The impact of prior malignancies on the development of second malignancies and survival in follicular lymphoma: A population‐based study

Abstract We assessed the impact of a prior malignancy diagnosis (PMD) – as a potential proxy for genetic cancer susceptibility – on the development of a second primary malignancy (SPM) and mortality in follicular lymphoma (FL) patients. From the nationwide Netherlands Cancer Registry, we selected all adult FL patients diagnosed in 1994‐2012 (n = 8028) and PMDs and SPMs relative to FL, with follow‐up until 2017. We constructed two Fine and Gray models – with death as a competing risk – to assess the association between a PMD and SPM incidence. A PMD was associated with an increased incidence of SPMs (subdistribution hazard ratio [SHR], 1.30; 95% confidence interval [CI], 1.03‐1.64) – especially carcinomas of the respiratory tract (SHR, 1.83; 95% CI, 1.10‐3.05) and cutaneous squamous cell carcinomas (SHR, 1.58; 95% CI, 1.01‐2.45) – and a higher risk of mortality in a multivariable model (HR, 1.43; 95% CI, 1.19‐1.71). However, when additionally adjusted for the receipt of systemic therapy and/or radiotherapy before FL diagnosis, only patients who received such therapies had an increased incidence of SPMs (SHR, 1.40; 95% CI, 1.02‐1.93). In conclusion, patients with a PMD had a higher rate of SPMs and mortality than those without a PMD, which might have resulted from therapy‐related carcinogenesis.

The improved longevity of patients with FL might come at a price, as these patients might live long enough to develop second primary malignancies (SPMs). A few studies have reported an increased risk of hematological and solid SPMs among patients with FL, as compared to the general population [12][13][14]. More specifically, patients had a statistically significantly elevated risk of Hodgkin lymphoma and acute myeloid leukemia, and solid tumors of the following sites: oral cavity and pharynx, stomach, lung, melanoma skin, nonmelanoma skin, urinary bladder, and kidney/pelvis [12][13][14].
SPMs could reflect late sequelae of treatment or the effect of shared etiologic factors, environmental exposures, and genetic and non-genetic host characteristics, as well as combinations of these influences -including gene-environment and gene-gene interactions.
Suggested risk factors for SPMs among patients with FL [12][13][14] and non-Hodgkin lymphoma in general, included age > 65 years, male sex, and receipt of radio-and/or chemotherapy for the lymphoma [15].
Among patients with multiple myeloma diagnosed in Sweden, a prior malignancy diagnosis (PMD) -as a potential proxy for genetic susceptibility to cancer -was associated with SPM development and mortality, as compared to those without a PMD [16]. At present, information on the impact of a PMD on the development of SPMs and mortality among patients with FL is lacking. This information is important as it could have consequences for surveillance on the development of SPMs among patients with a PMD. Also, whenever a newly diagnosed patient with FL has received prior anti-neoplastic therapy, a treatment regimen with a low toxicity profile might be considered to avoid excessive, cumulative toxicity and reduce treatment-related mortality. Therefore, this nationwide population-based study aimed to assess the impact of a PMD on the development of SPMs and mortality in patients with FL in the Netherlands.

Study population
We selected all adult (≥18 years) patients diagnosed with FL grades

Statistical analysis
Descriptive statistics were employed to compare patient characteristics between those with and without a PMD. The SHR describes the relative change in the instantaneous rate of the occurrence of an SPM in those who did not develop an SPM during follow-up (ie, the event of interest) and those who died before that event occurred (ie, the competing risk) [19]. Given the relationship with the cumulative incidence function for the subdistribution hazard function, the SHRs can also be interpreted as the effect of a PMD on the incidence of SPMs. Of note, the magnitude of the effect of a PMD on the incidence of SPMs cannot be directly quantified by using SHRs.
Similar to the competing risk models, we constructed two Cox proportional hazard models to calculate hazard ratios (HRs) with 95% CIs for the association between a PMD and mortality (ie, overall survival).
Patients were censored at the time of emigration or at the end of the study (ie, December 31, 2016), whichever occurred first. Both models were additionally adjusted for baseline characteristics at FL diagnosis, namely sex, age at diagnosis, year of diagnosis, and stage at diagnosis.
We performed sensitivity analyses in which we excluded patients with synchronous malignancies within a time-interval of 6 months before or after FL diagnosis. The impact of applying different definitions of synchronous malignancies in relation to the outcome has been appraised previously [20,21].
A P-value of <.05 indicated statistical significance. Statistical analyses were performed with STATA Statistical Software version 14.2 (Stat-aCorp, College Station, TX).

Patient characteristics
A total of 8028 patients with FL-of whom 483 (6%) had a PMD and 1106 (14%) developed an SPM-were included in the study. Characteristics of these patients at the time of FL diagnosis are presented in Table 1 according to the history of a PMD. The majority of patients with a PMD did not receive systemic therapy and/or radiotherapy for the treatment of their PMD (52%; Supplemental Table 1). Lastly, patients with a PMD were more often female (57% versus 50%; P < 0.001) and older at FL diagnosis (median age 69 versus 60 years; P = 0.005), as compared to patients without a PMD. The specific subtypes of PMDs and SPMs are presented in Figure 2.

Association between a PMD and SPM development
The 5-year cumulative incidence of SPMs was 5.8% (95% CI, 5.3%-6.4%) and 10.1% (95% CI, 7.6%-13.2%) for patients without and with a PMD, respectively ( Figure 3A). In the univariable Fine and Gray regression model with PMD regarded as a binary variable, a PMD was associated with an increased incidence of SPMs (SHR, 1.44; 95% CI, 1.15-1.80; P = 0.001; Table 2). Subgroup analyses revealed a higher incidence for carcinomas of the respiratory tract (SHR, 1.91; 95% CI, 1.16-3.15; P = 0.011) and squamous cell carcinomas (SHR, 2.02; 95% CI, 1.31-3.12; P = 0.001; Supplemental Table 2). Adjustment for baseline characteristics did not largely affect the SHRs for the overall (Model 1; Table 2) and subgroup analyses (Supplemental Table 2). Furthermore, multivariable analyses showed that male sex and age at FL diagnosis per ten-year increase were independently associated with a greater cumulative incidence of SPMs (Model 1; Table 2). SPM incidence was not influenced by the year of FL diagnosis per one-year increase and the disease stage of FL (Model 1; Table 2).
We specifically assessed the contribution of systemic therapy and/or radiotherapy for a PMD on SPM development. The 5-year cumulative incidence of SPMs after FL diagnosis was 10.4% (95% CI, 6.8%-15.1%) and 9.9% (95% CI, 6.5%-14.3%) for patients with a PMD

Number of malignancies
Prior malignancy diagnosis Second primary malignancy F I G U R E 2 Types of prior and subsequent malignancies among patients with follicular lymphoma. The absolute number of prior and subsequent malignancies according to type is also presented in Supplemental Table 3 who were previously treated and not treated with systemic therapy and/or radiotherapy, respectively ( Figure 3B). The univariable Fine and Gray regression model demonstrated that the association of a PMD with an increased incidence of SPMs was irrespective of whether a PMD was treated with systemic therapy and/or radiotherapy (Table 2).
However, when adjusted for potential confounding factors, only patients with a PMD who were previously treated with systemic therapy and/or radiotherapy-as compared to those without a PMD-had a statistically significant increased incidence of SPMs (SHR, 1.40; 95% CI, 1.02-1.93; P = 0.039; Model 2; Table 2). Besides, the remaining covariates associated with the cumulative incidence of SPMs in Model 2 were comparable to those observed in Model 1 ( Table 2). Sensitivity analyses showed results that were comparable to the results of the primary analyses (data not shown). patients with FL died. Five-year overall survival was 73% (95% CI, 72%-74%) and 62% (95% CI, 58%-66%) for patients without and with a PMD, respectively (P < 0.001; Figure 4A). In the univariable Cox model with PMD regarded as a binary variable, the risk of mortality was higher in patients with a PMD, as compared to patients without a PMD (HR, CumulaƟve incidence  Table 3). Male sex, age per tenyear increase, and higher disease stage at diagnosis were independently associated with a higher risk of mortality, whereas the year of FL diagnosis per one-year increase was associated with a lower risk of mortality.

Years aŌer diagnosis
Five-year overall survival was 61% (95% CI, 54%-67%) and 63% (95% CI, 57%-69%) for patients with a PMD who were previously treated and not treated with systemic therapy and/or radiotherapy, respectively (P = 0.001; Figure 4B). In the univariable Cox model where patients with a PMD were broken down according to the receipt of systemic therapy and/or radiotherapy, the risk of mortality was higher for patients with a PMD-as compared to those without a PMDirrespective of whether a PMD was treated with systemic therapy and/or radiotherapy (  Table 3). Additional factors that were associated with mortality in Model 2 were comparable to those observed in Model 1 (Table 3). Sensitivity analyses again showed results that were comparable to the results of the primary analyses (data not shown).

DISCUSSION
In this nationwide, population-based study, we demonstrated that FL patients with a PMD had an increased incidence of SPMs-in particular of carcinomas of the respiratory tract and cutaneous squamous cell carcinomas-as compared to patients without a PMD. Also, patients with a PMD had a higher adjusted risk of mortality, as compared to patients without a PMD. The increased incidence of SPMs and the higher risk of mortality likely resulted, in part, from therapy-related carcinogenesis. To our knowledge, our study is the first to assess the association of a PMD with SPM development and mortality in FL.
Second, an increased incidence of skin and lung cancer is suggested to be related to exposure to radiotherapy, in particular in combination with systemic therapy [12,13]. This phenomenon is analogous to what has been observed among patients with Hodgkin lymphoma treated with radiotherapy [37][38][39].
To build upon the potential etiologies discussed earlier, immunosuppression and the late effects of systemic therapy and/or radiotherapy might also explain the excess risk of lung and skin cancer among FL patients with a PMD, as compared to those without a PMD.
First, patients with a PMD might have prolonged immune dysfunction related to a PMD and its treatments. Second, the carcinogenic effect of systemic therapy and/or radiotherapy is dose-dependent [26,38,40,41]. Thus, whenever a PMD was treated with systemic therapy and/or radiotherapy, the cumulative dose of potential carcinogens TA B L E 3 Cox regression models for the association between a history of malignancies and mortality among follicular lymphoma patients in the Netherlands In summary, in this nationwide, population-based study, FL patients with a PMD had an increased incidence of SPMs -particularly carcinomas of the respiratory tract and cutaneous squamous cell carcinomasand a higher risk of mortality, as compared to patients without a PMD.
The mechanism behind this was likely multifactorial, albeit our data suggest that it may have resulted, in part, from therapy-related carcinogenesis. As the longevity of patients with FL is expected to increase, physicians should be aware of SPMs within this patient population, especially among patients with a PMD who were treated with systemic therapy and/or radiotherapy. We encourage forthcoming studies to validate our study findings through analysis of population-based cancer registry data.

ACKNOWLEDGMENTS
The authors would like to thank the registration clerks of the Netherlands Cancer Registry (NCR) for their dedicated data collection. The nationwide population-based NCR is maintained and hosted by the Netherlands Comprehensive Cancer Organisation (IKNL).

CONFLICT OF INTEREST
MJK has received research and travel support, as well as honoraria for presentations from Roche.

AUTHOR CONTRIBUTIONS
AGD designed the study; MAWD analysed the data; OV collected the data; MAWD wrote the manuscript with contributions from all authors, who also interpreted the data, and read, commented on, and approved the final version of the manuscript.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available via The Netherlands Comprehensive Cancer Organisation. These data are not publicly available and restrictions apply to the availability of the data used for the current study. However, these data are available from the