Dr. Jonathan D. Tward had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
The risk of secondary malignancies over 30 years after the treatment of non-Hodgkin lymphoma
Article first published online: 17 MAY 2006
Copyright © 2006 American Cancer Society
Volume 107, Issue 1, pages 108–115, 1 July 2006
How to Cite
Tward, J. D., Wendland, M. M. M., Shrieve, D. C., Szabo, A. and Gaffney, D. K. (2006), The risk of secondary malignancies over 30 years after the treatment of non-Hodgkin lymphoma. Cancer, 107: 108–115. doi: 10.1002/cncr.21971
- Issue published online: 16 JUN 2006
- Article first published online: 17 MAY 2006
- Manuscript Accepted: 28 FEB 2006
- Manuscript Revised: 16 FEB 2006
- Manuscript Received: 11 JAN 2006
- absolute excess risk;
- observed-to-expected ratio;
- second malignancies;
- standardized incidence ratio;
- Surveillance, Epidemiology, and End Results Program
Survivors of non-Hodgkin lymphoma (NHL) are at increased risk for developing secondary malignancies. For the current study, the authors quantitated this risk in a group of NHL survivors over 30 years of follow-up.
Standardized incidence ratios (observed-to-expected [O/E] ratio) and absolute excess risk of secondary malignancies were assessed in 77,876 patients who were diagnosed with NHL between 1973 and 2001 from centers that participated in the National Cancer Institute's Surveillance, Epidemiology, and End Results Program.
There were 5638 patients who developed secondary malignancies, significantly more than the endemic rate (O/E, 1.14; P<.001). Overall, irradiated patients had a similar risk of secondary malignancies compared with unirradiated patients (relative risk, 1.04; 95% confidence interval, 0.98-1.10; P = .21). Irradiated patients had excess risk for sarcomas, breast cancers, and mesothelioma compared with unirradiated survivors (P<.05). Patients age <25 years at the time of their NHL diagnosis had the highest relative increased risk (no radiation: O/E, 2.1; P<.05; radiation: O/E, 4.51; P<.05). Overall, no statistical difference was observed for secondary cancer incidence between females and males (O/E, 1.12 vs. 1.15, respectively). Female survivors of NHL were less likely to develop breast cancer than the general population (O/E, 0.85; P<.05), but women age <25 years at the time of their NHL diagnosis were more likely to develop breast cancer (no radiation: O/E, 2.1; P<.05; radiation: O/E, 4.51; P<.05).
The overall risk of secondary malignancies was increased for NHL survivors and varied according to age at NHL diagnosis, gender, and treatment. Cancer 2006. © 2006 American Cancer Society.
This year, an estimated 56,300 patients will be newly diagnosed with non-Hodgkin lymphoma in the U.S., and it is estimated that there will be nearly 19,200 deaths from NHL.1 The survivors of NHL have a significantly elevated risk for secondary cancers, especially lung cancer, brain cancer, renal cancer, bladder cancer, melanoma, Hodgkin lymphoma, and acute nonlymphocytic leukemia.2 Chemotherapy,3–7 high-dose chemotherapy, autologous stem cell transplantation,8–11 and radiation therapy2, 12, 13 all have been implicated as risk factors. Currently, patients with lymphoma receive various combinations of chemotherapy, radiation, radioimmunotherapy, and high-dose chemotherapy with autologous stem cell transplantation. Therefore, it is important to understand how different patient characteristics and their therapies alter their future cancer risk profiles, so that appropriate follow-up and management may occur. Accordingly, we evaluated the absolute excess risk (AER) and relative site-specific risks (standardized incidence ratio [SIR] or observed-to-expected [O/E] events) among 77,823 patients who were treated for NHL between 1973 and 2001, taking into account age at diagnosis, gender, radiation therapy, and latency period, with regard to the development of secondary cancers.
MATERIALS AND METHODS
The Surveillance, Epidemiology, and End Results Program
The Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute collects and publishes cancer incidence and survival data from 14 population-based cancer registries and 3 supplemental registries that cover approximately 26% of the U.S. population. SEER began collecting data on January 1, 1973 and currently has information on >3 million patients with in situ and invasive cancer. The SEER Registries routinely collect data on patient demographics, primary tumor site, morphology, stage at diagnosis, first course of treatment, and follow-up for vital status.
The SEER Program statistical analysis software package (SEER*Stat, version 5.3.1)14 was used to identify patients who were diagnosed with any stage of NHL as their first primary malignancy between 1973 and 2001. The SEER*Stat Multiple Primary-SIR tool was used to calculate SIRs and excess risk for secondary malignancies by comparing these patients' subsequent cancer experience with the number of cancers that would be expected based on incidence rates for the general U.S. population. These analyses were adjusted for age, gender, race, and year of NHL diagnosis. Information regarding lymphoma staging, chemotherapy, radiation dose, and sites to which radiation therapy was directed is not available currently in the SEER public-use data base.
Individuals who developed a secondary malignancy within 2 months of their NHL diagnosis or whose NHL diagnosis was not their first primary cancer were excluded from the analysis. The time to development of secondary malignancies was calculated from the date patients were diagnosed with NHL.
The risk of second cancers was estimated by compiling person-years (PYs) of observation according to age, gender, and calendar-year periods from 2 months after the date of NHL diagnosis to the date of death, the date of last follow-up evaluation, the date of diagnosis of second cancer, or the end of the study (December 31, 2001), whichever occurred first. Cancer incidence rates that were specific for 5-year age groups, gender, and calendar-year intervals were multiplied by the accumulated PYs at risk to estimate the number of cancer cases expected. The observed and expected numbers of second cancers were then summed, with the SIR expressed as the ratio of O/E cases. The AER was determined by subtracting the expected number from the observed number of second cancers and then dividing the difference by the number of PYs at risk. The number of excess second cancers was expressed per 10,000 PYs. Risks of second cancers were stratified by gender, age group at NHL diagnosis, time since NHL diagnosis (latency), and treatment (radiotherapy vs. no radiotherapy). Statistical tests and 95% confidence intervals (95% CIs) were based on the assumption that the observed number of second cancers was distributed as a Poisson variable. Tests for heterogeneity and linear trend were conducted by using the methods of Breslow et al.15
SIR and AER are 2 complimentary measures of the incidence of an event of interest (in this instance, secondary cancer) in a subpopulation compared with the entire population. Both are based on comparing the observed number of events in the subpopulation (O) with the number of events that would be expected (E) if the risk profile for the subpopulation was identical to that for the full population. Because, as an individual ages, his or her risk of an event typically changes (because of both attained age and attained calendar year), the calculation of the expected number of events is adjusted for these variables. In addition, fixed characteristics that affect event rates, such as gender and race, are incorporated into the calculation of the expected number of events. These adjustments make subpopulations of different structures comparable: for example, longer follow-up in a particular group. SIR and AER differ in the way they combine O and E: SIR measures the fold difference between the observed and expected number of events (SIR = O/E), whereas AER measures the actual number of excess events normalized to the number of PYs observed (AER = [O − E]/PY). Therefore, SIR measures the relative risk of the event on an individual level, and it does not depend on the frequency of the event in the population; whereas AER measures the population impact, i.e., small increases in the relative risk of a common event affect more individuals than a large increase in a rare event. It should be noted that the length of follow-up does not affect either measure: the number of observed events (O) increases, but so do the number of expected events (E) and the number PYs (PY).
The effect of external-beam radiation therapy for the treatment of NHL on the development of secondary malignancies was analyzed for cohorts based on age at diagnosis of NHL, gender, and the latency period in decades.
There were 77,823 patients included in the current analysis. Patient characteristics for the entire cohort are shown in Table 1. Second cancers developed in 5638 patients. In some patients, >1 secondary cancer was diagnosed; therefore, 6188 additional malignancies were observed (Table 2).
|Variable||All Patients||External Beam||No Radiation|
|Total no. of patients with NHL||77,823||21,100||55,351|
|No. who developed second malignancy||5638||1595||3950|
|Average age at NHL diagnosis, y||61||59||62|
|Average age at first secondary, y||70||69||71|
|Person-years at risk||380,505||112,020||261,073|
|Average months at risk||60||60||60|
|American Indian/Alaska Native||218||0.3||6||0.4||3||0.1|
|Asian or Pacific Islander||3867||5.0||66||4.3||120||3.1|
|Characteristic||All Patients||External Beam||No Radiation|
|Total no. of patients||5638||1595||3950|
|Age at NHL diagnosis, y|
|Person-years of follow-up||49,233||15,518||32,782|
|American Indian/Alaska Native||11||0.2||8||0.5||3||0.1|
|Asian or Pacific Islander||201||3.6||71||4.5||130||3.3|
Overall, second malignancies occurred at a higher rate among the study population than what would be expected in the general population (O/E, 1.14; 95%CI, 1.11-1.17; AER, 20.12) (Table 3). There was a significantly increased risk for cancers of the head and neck, melanoma, lung cancer, colon cancer, bladder cancer, renal cancer, Hodgkin disease, leukemia, and Kaposi sarcoma compared with the general population. Significantly decreased risks of female breast cancer, prostate cancer, and myeloma also were observed.
|Malignancy/Tumor Site||All Patients|
|Patients (77,876)||Person-Years at Risk (381,257)|
|Observed||Excess Risk*||O/E||95% CI|
|All sites excluding NHL||5,960||19.24||1.14†||1.11–1.17|
|All solid tumors‡||5,363||13.78||1.11†||1.08–1.14|
|Head and neck||223||1.29||1.28†||1.12–1.46|
|Lung and mediastinum||1,095||6.76||1.31†||1.23–1.39|
|Colon excluding rectum||599||1.42||1.1†||1.01–1.19|
|Rectum and rectosigmoid junction||187||−0.28||0.95||0.81–1.09|
|Anus, anal canal, and anorectum||16||0.11||1.36||0.77–2.2|
|Liver, gallbladder, and biliary||82||−0.06||0.97||0.77–1.21|
|Kidney and renal pelvis||187||1.75||1.56†||1.34–1.8|
External-beam radiation did not alter the overall risk significantly for secondary cancers versus unirradiated patients with NHL (all sites: O/E, 1.18 vs. 1.13, respectively; AER, 23.36 vs. 19.36, respectively) (Table 4). However, when comparing specific cancer subtypes between the irradiated and nonirradiated cohorts, radiation therapy resulted in significantly more soft tissues cancers (sarcomas), female breast cancers, and mesotheliomas (P<.05). Although the unirradiated patients had a significantly decreased risk of developing female breast cancer (O/E, 0.79; AER, − 3.57), the risk for the irradiated group was equivalent to the endemic rate (O/E, 1.00; AER, 0.01). The prostate cancer risk also was significantly lower for unirradiated patients compared with the risk among the standard population (O/E, 0.89; AER, − 3.11).
|Malignancy/Tumor Site||No Radiation||Radiation*|
|Patients (55,392)||Person-Years at Risk (261,545)||O/E||95% CI||Patients (21,111)||Person-Years at Risk (112,281)||O/E||95% CI|
|Observed||Excess Risk||Observed||Excess Risk|
|All sites excluding NHL||4191||19.14||1.14†||1.1–1.17||1671||20.9||1.16†||1.11–1.22|
|All solid tumors‡||3752||12.84||1.1†||1.06–1.13||1521||17.06||1.14†||1.09–1.2|
|Head and neck||159||1.42||1.3†||1.11–1.52||61||1.1||1.25||0.96–1.61|
|Lung and mediastinum||782||7.28||1.32†||1.23–1.42||300||6.25||1.31†||1.16–1.46|
|Colon excluding rectum||440||2.12||1.14||1.04–1.26||152||0.21||1.02||0.86–1.19|
|Rectum and rectosigmoid junction||130||−0.34||0.94||0.78–1.11||50||−0.45||0.91||0.67–1.2|
|Anus, anal canal, and anorectum||10||0.06||1.2||0.58–2.21||5||0.15||1.53||0.49–3.57|
|Liver, gallbladder, and biliary||54||−0.2||0.91||0.68–1.19||24||0.05||1.02||0.66–1.52|
|Kidney and renal pelvis||134||1.87||1.58†||1.32–1.87||52||1.69||1.58†||1.18–2.07|
In general, the O/E risk of all secondary cancers was similar in men and women (O/E, 1.15 vs. 1.12, respectively; relative risk, 1.02; P = .92). Lung cancer and mediastinal cancers accounted for the greatest AER in both men and women (6.36 and 7.19, respectively). The greatest decreases in risk were seen for breast cancer in women (AER, − 5.05; O/E, 0.85; P<.05) and for prostate cancer in men (AER, − 5.29; O/E, 0.9; P<.05).
The risks of all secondary cancers and those for selected sites are listed in Table 5 according to age at diagnosis of NHL and treatment modality. The overall O/E risk of secondary malignancies was highest for the youngest cohort and trended downward with increasing age. NHL diagnosed in patients age >75 years carried a relative risk of secondary malignancies similar to that for the general population and a decreased AER (all sites, with radiation: O/E, 0.94; AER, − 15.77; all sites, without radiation: O/E, 0.93; AER, − 17.06). Patients age <25 years at the time of NHL diagnosis who had received therapy had a significantly elevated risk for leukemia, thyroid, and female breast cancer (O/E: 16.21, 8.69, and 5.03, respectively). Women who did not receive radiation therapy had a decreased risk of breast cancer across all age cohorts.
|Malignancy/Disease Site||No Radiation||Radiation*|
|Age <25 Years||Ages 25–49 Years||Ages 50–74 Years||Age ≥75 Years||Age <25 Years||Ages 25–49 Years||Ages 50–74 Years||Age ≥75 Years|
|Patients (1752)||Person-Years at Risk (12,288)||Patients (11,061)||Person-Years at Risk (65,442)||Patients (29,549)||Person-Years at Risk (148,890)||Patients (12,989)||Person-Years at Risk (34,453)||Patients (1041)||Person Years at Risk||Patients (5104)||Person-Years at Risk (31,744)||Patients (10,737)||Person-Years at Risk (59,811)||Patients (4218)||Person-Years at Risk (11,624)|
|Excess Risk||O/E||Excess Risk||O/E||Excess Risk||O/E||Excess Risk||O/E||Excess Risk||O/E||Excess Risk||O/E||Excess Risk||O/E||Excess Risk||O/E|
|All sites excluding NHL||3.03||1.71||32.06||1.81†||22.49||1.13†||−14.74||0.94||17.5||4.42†||39.39||1.93†||18.66||1.11†||−16.48||0.93|
|All solid tumors‡||1.35||1.38||26.33||1.71†||14.66||1.09†||−16.83||0.92†||11.52||3.67†||35.15||1.88†||14.21||1.09†||−14.85||0.93|
|Head and neck||−0.11||0||1.64||1.87†||1.37||1.23†||1.88||1.33||−0.14||0||0.81||1.4||1.56||1.26||0.58||1.11|
|Lung and Mediastinum||−0.12||0||6.82||2.3†||11.04||1.37†||−5.85||0.82||−0.15||0||8.58||2.45†||8.88||1.31†||−8.02||0.74|
|Colon excluding rectum||0.71||8.12||0.66||1.27||3.4||1.2†||−0.12||1||−0.13||0||0.49||1.18||1.58||1.09||−8.63||0.75|
|Rectum and rectosigmoid junction||−0.06||0||−0.04||0.97||−0.04||0.99||−2.18||0.78||−0.07||0||−0.14||0.9||−0.14||0.98||-3.07||0.69|
|Anus, anal canal, and anorectum||−0.01||0||0.62||5.49†||−0.18||0.53||0.07||1.15||−0.02||0||0.49||4.6||0.13||1.35||−0.51||0|
|Liver, gallbladder, bile ducts||−0.03||0||0.24||1.47||−0.03||0.99||−1.83||0.61||−0.04||0||0.37||1.65||−0.34||0.87||1.31||1.28|
|Kidney and renal pelvis||−0.07||0||1.6||2.4†||2.45||1.6†||0.66||1.14||−0.09||0||3.19||3.62†||1.41||1.36||−0.32||0.93|
The relative risk of developing a secondary malignancy appeared to increase with increasing latency from diagnosis. However, after adjusting for age at diagnosis, the effect of latency disappeared: The adjusted SIR was 1.59, 1.55, and 1.65 for latencies <10 years, from 10 years to 20 years, and >20 years, respectively, for a test of equality with P = .64.
Multiple studies3, 6, 8, 10, 11, 16, 17 in the last decade have demonstrated that patients with NHL are at significantly greater risk for secondary cancers, despite earlier reports to the contrary.18–22 Our current results update and expand on the 1993 findings of Travis et al.2 and now represent the largest population-based study of NHL survivors. Greater than 2000 ≥20-year survivors were included, allowing us to evaluate cancer risk over 30 years. Our report validates more recent reports that NHL survivors are at a significantly increased risk for secondary cancers and that both the O/E risk and AER increase with the length of survival.
Men and women NHL survivors have an overall equivalent risk for developing secondary malignancies. The 1993 report by Travis et al. indicated that men had a relative risk of 1.25 for all secondary cancers compared with 1.08 for women. Our update showed that the relative risk was 1.15 in men and 1.12 in women.
The current study demonstrated an O/E risk for prostate cancer of 0.9 (95% CI, 0.84-0.96; P<.05). This translated to a decrease in AER of − 5.3. This may be a significant underestimate of the effect, because the incidence of prostate cancer has risen sharply secondary to prostate-specific antigen screening.23 Several hypotheses may explain the reduced risk for prostate cancer in NHL survivors. Modern treatment for NHL may eradicate or delay the clinically detectable progression of occult prostate cancers.24, 25 Alternatively, men who have been diagnosed with a prior malignancy like NHL may be more conscientious of their health, resulting in lifestyle changes that have an impact on their risk of prostate cancer.26–28
In 1993, the risk of a secondary lung cancer in men was 1.61 times that of the reference population; whereas, in the current analysis, this risk had decreased to 1.23. Cancer survivors are less likely to smoke than the general population,29, 30 which may explain this lung cancer risk decrease; however, in women, we observed a relative risk of 1.45 for lung cancer development, identical to the 1993 observation.
When we excluded gender-specific malignancies, we discovered that women had a relative risk of 1.29 for secondary cancers over the general population. Female breast cancer accounted for the greatest number of observed secondary malignancies (548 patients) in women, yet NHL survivors are protected from this malignancy compared with the general population (Table 3). It is conceivable that lowered estrogen levels from the ablation of ovarian function from chemotherapy may explain this decline. In patients with Hodgkin disease who receive ovarian-ablative therapy with either chemotherapy or radiotherapy, a decrease in breast cancer has been observed.31, 32
Radiation therapy did not increase the overall, solid tumor, or hematologic cancer risk of secondary malignancies significantly compared with the unirradiated cohort (Table 4). The only significant differences that were observed among irradiated patients versus unirradiated patients were in the development of soft tissue malignancies (sarcomas), female breast cancer, and mesothelioma. An increased risk of sarcoma development in irradiated patients has been reported for other malignancies.33–35 Although the risk of a secondary breast cancer is greater for irradiated NHL survivors versus unirradiated women, the risk of a secondary breast cancer overall in irradiated women is not different from that in the general population (O/E, 1.00; 95% CI, 0.86-1.16). Additional analysis revealed that, for the irradiated women, age younger than 25 years at NHL diagnosis translated into a 5-fold risk increase for breast cancer, indicating the deleterious effect of radiotherapy in young patients. To the best of our knowledge, increased mesothelioma risk after irradiation has not been described in other population-based studies. The aggregate AER for soft tissue malignancies, female breast cancer, and mesothelioma in irradiated patients was 1.57 cases per 10,000 PYs.
The O/E risk of secondary cancers is greatest for the youngest cohort, whereas the AER for secondary cancers peaks among patients ages 25 to 49 years at the time of their initial NHL diagnosis, and then declines with advancing age (Table 5). This held true for both the irradiated and unirradiated patient cohorts. These data highlight the importance of the judicious use of radiotherapy in young patients, and particularly in growing children. Further studies with more complete data bases that can match cohorts with additional prognostic variables are needed to define these risks further.
We also considered whether latency affects the incidence of secondary cancers; that is, whether the risk relative to the general population changes as the time since NHL diagnosis increases. At first glance, it appeared that, with increased latency, the SIR increased: For all cancers, the SIRs were 1.12, 1.21, and 1.50 for latencies <10 years, from 10 to 20 years, and >20 years, respectively, for a test of linear trend with P<.0001. However, such an analysis is misleading, because age at diagnosis affects both the risk of secondary cancers and the probability of surviving for 10 or 20 years after diagnosis. The latter effect is a combination of higher cause-specific survival among younger patients and a lower rate of death from other causes. Consequently, among the patients who have a latency <10 years, 71% are age >50 years at diagnosis; whereas, among the patients with a latency >20 years, the proportion declines to 40%. After adjusting for age at diagnosis, the effect of latency disappears: The adjusted SIRs were 1.59, 1.55, and 1.65 for latencies <10 years, from 10 to 20 years, and >20 years, respectively, for a test of equality with P = .64. We performed similar evaluations for all the cancer sites (data not shown) and found no evidence of latency-related changes in the risk of a secondary cancer after adjusting for age at diagnosis. The number of patients in the latency period beyond 240 months was comparatively smaller than the number in other latency periods; therefore, conclusions must be interpreted cautiously.
Some clinical and pathologic data with known prognostic significance are not available in the SEER data base. Specifically lacking is information regarding lymphoma staging, details about types or amounts of chemotherapy, whether patients underwent bone marrow transplantation, the proportion of individuals who had the human immunodeficiency virus or who were otherwise immunocompromised, or the specifics of radiotherapy dose, treatment fields, and fractionation schemes.37 In addition, the SEER data base does not record history of treatment failure or time of recurrence. Therefore, we were unable to adjust for these factors in the current analyses.
In conclusion, the results of the current study support the observations of others that NHL survivors are at higher risk for developing both solid tumors and hematologic secondary malignancies than the general population. These risks are altered by the type of second cancer, age at diagnosis of NHL, effect of radiation therapy, and gender. Overall, radiation therapy does not cause a significant increase in excess risk of secondary malignancies for patients with NHL. Further investigations with more robust data bases that incorporate the specifics of staging, treatment delivery, genetics, and other prognostic variables will be crucial to our understanding of this growing patient population.
- 1American Cancer Society. Cancer Facts and Figures, 2005. Atlanta: American Cancer Society; 2005.
- 5Risk of acute nonlymphocytic leukemia and preleukemia in patients treated with cyclophosphamide for non-Hodgkin's lymphomas. Comparison with results obtained in patients treated for Hodgkin's disease and ovarian carcinoma with other alkylating agents. Ann Intern Med. 1985; 103: 195–200., , , et al.
- 10Moderate increase of secondary hematologic malignancies after myeloablative radiochemotherapy and autologous stem-cell transplantation in patients with indolent lymphoma: results of a prospective randomized trial of the German Low Grade Lymphoma Study Group. J Clin Oncol. 2004; 22: 4926–4933., , , et al.
- 14National Cancer Institute, Surveillance, Epidemiology, and End Results (SEER) Program. SEER*Stat, ver 5.3.1. [computer program]. Bethesda: National Cancer Institute.
- 23RiesLAG, EisnerM, KosaryCL, et al., editors. SEER Cancer Statistics Review, 1975-2002. Bethesda: National Cancer Institute; 2005.
- 36National Cancer Institute, Surveillance, Epidemiology, and End Results (SEER) Program. SEER data quality. Available at URL: http://seer.cancer.gov/about/quality.html Accessed May 6, 2005.
- 37National Cancer Institute, Surveillance, Epidemiology, and End Results (SEER) Program (available at URL: www.seer.cancer.gov). SEER*Stat data base: incidence-SEER 13 registries, public use, 1973-2002 varying (based on the November 2004 submission; released April 2005). Bethesda: National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, 2005.