³^,⁴ Other forms of chemotherapy, such as many of the antimetabolites, do not appear to be carcinogenic (Table 24-1). Advances in cytogenetic research have provided a model for subcategorization of secondary leukemias based on genetic pathways of leukemogenesis.⁵

In the following sections, we will focus on studies of patients treated with chemotherapy that have provided information on the risks and characteristics of secondary cancer. Neither the effects of long-term immunosuppression nor those of hormonal therapies, such as tamoxifen, are covered in this review.

LEUKEMIA AFTER CHEMOTHERAPY

Treatment-related acute myeloid leukemia (t-AML) is by far the most frequently reported cancer after chemotherapy. t-AML has been documented after the alkylating agent treatment of Hodgkin lymphoma,⁶^,⁷^,⁸^,⁹^,¹⁰^,¹¹^,¹²^,¹³^,¹⁴^,¹⁵^,¹⁶^,¹⁷^,¹⁸^,¹⁹^,²⁰^,²¹^,²²^,²³^,²⁴^,²⁵^,²⁶^,²⁷^,²⁸^,²⁹^,³⁰^,³¹^,³² multiple myeloma,³³^,³⁴^,³⁵^,³⁶^,³⁷^,³⁸^,³⁹ non-Hodgkin lymphoma (NHL),⁴⁰^,⁴¹^,⁴²^,⁴³^,⁴⁴^,⁴⁵^,⁴⁶^,⁴⁷^,⁴⁸ breast cancer,⁴⁹^,⁵⁰^,⁵¹^,⁵²^,⁵³^,⁵⁴^,⁵⁵^,⁵⁶ ovarian cancer,⁵⁷^,⁵⁸^,⁵⁹^,⁶⁰^,⁶¹^,⁶² lung cancer,⁶³^,⁶⁴^,⁶⁵^,⁶⁶^,⁶⁷^,⁶⁸ testicular cancer,⁶⁹^,⁷⁰^,⁷¹^,⁷²^,⁷³^,⁷⁴^,⁷⁵^,⁷⁶^,⁷⁷^,⁷⁸^,⁷⁹ various childhood cancers,⁸⁰^,⁸¹^,⁸²^,⁸³^,⁸⁴^,⁸⁵^,⁸⁶^,⁸⁷^,⁸⁸^,⁸⁹^,⁹⁰^,⁹¹ gastrointestinal cancer,⁹²^,⁹³^,⁹⁴ brain cancer,⁹⁵ and polycythemia vera.⁹⁶

The magnitude of risk has been estimated by various methods. Relative risk (RR) has been calculated from adverse outcomes in randomized controlled trials as well as case-control studies within large cohorts. Standardized incidence ratios (SIR) compare the incidence within a treated cohort to the expected incidence based on general population data. The range of reported RRs and SIRs for leukemia after chemotherapy is wide, between 1- and 100-fold, depending on the disease being treated and the intensity of treatment. Because leukemia is rare, however, the absolute risk and the cumulative probability (actuarial risk) are often used to indicate the true population impact and often may have more relevance to the clinician and patient, whereas the RR and SIR are more helpful in establishing causality. For example, an RR of 2 for leukemia that is associated with cyclophosphamide therapy for breast cancer would correspond to an absolute risk or excess of approximately 5 leukemias in 10,000 patients over a 10-year period.⁴⁹

The risk of t-AML after chemotherapy depends on many factors, including the drug(s) administered, duration of treatment, cumulative dose, dose intensity, age of the patient, and the concomitant use of radiotherapy. The risk estimates also differ by whether treatment-related myelodysplastic syndromes (t-MDS) are included. Similar to t-AML, these preleukemic conditions are frequently fatal.⁹⁷^,⁹⁸

The association of chemotherapeutic agents with leukemia is related to their mechanism of action. The alkylating agents bind covalently to DNA and have been convincingly linked to leukemia in many studies, although details of the relation between alkylating agents and specific karyotypic subtypes are not understood. Some alkylating agents are less leukemogenic than others; for example, cyclophosphamide appears to be less leukemogenic than melphalan.³⁴^,⁵⁹ MDS and t-AML developing after the use of cyclophosphamide for nonneoplastic conditions, such as rheumatoid arthritis⁹⁹^,¹⁰⁰ and Wegener granulomatosis,¹⁰¹ has been described, supporting the theory that the drug causes the leukemia in patients with and without malignancies. Dose-intense cyclophosphamide has been shown to be more leukemogenic than lower doses in adjuvant breast cancer.¹”¹⁰²

Nitrosoreas have also been associated with secondary leukemias. In a randomized trial of 3,633 patients with gastrointestinal cancer, 14 leukemias occurred in 2,067 patients who were given semustine, whereas only 1 occurred in 1,566 patients who were given other therapies.⁹² Higher doses of semustine were more strongly associated with leukemia development.⁹³ BCNU was linked to t-AML in a small clinical trial series of patients with brain cancer.⁹⁵

The epipodophyllotoxins, etoposide and teniposide, used alone or in combination chemotherapy, increase the risk of leukemia.⁴^,¹⁰³^,¹⁰⁴^,¹⁰⁵ The epipodophyllotoxins bind directly to DNA-topoisomerase II, leading to chromosome breakage and cell death. Epipodophyllotoxin-related leukemias differ from alkylating agent-related leukemias. They develop after a shorter latency, involve translocations rather than deletions at the molecular level, and have a better prognosis. Other

chemotherapeutic agents that inhibit DNA-topoisomerase II, such as doxorubicin, epirubicin and mitoxantrone, may cause leukemia, especially in combination with alkylating agents.⁹¹^,¹⁰⁶^,¹⁰⁷^,¹⁰⁸^,¹⁰⁹ Distinguishing between the effects of chemotherapeutics frequently used in multidrug regimens is challenging. A recent review of mitoxantrone use in multiple sclerosis, a nonmalignant setting in which mitoxantrone has been used as a single agent, suggested increased risk of posttherapy leukemia, and a preponderance of acute promyelocytic leukemia (APL) (63%) among those cases.¹¹⁰

Table 24-1 Carcinogenicity of Selected Anticancer Drugs

Drug	Human Carcinogen	Animal Carcinogen	References
Alkylating Agents
Busulfan (Myleran)	+++	+	^66,174
Carmustine (BCNU)	++	+++	^95,174
Chlorambucil (Leukeran)	+++	+++	^{6,8,41,57,96,174,282}
Chlornaphazine	+++	+	^{174, 239 [more]}
Cisplatin	+++	+++	^{62,79,174,283}
Cyclophosphamide (Cytoxan)	+++	+++	^{41,48,49,59,101,174,209,236,240,244}
Dacarbazine (DTIC)	+	+++	¹⁷⁴
Ifosfamide	ND	+	¹⁷⁴
Lomustine (CCNU)	++	+++	^9,12,174
Mechlorethamine (nitrogen mustard)	++	+++	^7,12,13,174
Melphalan (phenylalanine mustard/Alkeran)	+++	+++	^{34,49,52,59,174}
Mitomycin-C	+	+++	¹⁷⁴
Prednimustine	++	±	^41,284
Procarbazine	++	+++	^8,174
Semustine (methyl-CCNU)	+++	+	^92,93,174
Thiotepa	+++	+++	^57,174
Treosulfan	+++	ND	^57,60,174
Uracil mustard	+	+++	¹⁷⁴
Antimetabolites
5-Fluorouracil (5-FU)			¹⁷⁴
6-Mercaptopurine (6-MP)			¹⁷⁴
Methotrexate			^174,285
Mitotic inhibitors
Vinblastine	±		^8,174
Vincristine			¹⁷⁴
DNA intercalation
Dactinomycin (actinomycin D)		+	¹⁷⁴
DNA strand breakage
Bleomycin			¹⁷⁴
DNA intercalation and DNA topoisomerase II inhibition^a

Doxorubicin (Adriamycin)	+	+++	^{17,91,106,128,174}
Epirubicin (4′-epi doxorubicin)	±		^107,128
Mitoxantrone	±		¹²⁸
DNA topoisomerase II inhibition^a
Etoposide (VP 16)	++		^12,76,83,86
Teniposide (VM 26)	++		¹²⁸
Razoxane	±		¹²⁸
+++, carcinogenic; ++, probably carcinogenic; +, possibly carcinogenic; not known to be carcinogemoc; ±, uncertain due to limited, inadequate, or inconsistent data; ND.
^aFor practical purposes, these anticancer drugs could be classified as human leukemogens; however, the leukemogenic effect is most notably apparent in the presence of alkylating agent or cisplatin therapy.

One class of antimetabolites, the thiopurines, including 6-MP and azathioprine, is also associated with t-AML. A review of > 180,000 solid organ transplant recipients reported an increased incidence of AML, and a dose-response relationship between azathioprine and AML development.¹¹¹ Another review of 439 children treated for acute lymphocytic leukemia (ALL) suggested that the risk of leukemia correlated with levels of 6-MP metabolites in susceptible patients.¹¹² These patients were treated with multiple cytotoxic agents and had a history of ALL, which confounds interpretation of any direct effect of 6-MP on leukemogenesis. Both alkylator and thiopurineassociated t-AML cases are associated with defective DNA mismatch repair.¹¹³

It is difficult to disentangle the separate effects of cumulative dose, dose intensity,¹¹⁴ and duration of administration, because these measures are highly intercorrelated; for example, patients who receive large cumulative exposures usually have long durations of treatment. Dose-response relationships, however, have been reported among patients who are given alkylating agents for Hodgkin lymphoma,⁶^,²² breast cancer,⁴⁹^,¹⁰² gastrointestinal cancer,⁹³ NHL,⁴¹^,⁴⁴ ovarian cancer,⁵⁷^,⁵⁹^,⁶²^,¹¹⁵ testicular cancer,⁷⁹ and childhood cancer.¹⁷ Duration of treatment has been proposed as an additional factor, with continuous exposure of stem cells to a stream of cytotoxic agents, possibly enhancing the progression of a transformed cell.³⁴ The period of highest risk of t-AML is 2 to 10 years after initial treatment. A shorter latency of 1 to 3 years, is seen with topoisomerase II inhibitors than with alkylating agents, frequently 5 to 7 years. In three large series, t-AML risk decreased over time, but remained significantly elevated for >10 years after treatment.²⁵^,⁴⁷

The importance of gender and age on the risk of chemotherapy-related AML is unclear. A few series have indicated that women may be at higher risk than men for second cancer development,²⁵^,¹¹⁶^,¹¹⁷ but other series have challenged this conclusion.⁶^,⁷^,¹³^,¹¹⁸ Age at treatment has been associated with increased risk in several Hodgkin lymphoma cohorts, with treatment over the age of 40 years carrying a higher risk, perhaps 4-fold higher, than treatment at a younger_age.⁷^,¹²^,¹⁴^,¹⁵^,²² This interaction of age and risk, however, was not seen in one large series.²⁵ Chemotherapy for childhood cancer also appears to carry a high risk of leukemia,¹¹⁹^,¹²⁰^,¹²¹^,¹²²^,¹²³^,¹²⁴ with adolescents at higher risk than younger patients in one cohort.¹¹⁷

Because radiotherapy is often given in combination with chemotherapy, it is important to determine whether the two modalities together synergistically promote leukemia. One large study of patients with breast cancer suggested a possible interaction between radiation and chemotherapy,⁴⁹ but such an interaction is not generally seen in Hodgkin lymphoma patients.¹¹ The risk of t-AML is likely to be related to the extent to which bone marrow is exposed in radiation fields. Relatively small increased risks of t-AML have been observed after pelvic irradiation for the treatment of cervical, uterine, rectum, and anus in several large cancer registry-based studies.¹²⁵^,¹²⁶^,¹²⁷ The risk of radiation-induced t-AML appears to peak 5 to 10 years after pelvic irradiation, with risk persisting beyond 10 years.

Pathogenesis

Historically, treatment-related leukemias have been grouped into two distinct categories: leukemias related to alkylating agents and those secondary to topoisomerase II inhibitors.¹²⁸ Leukemia related to alkylating agents frequently exhibits deletion of chromosome 5q or 7q, whereas reciprocal translocations are associated with DNA-topoisomerase II inhibitors.¹²⁸^,¹²⁹ Longer latency periods and more frequent presentation with MDS prior to leukemia also characterize alkylator-associated leukemias.¹³⁰ Topoisomerase II is an enzyme that cuts doublestranded supercoiled DNA and then ligates the break, to relieve tension and facilitate transcription. Topoisomerase II inhibitors, including the epipodophyllotoxins (e.g., etoposide), anthracenediones (e.g., mitoxantrone), and anthracyclines (e.g., doxorubicin and epirubicin), interfere with the proper ligation of the topoisomerase II-induced double-strand break. One revealing study established that cases of APL (t(15;17)) in patients who received mitoxantrone for breast or laryngeal cancer had a characteristic chromosome 15 breakpoint at a location known to be cleaved by mitoxantrone-poisoned topoisomerase II.¹³¹ Similar chromosomal breakpoint “hotspots” at sites of epirubicin/topoisomerase II cleavage were noted in six cases of epirubicin-associated APL. These hotspots were distinct from the sites of mitoxantrone cleavage.¹³² Taken together, these observations offer a compelling mechanism for topoisomerase II inhibitor-mediated formation of leukemogenic translocations in APL after aberrant DNA splicing at topoisomerase cleavage sites and recombination by the nonhomologous endjoining machinery. This observation was reinforced by a study of t-APL patients who had received mitoxantrone for multiple sclerosis and had DNA breakpoints similar to those seen in the breast/laryngeal cancer-treated cohort.¹³³ Etoposide and anthracyclines are also frequently associated with translocations involving 11q23,⁶⁸^,⁸⁴^,¹³⁴^,¹³⁵ which lead to chimeric rearrangement of the MLL gene¹²¹^,¹³⁶ at a recognized site of epirubicin cleavage.¹³⁷ The mechanism by which alkylators promote leukemogenesis is less clear but possibly related to selection for cells with poor mismatch repair pathways.¹³⁸

To account for more of the genetic and clinical complexity of t-AML, Pedersen-Bjergaard and colleagues⁵ proposed subdividing the t-AMLs into eight distinct pathogenic pathways characterized by distinct genetic alterations. Many of the central regulators of cell growth, division, and apoptosis are implicated in the complex cytogenetic abnormalities in t-AML. RAS, RUNX1, and TP53 have all been noted to be mutated with high frequency in t-AML/t-MDS.¹³⁹^,¹⁴⁰^,¹⁴¹ RAS mutations in
particular have been associated with progression from t-MDS to t-AML.¹⁴² Investigation of the etiology and effect of the genetic abnormalities of t-AML and t-MDS is an area of active research.

Recent research on the pathogenesis of cancer explores epigenetic changes to the cancer genome, in the form of DNA methylation and histone deacetylation. Adding methyl groups to DNA promotor sequences and removing acetyl groups from histones decrease the translation of the affected genes and can have a profound effect on cell survival. Cyclin-dependent kinase inhibitor p15 is an important negative regulator of the cell cycle. In a study of tumor tissue from 81 t-AML patients, p15 methylation density and frequency increased with increasing stage of disease.¹⁴³ Death-associated protein kinase (DAPK), a potential tumor suppressor gene, was frequently hypermethylated in one study of AML and MDS.¹⁴⁴ Cases of t-AML in this study were significantly more likely to be associated with DAPK hypermethylation than cases of de novo AML. Studies of epigenetic abnormalities in t-AML are in their infancy but likely to contribute significantly to the understanding of this disease.

Prognosis

Population survival data from the 1970s and 1980s demonstrated a worse prognosis for t-AML when compared with de novo AML, with an estimated 12-month survival of 10% versus 30%.¹⁴⁵ The genetic alterations of t-AML are also closely linked to both prognosis and treatment response. As with de novo AML, favorable risk cytogenetics in t-AML, including t(8;21), inv(16), and t(15;17), are associated with improved survival relative to poor risk and intermediate cytogenetics in t-AML.¹⁴⁶^,¹⁴⁷ Intermediate risk cytogenetics include t(9;11) and other abnormalities not covered in poor or favorable risk. Poor risk cytogenetics include monosomy 5 or 7, deletion 5q or 7q, 11q23 abnormalities other than t(9;11), or complex karyotype consisting of >3 abnormalities in the absence of any of the favorable risk markers. t-AML is more likely than de novo AML to have poor risk cytogenetics, and prognosis within any subgroup is worse for t-AML patients than for de novo AML.¹⁴⁷ Anthracyclines are a mainstay of induction therapy in AML and frequently cannot be used in t-AML patients because of previous exposure and cumulative cardiotoxicity. Diminished treatment options likely contribute to the poorer outcome of t-AML. In a recent follow-up of 121 patients with t-AML, median survival among those with favorable, intermediate, and unfavorable karyotypes was 26.7, 15.5, and 5.6 months, respectively.¹⁴⁸ t-AML patients with favorable disease are a heterogeneous group with respect to prognosis. A recent analysis of 188 cases of AML with core-binding factor gene abnormalities (either t(8;21) or inv(16)) which included 17 patients with t-AML, reported inferior event-free and overall survival for the t-AML patients compared with matched AML controls.¹⁴⁹ Another review of 106 patients with treatment-associated APL found outcomes similar to the outcomes of de novo APL patients.¹⁵⁰ APL is unusually responsive to nonanthracycline medications, including all trans-retinoic acid and arsenic trioxide. These additional treatment options likely contribute to the improved outcomes for t-APL patients.

Genetic Predisposition

Efforts to understand host factors that increase susceptibility to iatrogenic malignancies have focused on genetic polymorphisms in genes known to be involved in drug metabolism and DNA repair.¹⁵¹ Candidate genes that are involved in drug metabolism include glutathione S-transferases (GST), nicotinamide adenine dinucleotide phosphate:quinone oxidoreductase (NQO1), and cytochrome P450 (CYP)¹⁵² as well as genes involved in nucleotide excision DNA repair, base excision DNA repair, DNA mismatch repair, and cell-death signaling.¹⁵³ Epipodophyllotoxins and cyclophosphamide are substrates for CYP3A. Several case-control studies have reported an association between wild-type CYP3A and leukemia.¹⁰³^,¹⁵⁴ GST participates in inactivating many chemotherapeutics, including doxorubicin, etoposide, cyclophosphamide, and mitoxantrone. GST also detoxifies potentially mutagenic chemotherapy metabolites, and specific polymorphisms of GST P1 are found more frequently in t-AML compared with AML.¹⁵⁵ NQO1 participates in epipodophyllotoxin breakdown and is more frequently mutated in patients with t-AML relative to patients with AML.¹⁵⁶^,¹⁵⁷

Genes involved in DNA repair are also implicated in susceptibility to t-AML. The xeroderma pigmentosum group D gene codes for a helicase involved in DNA nucleotide excision repair. A functional lysine to glutamine polymorphism at codon 751 was evaluated in 341 elderly patients with AML. Heterozygotes and glutamine homozygotes had worse prognoses and increased risk of developing t-AML.¹⁵⁸ Both RAD51 and XRCC3 participate in the homologous repair of doublestranded DNA breaks. Abnormalities in both genes are associated with AML and more strongly associated with t-AML.¹⁵⁹ Abnormalities of DNA mismatch repair, also essential to maintaining genetic integrity, have been implicated in development of t-AML. A case-control study of 133 patients with therapyrelated cancers reported an odds ratio of 5.31 (95% CI, 1.40 to 20.15) for the development of t-AML among those exposed to methylating agents who had a particular variant of MLH1 mismatch repair gene.¹⁶⁰ These studies support the hypothesis that genetically determined variation in the pharmacokinetics of chemotherapeutic agents or the ability to repair DNA damage caused by chemotherapy may alter the risk of chemotherapyrelated secondary leukemias. Future translational research is needed to evaluate the possible risk stratification or prevention strategies based on pharmacogenetic profiles.

Treatment

Treatment approaches for the management of t-AML and t-MDS are largely based on conventional approaches to AML. Complete remission (CR) rates in t-AML after alkylating agents range from 28% to 65%, compared with 80% to 90% CR rates after topoisomerase II inhibitor-associated t-AML.¹⁶¹ Relapse occurs frequently with long-term survival rates of only10% to 20%.¹⁶¹˜¹⁶⁵ A recent review of 96 patients with secondary AML, 53% with t-AML, treated with aggressive induction chemotherapy reported a median event-free survival of 8 months and median-overall survival of 13.6 months.¹⁶⁶

In selected cases, allogeneic stem cell transplant may be curative. A recent review of 56 patients with t-AML and 55 with t-MDS treated with allogeneic stem cell transplants at the Fred Hutchinson Cancer Center¹⁶⁷ reported 1- and 5-year disease-free survival among the t-AML patients of 20% and 9%, respectively. The 5-year relapse rate was 40% and the 5-year nonrelapse mortality was 51%. The largest series to date reported outcomes for 868 patients transplanted for either t-AML (545 patients) or t-MDS (323 patients).¹⁶⁸ Treatmentrelated mortality and relapse rates were 48% and 31% at 5 years, respectively. Overall and disease-free survival at 5 years were 22% and 21%, respectively. These investigators identified four predictors of worse overall and progression-free survival: age >35, poor risk cytogenetics, t-AML not in remission, or advanced t-MDS, and donor other than an HLA identical sibling or partially or matched unrelated donor. Five-year survival for patients with none of the risk factors was 50%, but only 4% for patients with four risk factors. These studies confirm a small but reproducible chance of cure, albeit with a high risk of treatment-related death, among the select group of patients who are eligible for allogeneic transplant.

Many patients, however, will be ineligible for allogeneic transplant but may be candidates for high-dose therapy and autologous stem cell transplant (autoSCT). In a retrospective evaluation of 65 patients treated for t-AML/MDS (86% t-AML) withautoSCT, 3-year overall and disease-free survival were 35% and 32%, respectively, with 12% treatment-related mortality.¹⁶⁹ However, it should be noted that few patients had evidence of high-risk cytogenetics. Age <40 and treatment in first CR predicted lower relapse risk. A report from the Mayo Clinic of only six patients suggested that perhaps transplanting cells harvested prior to the development of the secondary malignancy holds promise.¹⁷⁰ Early harvest autologous transplant may represent an intriguing possibility in these patients with very limited therapeutic options. APL as a second malignancy may respond to ATRA and idarubicin as well as de novo APL, although a large case-control series did not restrict cases to APL after chemotherapy.¹⁷¹

ACUTE LEUKEMIA AFTER HODGKIN LYMPHOMA

The treatment of Hodgkin lymphoma disease with radiation chemotherapy and combination chemotherapy is one of the great triumphs of modern cancer therapy. In 1960, the diagnosis of Hodgkin lymphoma meant nearly certain death within a few years; today, patients are mostly cured. This success, however, has resulted in long-term complications of clinical importance, including increased second cancer development.¹⁷²^,¹⁷³ More studies have been published describing the risk of AML after Hodgkin lymphoma than after any other cancer (Table 24-2). Several large series of many thousands of patients report >1,000 secondary leukemias.⁶^,⁷^,⁸

Type of Chemotherapeutic Agent

Patients with Hodgkin lymphoma can be treated with combinations of drugs and may receive additional cytostatic chemicals for relapse; it is, therefore, difficult to attribute independent effects to individual drugs. Combination chemotherapy, particularly MOPP, has been convincingly linked to secondary leukemias, with mechlorethamine (nitrogen mustard) the most potent leukemogen of the four agents,¹¹⁴ although procarbazine is also a leukemogen in animals.¹⁷⁴ Mechlorethamine was the strongest predictor of leukemia risk in most,⁶^,⁷^,¹² but not all, studies.⁸ Doxorubicin, bleomycin, vinblastine, and decarbazine (ABVD) appears to be less leukemogenic than MOPP,¹⁵^,¹⁷⁵ with the cumulative risks estimated at 1.5% and 5.9%, respectively.⁸ Among 12,411 patients in the International Database on Hodgkin lymphoma, an RR of 17 was linked to MOPP combination chemotherapy.⁷ The PAVe regimen (procarbazine, melphalan, and vinblastine) appears to be less leukemogenic, with oral melphalan replacing mechlorethamine.¹⁴^,¹⁷⁶ It is encouraging that the chemotherapy regimens introduced in the 1980s appear to carry a lower risk of leukemia than earlier drug combinations did.¹¹ A recent evaluation of 35,511 survivors of Hodgkin lymphoma from international registries reported an excess absolute risk (EAR) of 6.2 cases of AML per 10,000 patient-years (95% CI, 5.4 to 7.1).¹⁷⁷ EAR was significantly elevated throughout the 15 years of follow-up, but decreased over time. Throughout the follow-up, the EAR for patients treated between 1984 and 2001 was significantly less than the EAR for patients treated between 1970 and 1984.¹⁷⁷ Highdose chemotherapy and autoSCT have become an important therapeutic option in selected patients with relapsed Hodgkin lymphoma.¹⁷⁸ Current evidence does not suggest an increased incidence of t-AML among Hodgkin lymphoma patients treated with autologous transplant.¹⁷⁹

Dose-Response

The number of cycles or treatments received is directly related to AML risk.⁶^,⁹^,¹² Attempts have been made to correlate the risk of AML with measures of cumulative dose, and it is clear that the higher the dose, the higher the_ri_sk.¹²^,¹³^,¹⁷^,²²^,¹¹⁸ Although quantitative data are sparse, an RR of 10 was reported after 60 mg per m² mechlorethamine¹³ in the Union International Contra Cancer series, compared with 0 mg per m². In the Netherlands, an RR of 10 was linked to 100 mg, which reached an RR of 84 at approximately 220 mg.¹² Assuming a body surface area of 1.8 m² for patients in the Netherlands, these estimates are similar. Although most modern regimens for Hodgkin lymphoma use less leukemogenic agents, high doses of cyclophosphamide in combination with procarbazine and etoposide have been well studied in both the standard and the dose-escalated BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisone) regimens by the German Hodgkin Study Group in patients with advanced stage and high-risk Hodgkin lymphoma. With 10-year followup, standard and escalated BEACOPP have estimated 10-year cumulative incidence of t-AML of 2.2% and 3.0%, respectively; both greater than a 0.4% cumulative incidence seen in patients receiving a regimen with less intense alkylating agents and no etoposide, ABVD/COPP (doxorubicin, bleomycin, vinblastine, dacarbazine, alternating with cyclophosphamide, vincristine, procarbazine, and prednisone).¹⁸⁰^,¹⁸¹ The risk of t-AML,
therefore, remains an important consideration in treatment choices for patients with Hodgkin lymphoma.

Table 24-2 Acute Leukemia after Hodgkin Disease

Study (Reference)	No. of Patients	No. of Observed Leukemias	SIR or RR	Actuarial Risk (%/y)
IARC⁶	9,552	163	9.0 (RR)
IDHD⁷	2,411	158	27.5 (SIR)	2.4/20
Canada/US⁸	9,280	122	23.9 (RR)	–
BNLI^9,10,25	5,519	45	21.1 (SIR)	1.7/20
GHSG²⁸⁶	5,411	46	–	1/5
Netherlands^11,12	1,939	31	34.7 (SIR)	4.0/15
UICC¹³	1,681	18	64.0 (RR)	2.3/10
Nordic²⁶	1,641	7	17.0 (SIR)	0.8/30
Stanford¹⁴	1,507	28	66.0 (SIR)	4.2/10
LESG^24,27,287	1,380	24	–	2.8/14
Milan¹⁵	1,329	19	–	3.6/10
Harvard³²	1,319	23	82.5 (SIR)	1.4/10
German-Austrian^30,288	1,245	10	–	0.6/22
Oslo¹⁶	1,152	9	24.3 (SIR)
Munich²⁸	1,120	8	–	1.0/15
Houston¹⁸	1,013	14	–	–
Gustave Roussy¹⁹	892	8	27.6 (SIR)	1.7/15
CALGB²⁰	798	10	133.0	4.0/10
Stanford pediatric²⁹	694	8	–	1.5/20
SWOG²¹	659	21	–	6.2/10
Denmark²²	391	17	–	9.9/10
NCI 1964-1981²³	473	9	–	6.0/10 (est)
-, not reported.
BNLI, British National Lymphoma Investigation; CALGB, Cancer and Leukemia Group B; GHSG, German Hodgkin Lymphoma Study Group; IARC, International Agency for Research on Cancer; IDHD, International Database on Hodgkin Disease; LESG, Late Effects Study Group; NCI, National Cancer Institute; SWOG, Southwest Oncology Group; UICC, Union International Contra Cancer.
These institutions are not all independent series. For example, the International Database on Hodgkin Disease also includes data from the British National Lymphoma Investigation.