Secondary hematopoietic malignancies in survivors of childhood cancer

An analysis of 111 cases from the Surveillance, Epidemiology, and End Result-9 registry

Authors


Abstract

BACKGROUND:

Studying secondary hematological malignancies in a large cohort of patients can help predict risks and trends associated with current therapies.

METHODS:

The authors analyzed data from the Surveillance, Epidemiology, and End Resultsecondary 9 (SEER-9) database on patients with a primary malignancy (diagnosed before the age of 20 years) between 1973 and 2005 who developed a secondary hematological malignancy. Primary cancer and histological subtype, incidence, risk factors, outcomes, and changes in risk patterns of secondary hematological malignancies were analyzed for 1973 to 1985, 1986 to 1995, and 1996 to 2005. Standardized incidence ratios (SIRs) of observed to expected cancers were calculated.

RESULTS:

Of 34,867 patients with a histology-confirmed primary malignancy, 111 developed secondary hematological malignancies (median, 44 months). Lymphoma was the commonest primary cancer (n = 47). The main histological subtype of secondary hematological malignancy was acute myeloid leukemia (AML) (49%), which had the shortest median latency time and the worst 5-year survival (18% ± 5.3%; P = .044). Secondary Hodgkin lymphoma had the best 5-year survival (83% ± 15%). The 5-year overall survival for patients with secondary hematological malignancies was 31% ± 4.7%. The risk of secondary AML steadily increased from 1986 to 2005, whereas SIRs for acute lymphoblastic leukemia did not change over time. Non-Hodgkin lymphoma, the second most common secondary hematological malignancy, occurred at a median of 112 months, and its risk steadily increased over time periods.

CONCLUSIONS:

Childhood cancer survivors are at increased risk of developing secondary hematological malignancies, particularly secondary AML. This risk has continued to rise even in recent years, emphasizing the need to study other factors contributing to secondary hematological malignancies and closely monitor these patients. Cancer 2010. © 2010 American Cancer Society.

Advances in contemporary modalities for treating childhood cancer have resulted in improved survival rates in this population. However, childhood cancer survivors are at increased risk of developing secondary malignancies. Thus, it is essential to study the host-, environment-, and treatment-related factors predisposing this growing population of survivors to the development of secondary malignancies.1, 2 Secondary hematological malignancies are among the most common types of secondary malignancies that are frequently associated with serious long-term consequences and poor outcome.3, 4 One example of treatment-related secondary hematological malignancy is the monoblastic or myelomonocytic subtype that develops after a short median latency period after the use of topoisomerase II inhibitors.4, 5 Another secondary hematological malignancy is late-onset acute myeloid leukemia (AML), which is a dose-dependent late effect of treatment with alkylating agents and is classically preceded by myelodysplastic syndrome (MDS) with loss or deletion of chromosome 5 or 7.6, 7 Childhood cancer survivors can also develop AML or MDS after treatment with growth factors (GFs), in combination with etoposide, anthracyclines, and radiotherapy. However, it is difficult to distinguish the contribution of GF versus intensified therapy and the effect of intensity8 or cumulative doses of chemotherapy9, 10 in the development of secondary hematological malignancies.

Radiotherapy can also increase the risk of AML, chronic myeloid leukemia, and acute lymphoblastic leukemia (ALL), which is highest 5 to 9 years after exposure to radiation. However, this risk is approximately 2× lower than that of chemotherapy, and is a function of radiation dose to the active bone marrow, dose rate, and percentage of exposed marrow.11 Moreover, genetic and specific host factors can be major predisposing factors to leukemogenesis. For example, the presence of certain polymorphisms in detoxifying enzymes is associated with increased risk of secondary AML.12, 13

Most previous studies have analyzed all types of secondary malignancies,14, 15 but there are few comprehensive large-scale studies specifically on secondary hematological malignancies. Furthermore, most reports on secondary hematological malignancies focus on second leukemias only.6, 16, 17 In this study, we used data from the Surveillance, Epidemiology, and End Resultsecondary 9 (SEER-9) registry to evaluate incidence, types, and outcomes of secondary hematological malignancies in a large cohort of childhood cancer survivors and to determine changes in trends of secondary hematological malignancy histology over different time periods.

MATERIALS AND METHODS

Data Source and Study Population

Data regarding secondary hematological malignancies were obtained from the SEER-9 registry (http://seer.cancer.gov/data/). The SEER Program of the National Cancer Institute is an authoritative source of data on cancer incidence and survival in the United States and currently covers approximately 26% of the US population. We selected the SEER-9 registry because it offers the longest follow-up (1973-2005) for registered patients. We used the Multiple Primaries-Standardized Incidence Ratios session of the SEER*Stat 6.4.4 program to generate a matrix of all patients who had a primary pediatric malignancy (before the age of 20 years) and later developed a secondary hematological malignancy. We studied data from all patients diagnosed with a primary malignancy and secondary hematological malignancy between January 1973 and December 2005. The Multiple Primaries-Standardized Incidence Ratios session was also used to calculate excess risk of developing secondary malignancies. We studied patient characteristics; primary cancer type; and histological subtype, incidence, risk factors, outcomes, and changes in risk patterns of secondary hematological malignancies.

Data Analysis

The resulting matrix from SEER*Stat was transferred to MedCalc for Windows, version 10.0.1.0 (MedCalc Software, Mariakerke, Belgium) to perform statistical calculations.

We used the multiple primary session of the SEER*Stat software to identify childhood cancer survivors with a secondary hematological malignancy.

The Kaplan-Meier method was used to calculate survival estimates, and the log-rank test was used to compare survival curves. All-cause mortality was considered an endpoint.

On the basis of when the primary malignancy was diagnosed, we divided the study into 3 periods: 1973 to 1985, 1986 to 1995, and 1996 to 2005. These time periods were best suited to study changes in risk patterns for secondary hematological malignancies with the introduction of new agents that increased the risk of secondary hematological malignancies and with the start of risk-adapted treatment in the United States. We evaluated the change in standardized incidence ratios (SIRs; observed to expected ratios based on age- and sex-specific population-based incidence rates on the registry) for survivors to develop a secondary hematological malignancy. All SIRs are reported with their 95% confidence interval (CI). To compare the characteristics of patients with secondary hematological malignancies and of the remaining registered patients; the chi-square test was performed using OpenEpi software (http://www.openepi.com/Menu/OpenEpiMenu.htm).

RESULTS

Patient Characteristics

The SEER-9 database has records of 34,867 patients (46% female) diagnosed from January 1973 to December 2005 with a primary cancer before the age of 20 years. Of these, we identified 111 records of patients (43% female; median follow-up for living patients, 10 years; range of follow-up, 0-17.4 years) subsequently diagnosed with a secondary hematological malignancy (Table 1). Their median age of primary cancer diagnosis was 14 years (range, 0-19 years). Similar to other patients registered in SEER-9, the majority (77%) with a secondary hematological malignancy were white. The most common primary diagnosis in patients developing a secondary hematological malignancy was lymphoma (n = 47): Hodgkin lymphoma (HL) in 30 and non-Hodgkin lymphoma (NHL) in 17 patients. Compared with other patients in SEER-9, those with a secondary hematological malignancy were significantly more likely to be older than 10 years at diagnosis (P < .001), to be diagnosed with lymphoma (P < .001), and to have received radiotherapy to treat the primary cancer (P = .005). Secondary hematological malignancies developed at a median of 44 months (range, 35-63 months). Table 2 summarizes time to onset of secondary hematological malignancy. The histological subtypes of secondary hematological malignancies were secondary AML in 54, secondary NHL in 27, secondary ALL in 16, secondary HL in 6, and other histologies in 8 patients. The 5-year survival of patients with a secondary hematological malignancy was 31% ± 4.7%. The 5-year survival was worst for patients with secondary AML (18% ± 5.3%) and significantly better (P = .044) for patients with secondary ALL (38% ± 14%), secondary NHL (49% ± 9.9%), and secondary HL (83% ± 15%) (Fig. 1).

Figure 1.

Kaplan-Meier survival curves for patients with secondary hematopoietic malignancies are shown according to type, considering all-cause mortality after diagnosis of secondary hematological malignancy as the endpoint. Curves were compared using the log-rank test. ALL indicates acute lymphoblastic leukemia; AML, acute myeloid leukemia; HL, Hodgkin lymphoma; NHL, non-Hodgkin lymphoma.

Table 1. Patient Characteristics and Use of Radiotherapy in 111 Patients (Aged <20 Years) Diagnosed With a Primary Cancer From 1973 Through 2005 Who Subsequently Developed an s-HM
VariablePatients Who Developed s-HMPatients on SEER-9 RegistryPa
No.%No.%
  • s-HM indicates secondary hematologic malignancy; SEER-9, Surveillance, Epidemiology, and End Results-9; CNS, central nervous system.

  • a

    Chi-square test was used to compare values and calculate P values.

Total111 34,867  
Sex
 Female484316,07946.55
 Male635718,78854 
Race
 White867728,34281.31
 Black1110354810 
 Other141329778.5 
Index record
 Age at diagnosis, y
  Birth-9312816,59448<.001
  10-19807218,27352 
 Primary diagnosis
  Leukemia1413878825<.001
  Lymphoma4742550716 
  CNS tumors1211579217 
  Solid tumors383414,69042 
 Radiation
  Beam484310,79431.005
  None/others/unknown635724,07369 
 Time period of diagnosis
  1973-1985403612,38336.55
  1986-1995383410,58130 
  1996-2005333011,90334 
Table 2. Time of Onset of s-HM in Patients (1973-2005)
TypeNo.Duration to Develop s-HM, mo
Median95% CI
  • s-HM indicates secondary hematologic malignancy; CI, confidence intervals for observed/expected ratios; ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; HL, Hodgkin lymphoma; NHL, non-Hodgkin lymphoma.

  • a

    Includes mast cell leukemia (n = 1); acute leukemia, not otherwise specified (n = 1); chronic myeloid leukemia (n = 5); and myeloid sarcoma (n = 1).

All s-HMs1114435-63
ALL1671.530-142
AML543630-48
HL64219-81
NHL2711231-208
Othersa8350.28-146

Risk of Secondary Malignancy

The cumulative incidence for developing secondary hematological malignancy at 5 and 10 years were 0.2% ± 0.03% and 0.4% ± 0.04%, respectively (Fig. 2A). SIRs showed a significantly higher incidence of secondary hematological malignancies in pediatric cancer survivors than others in SEER-9 (observed/expected ratio, 3.47; 95% CI, 2.86-4.18), with an excess risk of 2.26 per 10,000 (Table 3). Specifically, there was an increased incidence of AML (excess risk, 1.46 per 10,000), ALL (excess risk, 0.3 per 10,000), and NHL (excess risk 0.51 per 10,000). The SIR for developing a secondary hematological malignancy was higher in patients diagnosed at age 10 to 20 years and those who received radiotherapy. Table 4 shows SIRs for secondary hematological malignancy according to primary cancer type. Among survivors of hematopoietic malignancies, the highest SIR was seen in primary NHL (observed/expected ratio, 9.36; 95% CI, 5.11-15.71), whereas patients with primary ALL had a significantly higher incidence of secondary AML (observed/expected ratio, 11.20; 95% CI, 4.09-24.38; Table 4). Among patients with primary solid tumors, an increased incidence of secondary hematological malignancy was noted in patients with the Ewing sarcoma family of tumors (ESFT; observed/expected ratio, 11.36; 95% CI, 4.15-21.85), followed by embryonal tumors in the central nervous system (CNS; observed/expected ratio, 11.09; 95% CI, 4.77-21.85), osteosarcoma (observed/expected ratio, 6.81; 95% CI, 2.49-14.81), and retinoblastoma (observed/expected ratio, 6.51; 95% CI, 1.75-16.67). Interestingly, retinoblastoma was associated specifically with an increased incidence of secondary ALL (observed/expected ratio, 16.14; 95% CI, 4.34-41.33) but not of secondary AML, secondary HL, or secondary NHL.

Figure 2.

(A) Cumulative incidence of developing a second hematologic malignancy (s-HM) in patients with a diagnosis of pediatric cancer and observed to expected (O/E) ratios for developing (B) secondary hematological malignancies at different latency periods after a cancer diagnosis are shown (n=111); inadequate duration to calculate observed/expected after a 59-month latency in the last study period. (C) Secondary hematological malignancies during the 3 time periods are shown. (D) Secondary acute myeloid leukemia (s-AML) is shown according to primary diagnosis in the 3 time periods (n = 54). CNS indicates central nervous system.

Table 3. Standardized Incidence (O/E Ratios and 95% CIs) Values for Development of s-HM According to Patient Characteristics and Treatment of Primary Cancer
 s-HMs-ALLs-AMLs-HLs-NHL
  • O/E indicates observed/expected; CI, confidence interval; s-HM, secondary hematologic malignancy; s-ALL, secondary acute lymphoblastic leukemia; s-AML, secondary acute myeloid leukemia; s-HL, secondary Hodgkin lymphoma; s-NHL, secondary non-Hodgkin lymphoma.

  • a

    Statistically significant (P < .05).

  • b

    The 95% CI values are shown in parentheses.

All pediatric malignancies3.47a (2.86-4.18)b2.91a (1.66-4.72)15.69a (11.82-20.42)0.54 (0.2-1.17)2.83a (1.88-4.09)
Excess risk per 10,0002.260.301.46−0.150.51
Beam radiation
 Received4.15a (3.05-5.52)4.66a (2.00-9.17)17.46a (10.80-26.69)0.25 (0-1.37)4.20a (2.35-6.92)
 Not received2.94a (2.23-3.80)2.19 (0.94-4.32)13.83a (9.33-19.75)0.46 (0.09-1.32)2.02a (1.04-3.53)
Sex
 Male3.45a (2.66-4.41)2.96a (1.42-5.44)18.12a (12.47-25.45)0.36 (0.04-1.29)2.19a (1.19-3.67)
 Female3.49a (2.58-4.63)2.82a (1.03-6.14)13.06 (8.18-19.77)0.72 (0.19-1.84)4.01a (2.19-6.74)
Age ,y
 <102.72a (1.85-3.86)2.80a (1.40-5.01)10.13a (5.53-16.99)0 (0-1.18)1.18 (0.24-3.46)
 10-203.88a (3.08-4.83)3.17a (1.02-7.40)19.31 (13.85-26.19)0.74 (0.27-1.62)3.40a (2.20-5.02)
Table 4. Standardized Incidence (O/E Ratios and 95% CI) Values for Development of s-HM According to Primary Cancer Type
Primary Cancer TypeAll s-HMs-ALLs-AMLs-HLs-NHL
  • O/E indicates observed/expected; CI, confidence interval; s-HM, secondary hematologic malignancy; s-ALL, secondary acute lymphoblastic leukemia; s-AML, secondary acute myeloid leukemia; s-HL, secondary Hodgkin lymphoma; s-NHL, secondary non-Hodgkin lymphoma; CNS, central nervous system; PNET, peripheral neuroectodermal tumor; RMS, rhabdomyosarcoma; NRSTS, nonrhabdomyosarcoma soft tissue sarcoma; GCT, germ cell tumor.

  • a

    Statistically significant (P < .05).

  • b

    The 95% CI values are shown in parentheses.

All pediatric malignancies3.47a (2.86-4.18)b2.91a (1.66-4.72)15.69a (11.82-20.42)0.54 (0.2-1.17)2.83a (1.88-4.09)
Type
 Precursor ALL1.74 (0.75-3.44)0 (0-2.81)11.20a (4.09-24.38)0 (0-2.62)0.88 (0.01-4.91)
 AML6.72a (1.81-17.20)14.24a (1.60-51.42)0 (0-50.99)0 (0-18.98)12.76a (1.43-46.09)
 HL6.21a (4.19-8.86)2.84 (0.04-15.81)31.59a (17.67-52.10)0 (0-1.98)7.46a (3.97-12.76)
 NHL9.36a (5.11-15.71)0 (0-18.49)45.15a (18.09-93.04)5.52a (1.11-16.12)6.07a (1.22-17.74)
CNS (all tumors)3.24a (1.81-5.34)4.41 (1.19-11.29)11.79 (4.31-25.67)0 (0-2.31)2.23 (0.45-6.50)
CNS (embryonal tumors)11.09a (4.77-21.85)21.87a (5.89-56.00)35.99a (7.23-105.14)0 (0-16.07)5.25 (0.07-29.19)
Retinoblastoma6.51a (1.75-16.67)16.14a (4.34-41.33)0 (0-46.28)0 (0-25.69)0 (0-30.34)
Neuroblastoma0.90 (0.01-5.02)0 (0-7.77)7.27 (0.09-40.43)0 (0-15.07)0 (0-17.05)
Wilms tumor2.46 (0.49-7.17)2.54 (0.03-13.92)13.51a (1.52-48.77)0 (0-10.28)0 (0-13.35)
Hepatoblastoma7.71 (0.10-42.88)0 (0-58.01)0 (0-222.91)0 (0-158.72)43.29 (0.57-240.87)
Osteosarcoma6.81a (2.49-14.81)11.12 (0.15-61.90)52.77a (17.01-123.14)0 (0-11.09)0 (0-12.35)
PNET/Ewing sarcoma11.36a (4.15-24.72)0 (0-51.36)71.17a (19.15-182.21)4.82 (0.06-26.80)6.25 (0.08-34.77)
RMS2.61 (0.29-9.42)0 (0-21.54)23.42a (2.63-84.57)0 (0-14.87)0 (0-16.68)
NRSTS1.70 (0.46-4.35)0 (0-10.30)15.80a (4.25-40.44)0 (0-4.46)0 (0-4.86)
GCT1.64 (0.53-3.83)3.33 (0.04-18.52)9.76a (1.96-28.51)0 (0-3.42)0.91 (0.01-5.6)

The risk of developing secondary hematological malignancy peaked 1 to 5 years after primary diagnosis (Fig. 2B). Therefore, calculating the total number of patients developing cancer in different periods was not sufficient to compare the risk of their developing secondary hematological malignancy. Although the percentage of patients developing secondary hematological malignancy seemed similar (Table 1), the SIR constantly increased through the 3 study periods (Fig. 2C). Patients with a primary leukemia showed an interesting pattern—the risk of secondary hematological malignancy was low in the first 2 study periods but increased sharply in the last period.

Secondary AML

Secondary AML accounted for 49% of cases of secondary hematological malignancy and had the shortest median time to onset after diagnosis of primary malignancy (36 months). The histological subtype of secondary AML was not defined in the majority of patients (n = 41), except for patients with acute monocytic leukemia (n = 6) and acute myelomonocytic leukemia (n = 5). Survival of patients with secondary AML was significantly poorer (P = .044) than other secondary hematological malignancies (Fig. 1).

We analyzed observed/expected ratios of developing secondary AML according to primary diagnosis in each time period (Fig. 2D). For malignancies diagnosed from 1973 to 1985, the highest increase in incidence of secondary AML was observed in survivors with primary lymphoma (observed/expected ratio, 19.45; 95% CI, 8.38-38.33); SIRs increased in the following 2 periods, particularly for children with a primary lymphoma. During 1996 to 2005, the observed/expected ratio for secondary AML was 75.69 (95% CI, 27.64-164.75), 43.88 (95% CI, 20.02-83.30), 40.57 (95% CI, 10.91-103.86), and 25.95 (95% CI, 2.91-93.67) for children treated for lymphoma, primary solid tumors, primary leukemia, and CNS tumors, respectively. Among primary solid tumors, those treated for ESFT had the highest excess risk of secondary AML (observed/expected ratio, 71.17; 95% CI, 19.15-182.21).

Secondary ALL

Sixteen cases of ALL were registered as secondary malignancies at a median time of 71.5 months. All cases were reported as precursor cell leukemias (12 not specified and 4 precursor B cell). Unlike AML, SIRs for developing secondary ALL after treatment for a primary pediatric malignancy did not change substantially over time periods: 2.71 (95% CI, 0.99-5.89) in patients diagnosed from 1973 to 1985, 2.92 (95% CI, 1.07-6.36) in patients diagnosed from 1986 to 1995, and 3.24 (95% CI, 0.87-8.30) in those diagnosed from 1996 to 2005. Patients with primary solid tumors (observed/expected ratio, 3.74; 95% CI, 1.50-7.70) and CNS tumors (observed/expected ratio, 3.52; 95% CI, 1.13-8.22) were at significantly greater risk of developing secondary ALL. In this diagnostic group, patients most at risk were those with a primary diagnosis of CNS embryonal tumors (observed/expected ratio, 21.87; 95% CI, 5.89-56.00), followed by those with retinoblastoma (RB; observed/expected ratio, 16.14; 95% CI, 4.34-41.33). The 5-year survival rates of patients with secondary ALL were much lower (23% ± 20%) than those with a primary ALL; 5-year survival of patients with ALL (5-30 years old) was 66% ± 0.6%, even from 1996 to 2005. Only data from 5- to 30-year-old patients were selected, as secondary ALL was reported in this age group.

Secondary Lymphomas

Secondary lymphomas developed in 33 patients (observed/expected ratio, 1.58; 95% CI, 1.09-2.22): secondary NHL in 27 patients (observed/expected ratio, 2.75; 95% CI, 1.81-4.0) and secondary HL in 6 (observed/expected ratio, 0.54, 95% CI, 0.20-1.18). Secondary HL developed at a median time of 42 months (range, 19-81 months) from diagnosis of primary cancer (median follow-up,7.8 years). The incidence of childhood cancer survivors developing secondary HL was lower than that of the general population for all primary diagnoses except NHL. The incidence of developing secondary HL after a primary NHL was increased (observed/expected ratio, 5.52; 95% CI, 1.11-16.12). The 5-year survival of patients with secondary HL was 86% ± 5.8%.

Conversely, the median time to develop secondary NHL was 112 months (range, 31-208 months) from diagnosis of primary cancer. The highest SIR for secondary NHL was observed in patients with a primary diagnosis of AML. The 5-year survival of patients with secondary NHL (44% ± 8.1%) was poorer than those with secondary HL. The SIR for developing secondary NHL was lowest in patients diagnosed from 1973 to 1985. The risk increased steadily in patients diagnosed from 1986 to 1995 (observed/expected ratio, 7.5; 95% CI, 2.03-19.33) and rose further in patients diagnosed from 1996 to 2005 (observed/expected ratio, 19.39; 95% CI, 4.01-58.30).

DISCUSSION

The increased risk of developing secondary hematological malignancies in childhood cancer survivors is a serious concern, as the use of aggressive contemporary therapy creates an ever-growing population of survivors. Our analysis of SEER-9 registry data shows increased risk rates of childhood cancer survivors to develop secondary hematological malignancy, which is comparable to previous reports.2, 18, 19 It also demonstrates that the incidence, risk trends, and outcome of childhood cancer survivors who develop secondary hematological malignancies vary by primary cancer type and treatment periods.

Between 1973 and 1985, lymphoma survivors had a higher SIR of developing a secondary hematological malignancy, especially secondary AML, than survivors of other primary malignancies. The risk of secondary hematological malignancy rose steadily in successive time periods. In particular, the risk of developing secondary AML increased markedly in the last study period. Unlike secondary AML, the SIR for secondary ALL in our study remained steady over time periods, which may suggest that the factors influencing the development of secondary ALL did not change over time. In contrast, the SIR for secondary NHL was lowest in patients diagnosed from 1973 to 1985, particularly primary lymphomas, and rose constantly in the following 2 periods. This may reflect improvements in the diagnostic modalities, including flow cytometry and cytogenetics.

Among survivors of hematopoietic malignancies, patients with primary NHL had the highest excess risk of secondary hematological malignancy, as shown previously.1 Among solid tumors survivors, those with ESFT had the highest excess risk of developing a secondary hematological malignancy, consistent with previous reports.

In our study, 54 patients with secondary AML (49% of patients with secondary hematological malignancy) had an excess risk of 1.46 per 10,000, with the shortest median time to onset (36 months). In this cohort, survivors of lymphomas most frequently developed secondary AML, as also shown by Kaldor et al.20 The relative risk of secondary AML increases 1 to 5 years after therapy for primary cancer, remains elevated through the first 15 years of follow-up, and increases with increasing calendar year for the latency interval between 1 and 5 years.1 The 5-year survival of secondary AML patients in our cohort was 18% ± 5.3% (P = .044), consistent with earlier reports.

Hijiya et al.21 reported 37 cases of secondary AML among 123 patients who developed secondary malignancies after treatment for childhood ALL; the 5-year survival of the group with secondary AML was poorer (18%) than of those with other types of secondary hematological malignancies. Another study by Hijiya et al.17 comparing outcomes of ALL, AML, and secondary AML using SEER data showed significantly lower 5-year survival (23.7%; P < .001) for children with secondary AML compared with those with de novo AML (53.2%). Twenty-four patients with secondary AML in a Children's Cancer Group study (CCG 2891) had lower remission induction rates, survival, and event-free survival than patients with de novo AML.16 The findings of Hijiya et al. agree with those of others reporting poor prognosis for childhood cancer survivors with secondary AML.6, 16, 21 In contrast, the German Cooperative Group22 and the Italian Group for Adult Hematologic Diseases23 showed that outcomes of adult survivors with secondary AML are not worse than those of de novo AML when adjusted for cytogenetic features.

Secondary ALL occurred at a median time of 71.5 months and comprised 14.4% of all secondary hematological malignancies in our study, which concurs with previous reports demonstrating that secondary ALL is very rare24 and constitutes 5% to 10% of secondary leukemias.25 In 14 ALL consecutive studies reported by St. Jude, only 2 of 2304 patients with primary ALL were later diagnosed with secondary ALL.21 In most cases, these leukemias were misdiagnosed as recurrences of primary ALL, especially before molecular detection of immunoglobulin and T-cell receptor rearrangements was available. Zuna et al.24 showed that 0.5% to 1.5% of 366 cases of recurrent ALL were actually secondary ALL and proposed diagnostic criteria for secondary ALL that mandate molecular evidence of a new leukemic clone emergence, plus immunophenotypic and/or cytogenetic shift and/or fusion gene gain or loss. In a study of 101 secondary ALL cases, Shivakumar et al.26 showed a uniform time interval to diagnosis of secondary ALL across age groups, but a significantly longer interval in survivors of primary HL or neuroblastoma. Furthermore, that cohort had complex karyotypes and an overall poor survival that did not differ by age, primary diagnosis, cytogenetic subgroups, or immunophenotype.

In our study, patients with primary solid and CNS tumors were at significantly greater risk of developing secondary ALL compared with the general population. In this diagnostic group, patients with a primary diagnosis of CNS embryonal tumors, followed by those with RB, were at most risk. None of the secondary ALL patients had a primary diagnosis of ALL. The 5-year survival of secondary ALL patients was far poorer than of those with de novo childhood ALL, even in the most recent study period.

It is not known whether secondary ALL occurs secondary to the primary cancer or represents a second primary cancer, suggesting other underlying mechanisms such as genetic susceptibility. The presence of certain genetic and molecular variables might play a major role in secondary ALL development. For example, polymorphisms of several detoxification enzyme genes (such as nicotinamide adenine dinucleotide phosphate: glutathione S-transferase (GST), cytochrome P450 (CYP3A), quinone oxidoreductase (NQO1). were related to secondary leukemia.13, 27 Also, the presence of germ line mutations such as those of p53 (Li-Fraumeni syndrome) may underlie the occurrence of primary cancers and secondary ALL,28 whereas those of ataxia-telangiectasia gene are specifically associated with T-cell lineage ALL.29 Likewise, somatic PTPN11 mutations, similar to those present in Noonan syndrome, underlie the risk of lymphoid malignancies.30 Unmasking the pathogenesis of secondary ALL may be possible by detailed polymorphism and mutational analysis of these genes.

The development of secondary HL is very uncommon.31-33 We observed secondary HL in 6 patients only, mostly in primary NHL survivors. Travis et al.34 showed a 3-fold risk of secondary HL in patients with NHL; simultaneous occurrence of NHL and HL or at short intervals suggest a biologic relationship between them. Similarly, others have reported secondary HL occurrence after low-grade B-cell lineage NHL32, 33 and high-grade NHL and also after primary ALL.31, 35 In our cohort, patients with secondary HL had the best 5-year survival.

We found 27 patients diagnosed with secondary NHL; the majority had a primary diagnosis of AML, in contrast to reports of significantly high risk of secondary NHL in HL survivors.36, 37 Others have shown a high risk of secondary NHL in the first year after treatment initiation, which declines in the next 5 years, and peaks in the subsequent 10 to 14 years,37, 38 but might continue to occur in patients having long follow-ups. Green et al.39 reported a cumulative incidence of 4.62% for secondary NHL 30 years after diagnosis of nodular sclerosing HL. Van Leeuwen,37 however, observed the highest incidence of 5.9% ± 2.1% in 744 patients with HL, at a median interval of 13.3 years between HL diagnosis and secondary NHL.37

In our cohort, patients with secondary NHL fared more poorly than those with secondary HL. In a report of the German Hodgkin Lymphoma Study Group, treatment outcome of secondary NHL was influenced by time of occurrence after first diagnosis of HL, and approximately 50% of patients achieved a long-term disease-free interval.40 However, in other reports, the median time of survival for secondary NHL was only 2.5 months.41

Despite the large sample size and long follow-up of the SEER-9 registry cohort, our study has several limitations that should be considered while interpreting results. As the registry has no information on biologic features of primary malignancies, diagnosis of secondary hematological malignancies may be overestimated. Conversely, lack of molecular diagnostic methods in the earlier treatment eras could cause underestimation of secondary hematological malignancies. Moreover, the SEER data do not allow a clear understanding of the cytogenetic subsets of therapy-related leukemia; therefore, the etiologically distinct groups are lumped together in the description and analyses; this limits our understanding of the predisposing and prognostic factors as well as the outcome. Also, although the risk of secondary hematological malignancy is closely related to certain types, dosages, intensities, sequencing and combinations of chemotherapeutic exposures, and concomitant use of other agents like radiotherapy and GFs, our results cannot be definitively correlated with treatment, as SEER-9 has no such detailed information on treatment modalities known to increase the risk of secondary carcinogenesis (eg, dosage and schedules of epipodophyllotoxins, anthracyclines, alkylating agents, and GF use). Similarly, although our results indicated an association between radiation and secondary hematological malignancy, we should be careful with this interpretation, because information on radiation dosages, fields, and concomitant chemotherapeutic agents are not available in the SEER registry, and because this was only tested in a univariate model. Multivariate analysis of the risk factors for developing secondary hematological malignancy was not conducted because of the limited number of patients in each category. Another limitation is the lack of patients identified with MDS, indicating some under-reporting. (The registration of MDS in the SEER program began in the year 2001,42 which may largely explain this finding). Furthermore, host factors were not available (eg, predisposing genetic abnormalities or family history), which may influence the development of secondary hematological malignancies. Finally, secondary hematological malignancies might have been under-reported because patients relocated from SEER registry catchment areas.

Despite these limitations, our results are very useful and broadly representative of pediatric cancers survivors in the United States. They demonstrate that the risk of secondary hematological malignancy is increasing over time and possibly will increase further with the growing population of pediatric cancer survivors. Furthermore, they describe the changing trends and risk patterns in the development of secondary hematological malignancy overtime. Notably, the risks of secondary ALL and secondary HL are not higher than expected. However, more cases of secondary hematological malignancy are expected to be identified using modern technologies. Long-term close monitoring and follow-up of pediatric cancer survivors for the heightened risk of secondary hematological malignancy are warranted.

Acknowledgements

We thank Melissa Hudson, MD, for reviewing and editing the article.

CONFLICT OF INTEREST DISCLOSURES

The King Hussein Cancer Center is supported by the King Hussein Cancer Foundation.

Ancillary