An important new finding in our study, based on 298 patients with HL-BC and 405,223 patients with BC-1, is the observation that HL survivors in the general population who develop BC, as compared with women with a first or only BC, are at a significantly increased seven-fold risk of death from other cancers. Moreover, these patients experience significantly elevated two- to four-fold greater risks of cardiac death. For the first time, to our knowledge, the OS and CSS beyond 10 years, as well as the influence of several clinicopathologic factors (ie, sociodemographic status) on survival outcomes, are evaluated among patients with HL who develop BC. We show that lower sociodemographic status adversely affects both BC-CSS (P = .04) and OS (P = .09) among women who develop localized BC after HL.
Whereas the increased risk of new primary cancers in patients with HL is well-established,2,6,7,22,29
few studies address the development of two or more cancers after HL2
or subsequent mortality due to these malignancies.6,7
To our knowledge, none have addressed cancer mortality among HL survivors who develop BC. In our study, deaths from third or high-order cancers occurred at sites for which significantly increased risks of second cancers have been reported.,2,4,22,29–31
cancers of lung,2,4,22,29,31,33
HL survivors have a significantly increased three- to 10-fold risk of lung cancer,2,4,22,29,31,33
with associated risk factors including prior radiation, alkylating-agent chemotherapy (both with significant dose-responses), and smoking history.33,34
Virtually all lung cancers develop in HL survivors who smoke.33,34
Thoracic radiation and smoking history also significantly increase the risk of lung cancer after BC.35,36
Although it is possible that deaths resulting from metastatic BC were erroneously ascribed to lung cancer for either patients with HL-BC or BC-1, the reported rate of such misclassification is low.37
Without independent histopathologic review, unequivocal classification of incident lung cancer after BC is not possible.
HL survivors have a significantly increased five-fold to more than 20-fold risks of NHL, with excesses persisting for more than 15 years after HL.2,4,5,22,31,32
Although death attributed to NHL in our study could have represented misclassified HL, this type of error is less likely in patients with a lengthy interval between the two lymphoma diagnoses. Similarly, although women originally assigned a diagnosis of HL could have actually had NHL, this type of misclassification occurs in only approximately 2% of patients.32
Using an actuarial comparison, patients in the HL-BC group had a significantly increased two- to four-fold risk of death from heart disease as compared with patients in the BC-1 group. Although patients with either HL23
are susceptible to treatment-related cardiac toxicities (attributable to radiation and/or anthracycline chemotherapy), HL survivors tend to receive these exposures at a considerably younger age. Moreover, compared with patients with BC, patients with HL are generally given higher anthracycline doses and are irradiated to larger cardiac volumes, albeit at lower doses.39
The higher risk of cardiac death after HL is also noteworthy in view of the common use of mechlorethamine, vincristine, procarbazine, and prednisone chemotherapy during much of the study period (1970s to mid 1980s), after which doxorubicin, bleomycin, vinblastine, and dacarbazine (which includes cardiotoxic doxorubicin) became more widely used.40
Among women with localized BC, patients in the HL-BC group had a significantly increased two-fold risk of death resulting from BC as compared with patients in the BC-1 group, which may in part be due to an approximately two-fold (albeit nonsignificant) greater prevalence of contralateral BC (). However, contralateral BC explains only a small part of the inferior BC-CSS (), which may also reflect patient susceptibility or treatment-induced factors and/or limitations in treatment options for BC after HL. For example, HL survivors are less likely to receive anthracycline chemotherapy for BC.21
Moreover, our study and another investigation21
also showed that HL survivors are significantly less likely to receive breast-conserving surgery or radiotherapy for BC than patients with BC-1. We also found that HL survivors who underwent breast-conserving surgery were less likely to receive radiation, most likely due to prior chest radiation for HL.
BC after treatment for HL may also exhibit a different biology compared with de novo BC. In an analytic population-based study of BC after radiation for HL, archived paraffin-embedded tissues from 19 women were compared with de novo BC.41
BC after HL was characterized by a 4.2-fold increase (P
= .16) in microsatellite alterations, reflecting widespread genomic instability.41
Moreover, loss of heterozygosity in several chromosomes were significantly greater among HL survivors versus those with de novo BC.41
In another series, BC after HL, as compared with de novo BC, had a gene expression profile characterized by high proliferation, more aggressive tumor type (in approximately 80% of HL survivors),42
and greater likelihood of chromosomal aberrations (unpublished work discussed in Broeks et al42
). Another investigation showed no differences in germ-line mutations in the ataxia-telangiectasia gene between BC after HL and de novo BC.43
Because of the long follow-up in our study, we are able to report for the first time 15-year OS of HL survivors who develop BC. Two matched-pair analyses have also compared OS of HL survivors against women with a first or only BC,20,21
although with far fewer patients and less follow-up than in our study. One series analyzed 21 HL survivors and another evaluated 53 patients (of whom 35 patients had HL). In neither study was OS significantly different between HL survivors and those with de novo BC at time points of 5 to 10 years.20,21
Compared with women with first or only BC, HL survivors were relatively younger at time of BC diagnosis; diagnosed in relatively later decades, in part reflecting the typical latency period of ≥ 10 years required for radiation-associated solid tumors27
; and had a higher sociodemographic status, which is known to be correlated with HL.44,45
BCs among HL survivors were significantly more likely to be characterized by pathologic features of poorer grade, ER negativity, and PR negativity. Whether the greater risk of ER- and PR-negative tumors in our study may somehow reflect chemotherapy-induced (specifically mechlorethamine, vincristine, procarbazine, and prednisone) ovarian ablation46,47
is not known. Yahalom et al17
compared the histologic features of 45 BCs in 37 patients previously treated for HL with those in 935 patients with first BC. Nuclear grade, histologic grade, lymphocytic reaction, and lymphatic invasion were similar between the two groups.
The primary strength of the current study is the large number of patients (> 400,000) including 298 with BC after HL. All patients were derived from population-based registries, with no selection biases and with long-term follow-up, strengthening the generalizability of our results. Weaknesses of the SEER database include lack of information about radiation dose and fields as well as whether or not patients received chemotherapy (or what agents they received), factors important in predicting BC risks in patients with HL.5,9,12,16
In addition, it should be recognized that even the initial course of radiotherapy is underreported in the SEER program.48
With respect to BC-CSS, the power to detect covariate effects was limited because only 52 patients with HL-BC died as a result of BC, the most common cause of death in these patients. There were also small numbers of deaths from other causes.
Nonetheless, our results underscore the importance of continuing to monitor HL survivors who develop BC for late cardiac complications23
and additional malignancies, as well as the need to counsel patients on preventive measures such as smoking cessation and healthy lifestyle modifications. Systematic lifelong follow-up is needed to examine the emergence of long-term toxicity and associated morbidity and mortality.49,50
Future research should be directed at examining the underlying cancer biology and etiology of treatment-induced cancers, as well as inherent and treatment-induced genetic susceptibility of HL survivors. Genetic susceptibility may not only impact risk of developing additional cancers, but also subsequent survival, and warrants careful investigation.