To evaluate the impact of reduced radiation and combined modality therapy (CMT) in the treatment of Hodgkin lymphoma, we assessed the risk of second malignant neoplasms (SMNs) in patients who received extended-field radiotherapy only and patients who underwent CMT. Among 404 patients treated at Yale during 1970–2004, the risk of solid SMNs was elevated in the radiotherapy only group (n = 198, median follow-up = 21·1 years) compared to the general population, with a standardized incidence ratio (SIR) of 1·85 [95% confidence interval (CI): 1·17–2·78]. No increase was observed in the CMT group (n = 206, median follow-up = 14·3 years), although potential differences in SMN risk were indicated across the age spectrum in subgroup analysis. Patients who received mustard-containing regimens had increased risks for haematological SMNs (SIR = 8·74) and all SMNs (SIR = 1·85). When the analysis was stratified by age at diagnosis, children (0–20 years) had a significantly higher risk of SMNs (SIR = 5·24, 95% CI: 2·26–10·33), regardless of the treatment received. These findings suggest that recent treatment options utilizing lower dose radiation and less intense alkylator chemotherapy might be associated with lower incidences of SMNs among adults but not necessarily children.
The tremendous success in cure rates for Hodgkin lymphoma (HL) has been diminished by the late adverse effects resulting from therapy. While the 10-year survival of HL patients now approaches 90% (Favier et al, 2009), survivors experience a higher mortality than age-matched healthy controls, particularly from secondary cancers (Ng et al, 2002a). Ongoing research efforts have focused on reducing the risk of second malignant neoplasms (SMNs). In the 1980s and early 1990s there was a transition from high dose extended field radiotherapy only to primary chemotherapy with lower doses and volumes of radiation (combined modality therapy, CMT) for the treatment of HL (Zittoun et al, 1985; Canellos et al, 1992). At Yale, clinicians started to adopt CMT with low-dose, involved-field radiation as early as 1969 for advanced stage HL and in the 1980s for paediatric early stage HL. Subsequently, the management of early stage adult HL also shifted to CMT. At the same time, chemotherapy regimens evolved from MOPP (mustargen, oncovin, procarbazine, prednisone) type regimens to less leukaemogenic regimens, such as ABVD (adriamycin, bleomycin, vinblastine, dacarbazine).
An earlier study of HL patients diagnosed between 1969 and 1985 at Yale found an increased incidence of second solid tumours in the primary radiation group (Salloum et al, 1996) recapitulating other studies (Tucker et al, 1988; Swerdlow et al, 1992; Abrahamsen et al, 1993; van Leeuwen et al, 1994; Ng et al, 2002b; Franklin et al, 2006); but importantly, there was not an elevated risk among patients who received CMT. This previous Yale study was completed prior to the more general switch to CMT for early stage HL in adults and was restricted to subsequent solid tumours. Since then, multiple studies have sought to determine how the risk of SMNs varies with the changing treatment regimens in children and adults. Ng et al (2002b) concluded that the risk of SMN was higher with CMT than radiation alone among 1319 patients with a median follow-up of 12 years. More recently, O'Brien et al (2010) reported that 112 patients treated during 1979–1990 with CMT had a similar risk for subsequent sarcomas, breast, and thyroid cancer to historical data, despite treatment with lower dose radiation. Multiple investigators have discovered that the risk of SMN is greater among those younger at diagnosis (Ng et al, 2002b; Hodgson et al, 2007; De Bruin et al, 2009).
Conflicting conclusions of past studies are probably due to the changing therapy strategies over time, as well as the limitations of inadequate follow-up in many instances. While subsequent haematological malignancies tend to occur in the first few years following HL diagnosis, solid tumours often occur more than 10 years after the initial treatment (Chronowski et al, 2004). It is important to conduct further research based on a large number of HL patients across the age spectrum who received contemporary therapy and were followed up for an extended period of time.
Patients and methods
The present study included adult and paediatric HL patients treated at Yale between 1970 and 2004, with follow-up until 2009. We evaluated the occurrence of second haematological malignancies, in addition to solid tumours. We assessed the risk of SMNs in two groups: (i) patients who received radiotherapy only, and (ii) patients who underwent CMT, overall and stratified by age of diagnosis (0–20 years vs. 21 years and above). In addition, we examined whether specific chemotherapy regimens and fields of radiation were associated with the development of SMNs. With several decades of using low dose involved-field radiation for the treatment of HL in children and adults, this single institution study is well-positioned to evaluate the risk of SMNs with current treatment across the age spectrum.
A total of 753 HL patients treated consecutively at the Yale-New Haven Hospital between 1970 and 2009 were identified through the Yale Tumor Registry. Medical records for these patients were obtained from the Department of Therapeutic Radiology. A total of 32 charts, mostly from patients treated in the early 1970s, could not be located. After further review, six patients were found to have been miscategorized as HL and were removed from the study. The remaining 715 patients' charts underwent abstraction by an internal medicine-paediatrics resident and a paediatric oncology nurse practitioner after extensive training by an attending radiation oncologist. Radiation doses to 29 major lymph node fields were abstracted and recorded, and the data were reviewed by the radiation oncologist for completeness and consistency. For the purposes of this analysis we excluded patients who did not undergo primary treatment at Yale (n = 104), patients with prior cancers (n = 10), patients treated with chemotherapy alone (n = 26), patients who were diagnosed during or after 2005 (n = 25), patients with recurrence (n = 119), and patients who died within three years after diagnosis of HL (n = 27), resulting in a final study population of 404 patients. We restricted our analysis to 3 years survivors because the development of SMNs is most relevant to patients cured of their original disease and including patients with early deaths would underestimate the true risk of SMNs. However, to assess the impact of possible SMNs that might have occurred shortly after treatment, we also conducted a sensitivity analysis by including the 27 patients who died within 3 years after diagnosis. Avoiding patients who relapsed removed the confounding effects of salvage therapy on the risk of SMNs from primary therapy alone. Medical records and tumour registry data were thoroughly reviewed to identify patients who developed SMNs. Non-melanoma skin cancers were excluded from the analysis of SMNs. With a median follow-up time of 17·4 years, a total of 42 (32 solid and 10 haematological) SMNs were observed among the 404 HL patients. The study protocol was approved by the Yale University Institutional Review Board.
The radiation dose to major lymph node regions (i.e. abdominal, axillary, cervical, mediastinal, and pelvic) was grouped by 300 cGy intervals, and the distribution of patients falling in each group was presented as line graphs. Analyses of variance were conducted to compare the distribution of radiation doses among the three groups. The follow-up began on the date of HL diagnosis and ended on the date of SMN diagnosis, last contact with tumour registry, the date of death, or the end of the study period (December 31, 2009), whichever occurred first. The cumulative incidence of developing a SMN over time was calculated using the life-table (i.e. actuarial) method, and a log-rank test was carried out to compare the cumulative incidences between groups. The impact of various treatment regimens on the risk of SMNs was assessed by using standardized incidence ratios (SIRs). A SIR was calculated as the ratio of the observed number of malignancies to the expected number of malignancies, which was derived by applying age-, sex-, race- and period-specific rates from the Surveillance, Epidemiology and End Results (SEER) program to the study population (‘period’ refers to calendar time in 5-year intervals). Statistical tests and the estimation of 95% confidence intervals (CI) were based on the assumption that the number of SMNs followed a Poisson distribution. It is important to note that SIR is a form of indirect standardization. While it is appropriate to use a SIR to make an inference about a study group and the standard population, it is not valid to compare two SIRs from two study groups.
The primary groups of interest were (i) patients who received radiotherapy only and (ii) patients who underwent CMT. For some of the analyses, the CMT group was further classified by HL stage (I/II vs. III/IV). Two statistical software packages, seer*stat (version 7.0.4; National Cancer Institute, Bethesda, MD, USA) and sas (version 9.3; SAS Inc., Cary, NC, USA) were utilized to conduct the analyses.
At the time of HL diagnosis, 121 patients were 0–20 years, 216 aged 21–40 years, and the other 67 were 41 years or older (Table 1). There was a slight male predominance (55%). The vast majority of patients were Caucasian, 4% were African-American, and <2% belonged to other racial groups. The number of patients whose pathology was classified as nodular sclerosis, mixed cellularity, lymphocyte-predominant, lymphocyte-depleted, nodular lymphocyte, and unclassified was 301 (74·5%), 64 (15·8%), 14 (3·5%), 7 (1·7%), 11 (2·7%), and 7 (1·7%), respectively. A majority (89%) of the 198 patients comprising the radiotherapy only group had stage I/II disease. Approximately half of the patients were treated with CMT, and the number of patients with stage I/II HL (n = 115) was slightly higher than the number of patients with stage III/IV disease (n = 91). As suggested by the line graphs, patients in the radiotherapy only group received higher radiation doses to all lymph node regions than patients in the CMT groups (Fig 1, P value from analysis of variance <0·0001 for all lymph node regions). The median doses of mediastinal, abdominal, axillary, pelvic, and cervical radiation were 4300, 3600, 3600, 3700, and 4300 (all in cGy), respectively, for the radiotherapy group; 2550, 2150, 2200, 2150, and 2500, respectively, for the CMT stage I/II group; and 2100, 2100, 2100, 2000, and 2100, respectively, for the CMT stage III/IV group. As the median radiation doses and cumulative incidences of SMNs were similar for the two CMT subgroups, we decided to combine the two CMT subgroups in the rest of the analyses.
Table 1. Characteristics of the study population (n = 404)
The 15-, 25-, and 35-year cumulative incidence of a solid SMN was 4% (95% CI: 2–8%), 16% (95% CI: 11–25%), and 20% (95% CI: 13–30%), respectively, for the radiotherapy only group, and 2% (95% CI: 1–5%), 9% (95% CI: 4–19%), and 9% (95% CI: 4–19%), respectively, for the CMT group. With a median follow-up of 21·1 years, the radiotherapy group appeared to have consistently higher cumulative incidences of solid SMNs than the CMT group (median follow-up = 14·3 years), but a log-rank test comparing the two curves of cumulative incidences yielded a P value of 0·09 (Fig 2). Characteristics of the 42 patients who developed SMNs are shown in Table 2. We looked at the effect of axillary lymph node irradiation, regardless of radiation dose, comparing patients who received any axillary radiation to patients who did not receive any radiation to the axillary fields. Only a single patient who had not received axillary radiation developed lung cancer, with no cases of breast cancer found in this group. This contrasts with 10 patients developing breast cancer and seven developing lung cancer in the group that did receive axillary lymph node radiation.
Table 2. Hodgkin lymphoma patients with second malignant neoplasms (n = 42)
Among 104 patients who received mustard-containing regimens and had a median follow-up time of 21·2 years, 14 developed SMNs (eight solid and six haematological), and the SIRs were 1·21 (95% CI: 0·52–2·38) for solid SMNs, 8·74 (95% CI: 3·21–19·03) for haematological SMNs, and 1·85 (95% CI: 1·01–3·11) for all SMNs combined. Among 77 patients who received ABVD and had a median follow-up time of 8·0 years, only two developed SMNs (one out of field endometrial cancer and one non-Hodgkin lymphoma), and the SIRs were 0·69 (95% CI: 0·02–3·84) for solid SMNs, 5·90 (95% CI: 0·15–32·86) for haematological SMNs, and 1·20 (95% CI: 0·15–4·35) for all SMNs combined. The cumulative incidence of SMNs at 20 years after HL diagnosis was 13% (95% CI: 7–22%) for those who received mustard-containing regimens and 4% (95%: 1–15%) for those treated with ABVD.
The SIR of solid SMNs in the CMT group was close to one, suggesting that the incidence rate of solid tumours among HL patients who received CMT was not much higher than in the general population (Table 3). In contrast, compared with the general population, HL patients who received radiotherapy only had a statistically significant 85% increase in the risk of solid SMNs (SIR = 1·85, 95% CI: 1·17–2·78). For the radiotherapy group, the SIR was particularly high for thyroid cancer (SIR = 9·03, 95% CI: 1·86–26·40) and breast cancer (SIR = 3·68, 95% CI: 1·59–7·25; Table 3). HL patients who received CMT had a significantly elevated risk of haematological SMNs (SIR = 8·66, 95% CI: 3·74–17·06), while the SIR for these malignancies in patients in the radiotherapy only group did not significantly deviate from one (Table 3). Of the 10 haematological SMNs observed in the study population, eight occurred in patients who underwent CMT for stage III/IV HL, including three patients with non-Hodgkin lymphoma, three patients with acute myeloid leukaemia, one patient with acute lymphoid leukaemia, and one patient with myelodysplasia.
Table 3. Standardized incidence ratios of second malignant neoplasms
When the analyses were stratified by age at diagnosis, the pattern of SMNs was different between children (0–20 years) and adults (≥21 years; Table 3 and Fig 3). The SIRs of SMNs in children were higher than five and statistically significant regardless of the treatment received (5·24 overall, 5·01 for radiotherapy only, and 5·49 for CMT). The SIRs of SMNs in adults were around 1·5 and only statistically significant for the overall study population (SIR = 1·51, 95% CI: 1·05–2·11). While adults who underwent CMT appeared to have a relatively low SIR (0·62, 95% CI: 0·20–1·45) for solid tumours, children who underwent CMT had a significantly increased risk of SMNs (SIR = 7·08, 95% CI: 1·93–18·14). The SMNs in children receiving CMT was confined to two breast cancers and two thyroid cancers out of the 77 total paediatric patients treated with CMT. Log rank tests comparing the cumulative incidences of solid and haematological SMNs in the radiotherapy only and CMT groups yielded P values of 0·30 and 0·01 for children and adults, respectively (Fig 3).
When we conducted a sensitivity analysis by including the 27 patients who died within 3 years after diagnosis, the differences between the results with and without the 27 patients were minimal and negligible.
Hodgkin lymphoma is a disease associated with relatively high cure rates but considerable morbidity and mortality among survivors from treatment-associated complications. Clinicians and investigators have modified therapy for HL to reduce subsequent malignant neoplasms. This study examined the potential effects that recent reductions in radiation doses and changes in chemotherapeutic agents had on the incidence of SMNs, with the advantage of following up a large cohort of patients, treated as recently as 2004, with follow-up for an extended duration. We observed no elevated risk of solid SMNs among adult patients who received CMT with reduced radiation. Paediatric patients, however, still have a predisposition to thyroid and female breast cancers, despite the switch to CMT. The risk of haematological SMNs was elevated in patients who received nitrogen mustard-containing CMT, but not among other patients.
For patients treated with radiotherapy alone, the incidence of all SMNs was significantly higher than the expected rate in the general population and comparable to other studies of long term HL survivors (Hoppe, 1997; Foss Abrahamsen et al, 2002; Ng et al, 2002a,b; Bhatia et al, 2003; De Bruin et al, 2009). Specifically, rates of breast cancer, lung cancer and thyroid cancer were similar to those observed in a recent population-based study of 18 862 5-year HL survivors from 13 population-based cancer registries in North America and Europe (Hodgson et al, 2007).
For patients who underwent CMT, there was not an increased risk of solid SMNs, and subsequent haematological malignancies were limited to patients who received MOPP-like regimens. In addition, there was a higher rate of haematological malignancy in patients with stage III–IV disease, presumably reflecting higher cumulative doses of leukemogenic drugs. In a relatively small study that compared early-stage HL patients treated with extended field radiotherapy using slightly lower doses of radiation (‘moderate dose radiotherapy’) to patients treated with chemotherapy combined with low dose involved field radiation, no SMNs were observed in the CMT group, while solid SMNs were increased in the primary, moderate dose radiotherapy group (Koontz et al, 2006).
In our study, patients who received radiation to axillary lymph node fields had an increased risk of solid SMNs, particularly breast and lung cancer, which is probably attributable to the larger volume of breast and lung tissues exposed to radiation from anterior-posterior fields used to encompass the axillary nodes. This result was consistent with a Dutch study showing a lower risk of breast cancer after smaller radiation volumes that excluded the axilla despite similar doses of radiation (De Bruin et al, 2009). Most patients in our study treated with smaller radiation volumes excluding the axillary fields also received lower doses of radiation, proabably leading to a further reduction in the risk of SMNs.
Mustard-containing regimens were associated with an increased risk of haematological malignancies. Some studies have suggested a complex relationship between alkylating chemotherapy agents and the risk of SMNs. Lung cancer risk may be increased with such agents (Travis et al, 2002). Others have suggested that alkylator therapy may be protective for subsequent breast cancer, possibly due to a premature menopause induced by the hormonal effects of those agents (Travis et al, 2005). A similar phenomenon of a reduced risk of secondary breast cancer has been observed with pelvic radiation (Basu et al, 2008). Our observation of the risk of solid SMNs among HL patients who received ABVD should be considered preliminary, due to the relatively short duration of follow-up.
Dosimetric risk modelling approaches (Koh et al, 2007) have suggested a 65% decrease in the incidence of secondary breast cancer and lung cancer, when using low dose involved field radiation instead of higher dose mantle radiation, with additional decrease in risk when lowering radiation doses. These findings are consistent with the different risk observed between the radiotherapy and CMT groups in our patient cohort. Other investigations have also shown a reduction in secondary solid cancers with a reduction in the volume (Ng et al, 2002b) and dose (Travis et al, 2002, 2005; Inskip et al, 2009) of radiation. The thyroid may possibly be an exception to this concept. A quadratic dose-SMN risk relationship may exist in which intermediate radiation dose levels (20–30 Gy) to this gland may paradoxically result in a higher risk of thyroid cancer (Meadows et al, 2009). Probably due to the relatively low absolute number of second thyroid cancers in our study, we did not observe such a risk relationship.
Among the 77 patients who were diagnosed with HL under the age of 21 years and underwent CMT, the SIRs for breast and thyroid cancers were significantly elevated at 25·8 (95% CI: 3·1–93·2) and 33·6 (95% CI: 4·1–122), respectively, and the 30-year incidence of non-haematological SMNs was 20%. These findings are comparable to those recently reported by a study of paediatric HL patients treated at Stanford University (O'Brien et al, 2010). Together, the results suggest that the paediatric population may have higher risks for SMNs even with low dose involved field radiation, despite other findings that a reduction in radiation dose does lower this risk (Constine, 2008; Castellino et al, 2011). Efforts to reduce normal tissue exposure to radiation in HL may be particularly fruitful in the paediatric population (e.g. involved nodal radiation, intensity modulated radiation, proton radiation, response-based omission of radiation; Girinsky & Ghalibafian, 2007; Chera et al, 2009; Weber et al, 2009).
While there are advocates for eliminating the radiation component for the treatment of early stage HL (DeVita, 2003; Provencio et al, 2003), several randomized studies suggest that adjuvant radiotherapy following primary chemotherapy is important in reducing the risk of relapse (Nachman et al, 2002; Laskar et al, 2004; Nishioka et al, 2004; Straus et al, 2004; Meyer et al, 2005). Additional salvage therapy at the time of relapse introduces further potential of late toxicities including SMNs. The NCIC HD.6 trial comparing ABVD chemotherapy alone, ABVD with radiotherapy, and radiotherapy alone is a unique randomized trial with a relatively long median follow-up of 11·3 years showing that the late toxicity of radiotherapy negates any benefit in progression free survival (Meyer et al, 2011). While radiotherapy resulted in a worsened overall survival, an important caveat is that this trial used antiquated subtotal nodal irradiation to 35 Gy. To what degree a lower radiation dose and volume might reduce the risk of late toxicity remains unanswered. The recently published results from the German Hodgkins Study Group HD10 randomized trial cautiously support a role for low dose involved field radiation after ABVD chemotherapy for early stage HL (Engert et al, 2010). With a relatively short median follow-up of 7·5 years, this trial observed 55 SMNs (38 solid tumours, 15 non-Hodgkin lymphomas, and 2 acute myeloid leukaemias) amongst the 1190 patients, while a further analysis of late effects in this large trial is awaited.
The current study is distinguished by the relatively large number of adult and paediatric patients included, the extended duration of follow-up, the inclusion of patients treated as recently as 2004, the availability of detailed treatment records, and the comprehensive assessment of pathologically-confirmed solid and haematological SMNs. In addition, at Yale, low-dose radiotherapy as part of CMT, including in adult early-stage HL patients, was adopted much earlier than most other cancer centres and we thus were able to assess long-term solid tumour risk in HL patients with lower doses of radiation exposure. Consistent with prior studies, a large number of SMNs in our study occurred more than 20 years after the initial diagnosis of HL, and the long duration of follow-up is a strength of our study. Still, additional years of follow up will be needed before the effects of the most recent HL treatment modalities can be fully evaluated.
Our results should be interpreted in the context of potential caveats. Despite a relatively large number of patients and an extended period of follow-up, only 42 SMNs were identified, which affected the statistical power of the study, especially for subgroup analyses involving children. The imprecision of some of the estimates is reflected by the relatively wide confidence intervals. The treatment protocol for HL is complex and has been evolving over the years. Given the relatively recent clinical adoption of some HL treatment modalities, such as CMT and ABVD, additional years of follow up will be needed before the effects of these modalities can be fully evaluated. While the Yale Tumor Registry is highly efficient in ascertaining patients who were diagnosed with SMNs in good part due to the linkage with the Connecticut Tumor Registry, the oldest tumour registry in the United States and a member of the SEER program since its inception in 1973, we acknowledge the possibility of not identifying a SMN if the patient moved to another state and lost contact with the tumour registry. However, it is unlikely that patient's migration patterns were related to the treatment modalities they received for HL. In the present study, it is reassuring that the SMN rate for the radiotherapy group, which has the longest follow-up and hence a high chance of migration, is consistent with that reported in the literature (Tucker et al, 1988; Swerdlow et al, 1992; Abrahamsen et al, 1993; van Leeuwen et al, 1994; Ng et al, 2002b; Franklin et al, 2006). Lastly, SIR is a form of indirect standardization and as such, is not a measure that should be compared directly between study groups (e.g. radiotherapy only group and CMT group). It is only appropriate to use SIR to make an inference about a study group and the standard population.
With a median follow-up of 14·3 years in the CMT group and 21·1 years in the radiotherapy group, this study provides some data indicating that current radiation-containing regimens for HL using lower radiation doses might be associated with a lower risk of SMNs than high dose extended field radiotherapy. In adults, HL patients who underwent CMT had a risk of solid SMNs comparable to that of the general population, while those who were treated with extended field radiotherapy only had a higher risk. Children and adolescents with HL still had an elevated predisposition to thyroid and breast cancers despite the use of CMT with low dose involved field radiation. The present study is observational by nature. While it has the advantage of reflecting the “real world experience” by including all eligible patients treated at Yale, the most powerful data on the impact of any treatment will come from extensive follow-up of patients enrolled in randomized clinical trials. To date, most randomized clinical trials for HL have used short-term outcomes such as 5-year progression-free survival or freedom from disease progression as primary end points (Straus, 2011). The HD.6 trial (Meyer et al, 2011), which evaluated the type of radiation that is now considered outdated, was the first trial that used late survival as the primary end point (Straus, 2011). Observational studies like ours can generate results complementary to randomized clinical trials. We plan to continue the follow-up of our cohort, which is characterized by the inclusion of many patients who received lower-dose radiation therapy, to assess if the current findings are sustained.
The authors thank the Tumor Registry and particularly Ms. Teresita Vega at the Yale-New Haven Hospital for providing data.
BO, NSK, KBR, and XM designed the study and wrote the manuscript, BO, KBR and CD collected the data, RW and XM conducted statistical analyses, RW, CD, GMK, DC and SS helped to interpret the results and contributed to the writing. All authors approved the final version of the manuscript.