Maximizing Breast Cancer Therapy with Awareness of Potential Treatment‐Related Blood Disorders

Abstract

In this review we summarize the impact of the various modalities of breast cancer therapy coupled with intrinsic patient factors on incidence of subsequent treatment‐induced myelodysplasia and acute myelogenous leukemia (t‐MDS/AML). It is clear that risk is increased for patients treated with radiation and chemotherapy at younger ages. Radiation is associated with modest risk, whereas chemotherapy, particularly the combination of an alkylating agent and an anthracycline, carries higher risk and radiation and chemotherapy combined increase the risk markedly. Recently, treatment with granulocyte colony‐stimulating factor (G‐CSF), but not pegylated G‐CSF, has been identified as a factor associated with increased t‐MDS/AML risk. Two newly identified associations may link homologous DNA repair gene deficiency and poly (ADP‐ribose) polymerase inhibitor treatment to increased t‐MDS/AML risk. When predisposing factors, such as young age, are combined with an increasing number of potentially leukemogenic treatments that may not confer large risk singly, the risk of t‐MDS/AML appears to increase. Patient and treatment factors combine to form a biological cascade that can trigger a myelodysplastic event. Patients with breast cancer are often exposed to many of these risk factors in the course of their treatment, and triple‐negative patients, who are often younger and/or BRCA positive, are often exposed to all of them. It is important going forward to identify effective therapies without these adverse associated effects and choose existing therapies that minimize the risk of t‐MDS/AML without sacrificing therapeutic gain.

Implications for Practice

Breast cancer is far more curable than in the past but requires multimodality treatment. Great care must be taken to use the least leukemogenic treatment programs that do not sacrifice efficacy. Elimination of radiation and anthracycline/alkylating agent regimens will be helpful where possible, particularly in younger patients and possibly those with homologous repair deficiency (HRD). Use of colony‐stimulating factors should be limited to those who truly require them for safe chemotherapy administration. Further study of a possible leukemogenic association with HRD and the various forms of colony‐stimulating factors is badly needed.

Short abstract

This review summarizes the effect of various breast cancer therapies and related patient factors on the incidence of treatment‐induced myelodysplasia and acute myelogenous leukemia.

Introduction

In 1992 Curtis et al. published a landmark study of leukemia risk after radiation and chemotherapy in the adjuvant treatment of breast cancer 1. Since that time, many studies have evaluated the risks of these two modalities for the development of treatment‐related myelodysplastic syndrome and acute myelogenous leukemia (t‐MDS/ML). Recently, Morton et al. reported an observation of elevated t‐MDS/AML in many patients with solid tumors 2. Their report of the increased incidence of treatment‐related myeloid neoplasms among patients who received radiation and chemotherapy is consistent with reporting from a number of other studies. The chemotherapy drugs most often implicated were alkylating agents, topoisomerase II inhibitors (including anthracyclines), and platinum compounds. Looking at the issue of treatment‐induced t‐MDS/AML, at least four and perhaps more factors appear to be working independently or in concert to influence the risk of t‐MDS/AML: patient age, radiation therapy, leukemogenic chemotherapy drugs, the use of granulocyte colony‐stimulating factors (G‐CSFs), homologous DNA repair defects (including BRCA mutations), and the introduction of poly (ADP‐ribose) polymerase (PARP) inhibitors.

Materials and Methods

A literature search of the PubMed database was conducted on July 15, 2019. The search terms “breast cancer” and “treatment‐ or therapy‐related leukemia” and English language were applied. This yielded 800 results. Of these, 60 were clinical trials, 175 were reviews, and 8 were systematic reviews. A subsequent search was done to identify studies including observational and registry‐based studies related to the risk of treatment‐related blood disorders after breast cancer treatment and treatment with granulocyte colony‐stimulating factors. Including the term “acute myeloid dysplasia” yielded 42 studies and the term “myelodysplasia” yielded 30 studies with some duplications. The search criteria, which included articles characterizing induction of therapy‐related leukemias and their occurrence in following any or unspecified cancer sites including the terms “granulocyte colony‐stimulating factors” or “GCSF” or “neutropenia” or “safety” and English language, yielded 44 studies for AML and 594 for MDS, including two systematic reviews. Selected articles from this search are referenced accordingly in our overview of this topic.

Chemotherapeutic Agents

With regard to chemotherapeutic agents, the clinical picture for alkylating agents and topoisomerase inhibitors have been well described. Therapy‐related myeloid neoplasms are characterized by clonal abnormalities involving loss of all or part of chromosome 5 and/or 7 and loss of p53 with alkylating agents, translocations involving chromosome bands 11q23 or 21q22 with anthracyclines, and other balanced rearrangements that can also be observed with de novo leukemia 3, 4. In addition, the time latency for chemotherapy‐induced t‐MDS/AML tends to be shorter for anthracyclines (2–3 years) as opposed to alkylating agents (5–7 years) 4.

For platinum‐containing agents, there are fewer data. However, Travis has reported on the use of these agents in the treatment of ovarian cancer 5. The relative risks for treatment with carboplatin and cisplatin were 6.5 (95% confidence interval [CI]: 1.2–36.6) and 3.3 (95 CI: 1.1–9.4), respectively. They found evidence of a dose–response relationship, with relative risk reaching 7.6 at doses of 1,000 mg or more of platinum (p value for trend = <.001). A correlation was also found between the number of cycles of platinum therapy and the risk of t‐MDS/AML in the Solo2 trial, which evaluated the use of olaparib monotherapy after platinum‐based chemotherapy 6.

Prior to the development of the Bonnadonna regimen of cyclophosphamide, methotrexate, and 5‐fluorouracil (CMF), previous treatment regimens for breast cancer had included melphalan, which appeared to be quite leukemogenic among patients with breast and ovarian cancer 7, 8, 9. In the Curtis et al. study, which examined these studies as well as the CMF regimen and the classic SWOG CMFVP regimen, which added steroids and vincristine, the risk of acute nonlymphocytic leukemia was significantly increased after regional radiotherapy alone (relative risk [RR]: 2.4), alkylating agents alone (RR: 10.0), and combined radiation and drug therapy (RR: 17.4) 1. Dose‐dependent risks were observed after radiotherapy and treatment with melphalan and cyclophosphamide. The magnitude of leukemogenic risk with melphalan was 10 times that of cyclophosphamide (RR: 31.4 vs. 3.1). There was little increased risk associated with total cyclophosphamide doses below 20 g. The total cyclophosphamide dose administered in the often currently used weekly CMF regimen (60 mg/m²/day × 180 days) for an average‐sized woman would fall just short of this cumulative level.

Notably, Diamandidou et al. reported the existence of a trend toward increasing leukemia risk in association with intravenous bolus cyclophosphamide doses in excess of 6 g/m² in patients treated with various types of FAC (cyclophosphamide, doxorubicin, 5‐fluorouracil) regimens, which ranged in length from 6 to 24 cycles 10. Smith et al. observed an increased risk of leukemia in association with dose intensification of intravenous bolus cyclophosphamide (4.8–9.6 g/m²), whereas Laughlin et al. reported elevated risk of leukemia associated with high‐dose, bone marrow‐ablative chemotherapy involving alkylating agents in patients with breast carcinoma who were undergoing autologous bone marrow transplantation 11, 12. Neither Smith nor Diamandidou found an association between the amount of doxorubicin administered and increased leukemia risk.

Although topoisomerase inhibitors, including anthracyclines, are known to be leukemogenic, the experience in Hodgkin disease suggests that these agents do not pose quite as serious a problem as alkylating agents do 13, 14. In the treatment of Hodgkin disease, the MOPP regimen, which contains mechlorethamine, vincristine, procarbazine, and prednisone, is clearly leukemogenic, with peak incidence occurring in the 5‐ to 10‐year range 15. Delwail et al. reported that the 15‐year risk of developing acute leukemia was 3.4% after MOPP chemotherapy and 1.3% after receiving ABVD (doxorubicin, bleomycin, vincristine, and dacarbazine) 16. Thus, although doxorubicin‐containing chemotherapy does appear to be leukemogenic when used to treat patients with Hodgkin disease, it appears to have less leukemogenic potential compared with alkylating agents.

Increased MDS/AML risk was observed for patients with breast cancer who received both cyclophosphamide and doxorubicin (AC) compared with other chemotherapy regimens 17. Similar findings were observed in the 10‐year BCIRG 006 study results, which reported seven cases of leukemia in the AC arm and one in the non‐AC arm 18, as well as a Surveillance, Epidemiology, and End Results (SEER) database study of more than 56,000 patients in which AC‐containing‐regimen patients had an MDS/AML hazard ratio of 1.86, 95% CI: 1.33–2.61, compared with patients that did not receive chemotherapy 19. Because alternate regimens in all of these studies contain either an alkylating agent or a platinum agent, it is likely that the combination of cyclophosphamide and doxorubicin carries a higher risk of leukemia compared with non–doxorubicin/alkylating agent‐containing regimens. Long‐term follow‐up of the Tryphaena (NCT 00976989) and Train‐2 (NCT 01996267) trials, which conducted similar randomizations, will be helpful in confirming this observation.

Radiation Therapy

With regard to radiation therapy, Yu et al. observed a 1.8‐fold increased risk of leukemia following radiation treatment for breast cancer (95% CI: 1.2–2.8) in an inception cohort with women not receiving radiation as the comparison group 20. In an analysis of adjuvant breast cancer trials conducted by the NSABP, Smith reported a more than doubling of the risk of t‐MDS/AML in the arms that received radiation therapy in addition to chemotherapy 11. We observed a modest increase in t‐MDS/AML in patients with stage 0 breast cancer with radiation alone 21. In a study of invasive breast cancer, a larger increase in t‐MDS/AML was seen in those who received radiation plus chemotherapy versus chemotherapy alone 17. The experience in Hodgkin disease confirms the additive role of radiotherapy to chemotherapy in leukemogenesis 13, 14.

Age and t‐MDS/AML

In a SEER data study of t‐MDS alone after breast cancer diagnosis and treatment with surgery/radiation (43%) and surgery/radiation/chemotherapy (50%), a 30‐fold increased t‐MDS incidence was observed among younger patients (<55 years, RR: 31.8), as opposed to patients age 55–74 (RR: 4.29) and patients age > 75 (RR: 2.06) 22. Age‐related risk of AML was reported by Martin et al. in their analysis of SEER data following a diagnosis of stage I–III breast cancer 23, with increased relative risk in younger but not older patients (age < 50 years, RR: 4.14, p < .001; age 50–64, RR: 2.19, p < .001; age 65+, RR: 1.19, p = .123) and stage‐dependent risk observed in younger but not older women as well. In the study by Yu et al., t‐MDS/AML risk ratios for breast cancer survivors, for all site second cancers, were age 20–49, RR: 5.5 (95% CI: 5.0–6.1), age 50–64, RR: 1.3 (1.3–1.4), and age 65+, RR: 1.2 (1.1–1.2) 20. These findings are consistent with the strong association observed between young age and leukemia in the population exposed to the atomic bombings of Hiroshima and Nagasaki during World War II 24, 25.

Growth Factors and t‐MDS/AML

The role of G‐CSF in leukemogenesis has been controversial. Initially, there was concern that these cytokines might be leukemogenic, particularly given that G‐CSF receptors are present on the surface of hematopoietic stem cells 26. It has been speculated that the antiapoptotic effect of these drugs on stem cells could allow the survival of otherwise lethally mutated cells that go on, in their damaged state, to develop neoplasia. Fortunately, no increased risk was observed in healthy stem cell donors who were treated with G‐CSF, as reported from the RADAR project 27. More recent reviews of G‐CSF long‐term effects on thousands of unrelated donors found “no clear evidence for leukemogenesis” 28, 29.

From 1994, the Severe Chronic Neutropenia International Registry has monitored patients with different forms of neutropenia, including cyclic neutropenia, severe congenital neutropenia, and idiopathic neutropenia 30, 31. Among the 387 patients with severe congenital neutropenia, 35 developed MDS or AML, with a cumulative risk of 13% after 8 years of G‐CSF treatment. In contrast, none of the patients with cyclic neutropenia (n = 145) or idiopathic neutropenia (n = 238) showed signs of leukemic progression. A recent update on the treatment of patients with cyclic neutropenia by Dale has confirmed that when used alone, the risk of long‐term G‐CSF use is negligible in this patient population 32. These observations suggest that long‐term treatment with G‐CSF is not leukemogenic in patients with cyclic neutropenia or idiopathic neutropenia but might contribute to leukemic progression in patients with severe congenital neutropenia.

This latter notion is further supported by the discovery that acquired mutations in the gene encoding the G‐CSF receptor (CSF3R) are found in 30%–35% of patients with severe congenital neutropenia and has been reported in up to 78% of patients with congenital neutropenia who develop AML 30, 33. Because G‐CSFR and multiple other kinases associated with myelodysplasia and/or leukemia (e.g., FLT3, JAK2, KIT) share many signaling pathways, it is possible that mutant G‐CSFR provides a signal necessary for malignant transformation 34, 35, 36. These patients may require greater amounts of G‐CSF than those without mutation to maintain adequate white blood counts, and studies have shown a correlation between dose intensity of G‐CSF use in this patient population and AML risk 37. Whether the increased leukemic risk is a result of the increased G‐CSF dose these patients required or their stem cells were so damaged they required higher doses of G‐CSF and were inherently more likely to develop AML irrespective of G‐CSF dose could not be determined 37. The observation that G‐CSFR mutations are frequently found in chronic neutrophilic leukemia and treatment with ruxolitinib, a JAK2 inhibitor, can be effective treatment for this disease also supports a possible role for G‐CSFR mutations in the development of myeloproliferative disorders 38, 39.

Observations of a possible G‐CSF association with myelodysplasia and AML after breast cancer chemotherapy continue to raise concern. Smith, in the NSABP adjuvant trials, observed a higher risk of MDS/AML in doxorubicin/cyclophosphamide arms that mandated G‐CSF than in those that did not 11. This observation is complicated by the fact that those regimens contained two to four times as much cyclophosphamide as the arms for which G‐CSF was not included. There was a correlation between cumulative cyclophosphamide dose and risk of MDS/AML, although the actual amount of G‐CSF delivered in each arm is not reported. In a large meta‐analysis reported by Lyman et al., of the studies reporting blood disorders after treatment, a t‐MDS/AML RR of 1.92 (95% CI: 1.19–3.07) was observed in patients who received G‐CSF versus those who did not 40. An analysis from the SEER‐Medicare database by Hershman et al. 41 of 5,510 patients with breast cancer treated with chemotherapy, 16% of whom received G‐CSF or GM‐CSF, reported a 2.14 hazard ratio for t‐MDS/AML development for the G‐CSF/GM‐CSF‐treated group. In the SEER/Medicare study by Calip et al. of 56,000 chemotherapy‐treated patients with breast cancer, an increased dose–response‐related t‐MDS/AML risk was observed in G‐CSF‐treated patients, up to RR of 1.91 in the highest risk group 19.

Interestingly, no increased risk was observed in the Calip study in those receiving pegylated G‐CSF (pegfilgrastim, peg‐G‐CSF). A strikingly similar dichotomy has been reported for G‐CSF and peg‐G‐CSF in a cohort of more than 18,000 patients with non‐Hodgkin lymphoma treated primarily with rituximab plus multiple chemotherapy regimens including cyclophosphamide, vincristine, and prednisone with or without an anthracycline 42. G‐CSF and peg‐G‐CSF are thought to have the same mechanism of action but quite different pharmacokinetics. Therefore, if there is a true difference in leukemogenic potential between the two drugs, the mechanism is not clear at this time, unless peak drug blood levels or area‐under‐the‐curve assessments are relevant to the degree of G‐CSF exposure to susceptible patients 43.

Without sufficiently detailed dosage and chemotherapy drug schedule information, it is not possible to (a) separate the leukemogenic effect of G‐CSF from the effect of more intense potentially leukemogenic chemotherapy regimens used or (b) definitively identify a differential leukemogenic potential between G‐CSF and peg‐G‐CSF. The ability to treat patients for neutropenia has allowed for more intensive treatment regimens, when required, resulting in improved breast cancer survival, demonstrating treatment benefit in the presence of a small t‐MDS/AML risk, which is still an uncommon occurrence even in the highest risk groups 44.

HRD and t‐MDS/AML

BRCA and other mutations related to homologous DNA repair have been under scrutiny with regard to post–breast cancer treatment leukemia. Churpek has found indications of an association between both germline and somatic mutations of these genes and treatment‐related leukemia in breast cancer survivors 45. These mutations produce a defect in homologous DNA repair (homologous repair deficiency [HRD]) resulting in an increased risk of a variety of cancers, including breast, ovarian, prostate, and pancreatic cancer. Other mutations in the BRCA region, such as PALB2, ATM, CHEK2, RAD51D, and BARD1, have been shown to be important in DNA repair as well 46. BRCA pathway defects increase the risk for a subset of lymphomas and leukemias that are probably associated with genetic abnormalities 46, 47. A small study by Iqbal et al. suggested an association of BRCA2 mutation with an increased risk of leukemia 48, 49. However, BRCA pathway deficits do not themselves cause the underlying gene rearrangements but rather allow more mistakes in double strand break repair, increasing the numbers of cells with mistakes and resulting in the survival of increased numbers of abnormal cells 50, 51.

All of these genes are intimately associated with and function with the Fanconi anemia (FA) gene complex. Fanconi anemia is an inherited disorder involving a large number of genes involved in DNA repair in which cells cannot properly repair a particularly deleterious type of DNA damage known as interstrand crosslinks (ICLs). This defect in DNA repair results in genomic instability 52. ICLs form after exposure to radiation and DNA alkylating agents used as cancer chemotherapy, and from interactions with endogenous aldehydes formed as products of lipid peroxidation or exogenously derived aldehydes formed following alcohol consumption 52. Patients with FA have been estimated to have a 600‐fold and 700‐fold greater risk than the general population for developing MDS and AML, respectively, and up to 15% of patients with FA will develop AML 53, 54. Lymphoid malignancies including acute lymphoblastic leukemia and Burkitt lymphoma are also seen, although they are much less common 54.

Partner and localizer of BRCA2 (PALB2 ‐ Fanconi anemia complementation group N‐FANCN) is an important component of the Fanconi anemia DNA repair pathway and is involved in linking BRCA1 and BRCA2 to form a BRCA complex that is essential in preventing cells from accumulating DNA damage 46, 50. Patients with Fanconi anemia who carry PALB2 and BRCA2 mutations have an 800‐fold increased risk of MDS or AML when both alleles are inherited with a cumulative incidence of leukemia of 80% by age 10 55. Mutations in RAD51, which is also related to the HRD gene complex, have also been reported to cause an increased risk of leukemia 46, 56. The available data regarding the relationship of a number of HRD mutations to AML and other types of leukemia and lymphoma suggest a possible increased leukemogenic risk when these mutations are present 46, 56.

BRCA1 deficiency is strongly associated with both de novo and therapy‐related AML. Thirty‐six percent (32/112) of primary AML tumors and 75% (16/21) of therapy‐related AML tumors have reduced BRCA1 gene expression, and studies have noted increases in this activity correlating with successful response to AML therapy, suggesting that deficient BRCA activity is important in allowing malignant clones to grow and survive 57, 58. Loss of BRCA activity can occur by somatic or germline mutation, loss of heterozygosity, copy number alterations (regional or whole chromosome alterations or deletions), or promoter methylation with gene silencing 59, 60. Multiple investigators have demonstrated that BRCA gene silencing via hypermethylation occurs not only in AML but also in multiple solid tumors, including breast, ovary, and prostate cancer 56, 57, 59, 60, 61, 62, 63, 64. In addition, hypermethylation has been reported as a consequence of both cytotoxic chemotherapy and radiation 58.

BRCA mutated tumors have a particular sensitivity to platinum chemotherapy as well as PARP inhibitors 65, 66, 67. Whether this applies to mechanisms of HRD caused by other gene mutations is somewhat unclear at this time. Some reports suggest the mechanism of BRCA function loss is important 58, 67, 68, 69, 70, but much more study is required to evaluate this issue.

A variety of assays have been developed in an attempt to further define the properties of HRD tumor cells, which has been termed “BRCAness” or “BRCA‐like” in order to select patients most appropriate for different types of treatment 59, 69, 71, 72, 73, 74, 75, 76, 77, 78, 79, as it appears not all of the abnormalities carry the same implications for treatment. There are few data to evaluate the implications of the various types of “BRCAness” for leukemogenic risk 58. In addition, the problem is compounded by the known genomic and epigenomic instability of tumors over time 80, 81, 82, 83.

Although it is clear that BRCA mutations predict sensitivity to platinum agents and PARPs, it is unclear how other HRD types can be used to determine drug therapy as well as how they may contribute to the risk of t‐MDS/AML in patients with breast cancer. However, the experience in Fanconi anemia, the sensitivity of patients with FA to alkylating agents and radiation, and the finding of reduced BRCA activity without BRCA mutation in patients with AML all suggest the possibility that other gene mutations noted above (PALB2, ATM, and CHEK2, RAD51D, and BARD1) may contribute to leukemogenesis. Although many of these gene mutations are too infrequent to allow development of large data sets, the increasing use of genomic profiling may be helpful in identifying significant gene mutations as well as important “BRCA‐like” conditions relevant to leukemogenesis.

PARP Inhibitors and t‐MDS/AML

Interestingly, PARP inhibitors, which counteract the effects of the BRCA mutations and are currently in clinical trial for the treatment of AML 68, may have a role in the development of AML. These drugs come with a Food and Drug Administration warning of potential for t‐MDS/AML resulting from observed adverse events during clinical trials of PARP inhibitors in advanced BRCA mutated cancer 81. Currently, it is estimated that the incidence in patients treated with PARP inhibitors ranges from 0.5% to 2% 84. It is possible that PARP inhibition allows some BRCA mutated cells that would otherwise have died from their mutations to survive with nonlethal mutations that lead to leukemia, as suggested for G‐CSF 3, 26.

Because most of the patients treated with PARP inhibitors were selected precisely because they had BRCA mutations, and virtually all of them had previously received leukemogenic drugs (presumably with varying amounts of G‐CSF), the interactive effects of prior treatment, BRCA mutation, G‐CSF use, and PARP treatment is unclear at this time. Given the parallel observations of both leukemogenesis and effective antileukemic treatment with anthracyclines, however, it is possible that leukemogenesis and antileukemic treatment efficacy are two separate phenomena that coexist for PARP inhibitors as well.

The goal for our patients is to maximize clinical treatment benefit and minimize toxicity risk. The perfect storm for this dilemma is triple‐negative breast cancer. Triple‐negative patients are often young, with a higher incidence of BRCA mutations or other homologous repair deficiencies, and receive alkylating agents combined with anthracyclines and increasingly platinum agents as well. These myelosuppressive regimens thus necessitate the use of G‐CSF. Additionally, many receive short‐acting G‐CSF rather than peg‐G‐CSF for reimbursement reasons. In view of the studies cited above, the exposure of patients to additional risk may perhaps be mitigated by the use of peg‐G‐CSF. Many of these patients receive postoperative radiation therapy as well, adding to leukemogenic potential. Finally, adjuvant PARP inhibitor trials in patients with breast cancer with BRCA mutations and residual disease after neoadjuvant chemotherapy are under way (NCT02032823), potentially creating another potential leukemogenic risk factor.

Evidence from these studies taken together suggest a likelihood that maximal leukemogenic risk requires multiple insults from cancer therapeutic agents (possibly dose‐dependent), radiation, and G‐CSF combined with an age or genetic relationship that could amplify defects in DNA repair, creating a biochemical cascade effect. Multiple efforts to reduce the amount of treatment patients receive are under way. The use of genomic tumor profiling has reduced the use of chemotherapy with leukemogenic potential in the adjuvant setting 85, 86. Additionally, studies of newer agents in trials in addition to the BCIRG 006 study have shown success in removing doxorubicin/cyclophosphamide from the neoadjuvant treatment of many patients with human epidermal growth receptor 2/neu‐positive breast cancer. Early results in neoadjuvant trials in triple‐negative breast cancer from the Keynote 522 trial (NCT03036488) have recently shown improvement in outcomes with the addition of programmed cell death protein 1 blockers to standard therapy, raising the question of whether it might be possible to curtail the use of alkylating agents and anthracyclines in this setting. Hughes has found evidence that for women above the age of 70 with low‐stage breast cancer, radiation therapy may be safely omitted if adjuvant hormone therapy is used 87. In addition, it is possible that triple‐negative breast cancer may be treated with less aggressive therapy in older than in younger patients with similar outcomes 88.

Although the absolute risk of t‐MDS/AML is small, estimated to be in the 0.25%–2% range depending on the setting, it remains a disease with a usually fatal outcome 22. The recent approval of a new treatment option with dual‐drug liposomal encapsulation of daunorubicin and cytosine arabinoside shows some improvement over other therapies in older patients 89. Although this may be encouraging, better treatment options for treatment‐related myelodysplasia need to be identified 90. It will be important to further characterize possible risks associated with G‐CSF, peg‐G‐CSF, and PARP inhibitors in patients with and without homologous DNA repair deficiency mutations as well as other forms of “BRCAness” and to develop safer effective treatments for breast cancer.

Conclusion

Improved risk assessment strategies, pharmacovigilance, and safer therapies for all malignancies, especially for younger patients with long expected survival times, are critical as cancer treatment continues to evolve. The risk of secondary cancer in the form of t‐MDS/AML when possible needs to be factored into treatment decisions. Current evidence from randomized clinical trial reviews, population‐based cohort studies, and emerging tumor and genomic studies of chemotherapeutic agents, drug dosage, radiation, and neutropenia treatment in relation to adverse myelodysplastic outcomes after breast cancer treatment can inform treatment choice considering patient survival and disease threat elimination are paramount.

Author Contributions

Conception/design: Henry G. Kaplan, Gregory S. Calip, Judith A. Malmgren

Provision of study material or patients: Henry G. Kaplan, Gregory S. Calip, Judith A. Malmgren

Collection and/or assembly of data: Henry G. Kaplan, Gregory S. Calip, Judith A. Malmgren

Data analysis and interpretation: Henry G. Kaplan, Gregory S. Calip, Judith A. Malmgren

Manuscript writing: Henry G. Kaplan, Gregory S. Calip, Judith A. Malmgren

Final approval of manuscript: Henry G. Kaplan, Gregory S. Calip, Judith A. Malmgren

Disclosures

The authors indicated no financial relationships.