Incidence, risk and survival outcomes of second primary malignancy among renal cell carcinoma survivors: A nested case-control study

Background: Second primary malignancy (SPM) challenges survival and surveillance protocols among renal cell carcinoma (RCC) survivors. The incidence, temporal patterns, survival outcomes, and risk factors of SPM after T1-4N0-1M0 RCC diagnosis need to be investigated. Method: A nested case-control study that was designed using the Surveillance, Epidemiology, and End Results database from 2004-15; A cohort of 6204 SPM were matched with a control group of 37224 non-SPM. Results: SPM shortens the overall survival (hazard ratio [HR], 1.34; 95% condence interval [CI]: 1.28-1.42, P< 0.001). The median time interval to SPM was 54.5 months. The adjusted standardized incidence ratio (SIR) of SPM increases by survival time (SIR 12~35-month : 12.04; SIR 36~59-month : 12.67; SIR 60~19-month : 16.08; SIR 120+-years : 25.01, all P< 0.001), and decreased with age (SIR 18~44-years : 86.68; SIR 45~59-years : 26.95; SIR 60~74-years : 12.43; SIR 75+-years : 10.66, all P< 0.001). The second primary RCC onset, especially contralateral kidney, has the highest SIR (SIR: 54.6; 95%CI: 51.0~58.4) among all sites of SPM. Prostate cancer (29.8%) in male and breast cancer (23.5%) in female were the most common SPM site. Older age, black race, male gender, higher family income statues, papillary RCC, and lower TNM stage signicantly increases the risk of SPMs diagnosis. A longer time to SPM interval positively associated with a higher tumor stage of a SPM onset (P trend <0.001). The overall survival since the SPM diagnosis was associated with SPM’s stages, site, and surgical treatment, but not associated with time-to-SPM. Conclusion: Collectively, our study described the epidemiological characteristics of SPM among RCC survivors and identied the independent predictors of the SPM onset and its survival outcomes, which provides the clinicians for patients consulting and long-term individual-, tailored site-, and time-specic surveillance to improve survival outcomes. region. We also collated a range of data related to tumors, as follows: tumor size (cm), histological cell type for RCC (clear cell, and non-clear cell [papillary and chromophobe]); tumor grade (well-differentiated [grade I], moderately differentiated [grade II], poorly differentiated [grade III], and undifferentiated [grade IV]); and American Joint Committee on Cancer (AJCC) 6th tumor node metastasis (TNM) staging classication. The surgical intervention included partial or radical nephrectomy, cryosurgery, or radiofrequency ablation. range; CI:Condence interval; HR:Hazard ratios; OS:Overall survival; SEER:Surveillance, Epidemiology, and End Results; NCCN:National Comprehensive Cancer Network; EAU:European Association of Urology; AUA:American Urological Association.

Introduction selected by the SEER *Stat software (version 8.3.6). All data obtained from the SEER database were anonymized and were not associated with research studies. Consequently, the need for ethics approval was waived by the Institutional Review Board.

Patient identi cation and study variables
Using the SEER database, we prospectively identi ed and extracted patients who had been treated for primary RCC but then went on to develop SPM during follow-up. These patients were extracted from the MP-SIR session and were considered as the case group. Patients who did not develop SPM were considered as the control group. Figure 1 shows a owchart that describes how patients and data were selected from the SEER database. If the selected individuals had experienced multiple incidences of SPM, we only retrieved data relating to the rst occurrence of SPM. We ensured that the patients who had been treated for their rst primary RCC did not have any other metastatic diseases, were aged ≥ 18 years, and had undergone surgical treatment involving cryosurgery/radiofrequency ablation (RX Summ-Surg Prim Site codes 13, 15, and 23, in the SEER database) and/or partial/radical nephrectomy (RX Summ-Surg Prim Site codes 30, 40, 50, 70, and 80, in the SEER database). Furthermore, all diagnoses of RCC needed to have been nalized by histology and not by autopsy or death certi cation only. Tumor diameter needed to be no greater than 15 cm and SPM needed to have been diagnosed within 6 months of the diagnosis of primary RCC. Patients with pre-existing or synchronous tumors at the diagnosis of RCC were excluded. The primary cancer site was identi ed by referring to The International Classi cation of Diseases for Oncology, 3rd Edition (ICD-O-3) [12]. RCC was identi ed by the ICD-O-3 code C64.9.
We collated a wide range of demographical and clinical variables, as follows: year of diagnosis, age at diagnosis, gender, race or ethnicity (White and Others [Black, American Indian/Alaska Native, Asian Native, and Asian/Paci c Islander]), marital status, family income quartile, population, and region. We also collated a range of data related to tumors, as follows: tumor size (cm), histological cell type for RCC (clear cell, and non-clear cell [ and American Joint Committee on Cancer (AJCC) 6th tumor node metastasis (TNM) staging classi cation. The surgical intervention included partial or radical nephrectomy, cryosurgery, or radiofrequency ablation.

Outcomes for analysis
The primary objectives of this study were to investigate the risk factors and survival outcomes associated with SPM following a primary diagnosis of RCC. Therefore, the main outcomes of interest were (i) SPM diagnosis and the time duration for SPM to develop following a diagnosis of RCC, and the strati cation of SPM as kidney cancer, including contralateral and ipsilateral kidney cancer and other non-kidney malignancies; and (ii) overall-mortality, RCC-mortality, and other causes of mortality in patients with RCC. From the SEER database, we were able to collate data relating to the sequence of multiple primary malignancies and the time duration associated with their occurrence; we could also identify the speci c causes of death. Patients who died from RCC were identi ed as RCC-cause mortality, those who died from other causes were designated as 'competing events' prior to RCC-cause mortality. Any other cause of death was considered as all-cause mortality; this was considered as a competing event for the occurrence of SPM. The time interval between RCC diagnosis and SPM, and the duration of survival, were de ned as the time elapsed from the date of RCC diagnosis to the date of SPM diagnosis, and death or last contact, respectively.

Statistical analysis
Continuous variables are described as means ± standard deviations (SDs) if normally distributed and compared with the Student's t-test. Continuous variables that were not normally distributed are described as medians and interquartile range (IQR) and compared with the Wilcoxon rank-sum test. Categorical variables are presented as frequencies (%) and were compared with the chi-squared test. The reverse Kaplan-Meier method was used to calculate the median follow-up time. To control bias related to case selection between cases and control groups, we performed propensity score matching (1:5; cases: controls) using the 'nearest-neighbor' method and the 'MatchIt' package in R. The year in which the primary RCC diagnosis was made, and the rst SEER registries of the primary RCC, were designated as adjusted covariables and a multivariable logistic regression model was used to calculate the propensity score for each patient so as to maintain the case and control cohorts diagnosed during the same latency period, and to control for differences across different registries. The adjusted standard incidence ratio (SIR), along with its 95% con dence interval (CI), was calculated as the ratio of SPM in patients diagnosed with primary RCC to the number of expected events in the general population. SEER*Stat version 8.3.6 software was used for this analysis as it could compare the relative risk of SPM for RCC patients as compared with the general population. Age, time interval, and tumor site, were adjusted for the SIR calculation and the statistical signi cance of SIR was assessed using a likelihood ratio test. Since all-cause mortality could prevent the occurrence of SPM and was considered as a competing event, we used a Fine and Gray competing risks regression model to evaluate the risk factors for the rst occurrence of SPM (all SPM, contralateral-and ipsilateral-metachronous kidney cancer, and other non-kidney cancers), and calculated the hazard ratio (HR) with 95% CIs for all risk factors. For Fine and Gray competing risks regression analysis, we de ne the time interval as the date from the primary RCC diagnosis to the earliest date of the SPM diagnosis (event observed) or the last follow-up in December 2017 (censored). For each Fine and Gray competing risk regression analysis, we compared case and control groups after propensity score matching (using a 1:5 ratio). When investigating the risk factors associated with survival outcomes, we used Cox proportional hazards regression to analyze the risk factors for overall mortality from the rst RCC diagnosis and the rst SPM diagnosis. Subsequently, we included all variables in multivariate analysis. All analyses were conducted using the R statistical package (v.3.6.3; R Foundation for Statistical Computing, Vienna, Austria; https://www.r-project.org). All p values are two-sided, and P < 0.05 indicates statistical signi cance.

Patients and baseline characteristics
From 2004 to 2015, a total of 93,406 patients (aged over 18 years-of-age) were diagnosed with rst primary T1-4N0-1M0 RCC in the 18 registries of the SEER database. During follow-up to December 2017, 13,347 of these patients (14.3%) reported two or more subsequent cancer diagnoses. After screening patients with our inclusion and exclusion criteria (Fig. 1), we were able to identify a case-control cohort (SPM vs. non-SPM) with 6,204 patients (16.7%) in the case group and 31,020 patients (83.3%) in the control group after propensity score matching (1:5) for the year of diagnosis and different registers.
It is noteworthy that contralateral RCC was a common malignancy following the primary RCC diagnosis in several different subgroups, accounting for 12.2% of males, 10.9% of females, 10.3% of white patients, 22.1% of black patients, and 9.8% of other races. When considering different follow-up periods (< 1 year, 1-2 years, 3-5 years, and > 5 years), there were no obvious changes with regards to the proportion (%) of each type of SPM ( Supplementary  Fig. 2). We observed clear histological changes associated with SPM in the kidney when compared with the primary RCC ( Supplementary Fig. 3 The time interval to SPM Figure 2A and Supplementary Fig. 4 present the distribution of the time taken to diagnose SPM in all cohorts strati ed by gender and different SPM sites. For the entire cohort, the median time to SPM onset after the rst primary diagnosis was 54.5 months, 53.0 months, and 61.0 months, in all sex, male and female patients, respectively. With regards to different SPM sites, we found that SPM of the ipsilateral and contralateral kidneys had the shortest median time interval to SPM: 48.0 and 49.0 months, 41 and 46.5 months, and 21 and 31 months, in the entire cohort, the male cohort, and the female cohort, respectively. SPM of the ovary (79.5 months) and bladder (69 months) was associated with the longest time interval to SPM. Another interesting nding was that tumor stage, tumor size, and TNM stage of SPM were signi cantly associated with the time interval to SPM (Fig. 2B). A larger tumor size (P trend <0.001), more regional and distant metastasis (P trend =0.012), and stage III/IV tumors (P trend <0.001) were signi cantly associated with an increase by the time interval to SPM.
The age-, time interval-, and site-speci c SIR for SPM Risk factors for SPM following primary RCC diagnosis

Survival and risk predictors following the diagnosis of primary RCC and SPM
At the nal follow-up (crude median: 76 months), we found that 4.7% and 7.3% of patients had died from other forms of malignancy and RCC-speci c malignancy, respectively (results from unmatched data; data not shown). Survival analysis showed that the 5-year OS for RCC patients who experienced SPM was 85.9% (95% CI: 84.9-87.2%) from the primary RCC diagnosis, and 58.9% (95% CI: 57.1-60.8%) from the SPM diagnosis (Fig. 3). The development of SPM in the pancreas, brain, gallbladder/bile duct, liver, miscellaneous tissues, esophagus/stomach, and lungs/bronchi, was associated with poor survival outcomes, with a 5-year OS of 13.4%, 13.5%, 17.7%, 18.0%, 19.6%, 22.8%, and 24.2%, from SPM diagnosis, respectively. Similar results were obtained for the female and male cohorts ( Supplementary Fig. 5).
Following the adjustment of other covariables by multivariable Cox regression, we found that when compared with non-SPM patients, those with SPM were associated with a signi cant impact in terms of OS following the primary RCC diagnosis signi cant differences in the risk of mortality when compared between cases of SPM in the cecum/small intestine, female breast, and thyroid, when compared to ipsilateral cases. Interestingly, we found that male patients with SPM in the prostate were associated with a lower risk of mortality than those with SPM in the ipsilateral kidney (HR: 0.43; 95% CI: 0.22-0.83; P = 0.011). We also investigated the effects of SPM tumor stage and the site of SPM on OS.   ) differ in terms of their recommendations for imaging modalities and frequencies [12]. Although there is no consensus with regards to an optimal surveillance strategy for patients after RCC treatment, it is often recommended that patients are followed-up closely for 3-5 years after surgery since relapse is most commonly seen within the rst 5 years of surgery. Another common recommendation is to select different examination modalities based on individual risk strati cation. The number of follow-up assessments can be reduced after the rst 3 years but should continue for at least 5 years [13,14]. In our study, it suggested that it's important that long-term follow-up should not be stopped; rather, the surveillance modality and frequency should be adjusted based on individual risk.
Interestingly, the contralateral kidney showed the highest SIR (54.6) and accounted for the largest proportion of SPM cases (11.8%) in survivors of primary RCC. This indicates that it is crucial to perform lifelong follow-up assessments for RCC survivors, and that we pay particular attention to contralateral kidney carcinoma. Furthermore, it is important to routinely screen general patients or RCC survivors for malignancies associated with the prostate, breast, lung, bladder, colon/rectum, and thyroid [1]; these types of tumors accounted for most of the SPM cases seen in our current study. Although our results indicated that the early occurrence of SPM can exert impact on the OS, it was evident that survival can be prolonged if new cases of SPM are detected and treated early. Moreover, we found that if SPM cases were detected later, or involved a higher stage of solid SPM, then curative surgical treatment would be di cult and threaten long-term OS. The early detection of a small local SPM in the contralateral kidney would bene t from PN, particularly in patients who experienced RN for their primary RCC. Conversely, considering the risk of contralateral kidney cancer following surgery for primary RCC, it is recommended that we perform PN for the rst kidney surgery in order to retain renal function, even if a contralateral SPM occurs. The early detection of a local small RCC is critical and could bene t from PN, and this technique may protect the normal renal parenchyma [15][16][17][18].
With regards to the design of individual surveillance plans, the selection of patients at risk could help to balance follow-up bene ts and save costs, thus allowing the close monitoring of high-risk patients while reducing the over-medical treatment of low-risk patients. In the present study, we identi ed a series of independent high-risk predictors for the occurrence of SPM: increased age at the primary RCC diagnosis, black race, male gender, patients with high family income, papillary types of cancer, small tumors, the absence of lymph node invasion, a lower T stage, and a lower AJCC TNM stage. Our results further suggest that although some patients, such as those with a low TNM stage of RCC, may represent a low-risk group for recurrence after RCC [19], further follow-up is still necessary because this group is at high risk of developing SPM. Furthermore, the emergence of SPM is related to gender and the pathological type of RCC, thus indicating that SPM may be related to certain genes or other exposure environmental. Interestingly, SPM is known to be related to economic factors, and may be more prevalent in patients with a certain nancial status; some patients may pay more attention to their follow-up and general health status and also have the nancial means to afford the costs associated with follow-up.
Age is an independent risk factor for the occurrence of SPM. Interestingly, when predicting non-RCC SPM, the risk of SPM increases with age. However, when considering contralateral kidney tumors, it is evident that the elderly patients do not have an increased risk. Our univariate analysis found that the risk of patients over 60 years-of-age was lower for contralateral kidney SPM than that of patients under 44 years-of-age; this was consistent with previous literature [20,21]. Age itself is an independent risk factor for cancer [22], and because the risk of developing tumors increases with age. As long as patients survive long enough, they are likely to develop more than two types of malignant tumors in their lives. It has been reported that 33% of cancer survivors over the age of 60 will be diagnosed with another type of cancer [23]. For some patients, RCC is only one of the earliest malignant tumors that could develop; however, with improvements in the prognosis and survival of RCC patients, it is now possible to detect SPM during the follow-up period. In elderly patients with RCC as their rst cancer, SPM may be recorded shortly after RCC diagnosis. In contrast, younger patients may need a longer follow-up time for SPM to be detected, perhaps because the follow-up time required is relatively longer (in other words, the patients need to reach a certain age for successful detection). It is therefore possible that SPM was not detected by the end of the follow-up period in this study, or that the occurrence of a variety of competitive events prevented the occurrence of SPM.
The increased risk of secondary SPM in RCC patients still needs further veri cation and in-depth research in order to identify the precise connection between these two conditions. Possible mechanisms include renal insu ciency, and even renal failure after RCC, thus leading to long-term metabolic disorders and an increased risk of cancer in patients surviving from RCC. Furthermore, RCC patients may have experienced frequent episodes of computer tomography to monitor the disease, thus causing increased levels of exposure to radiation, or changes in the complex immune microenvironment. The biological characteristics of RCC confer all patients with kidney cancer with a higher risk of SPM. Renal cell carcinoma is also related to smoking [24]; however, smoking does not just affect the kidney. There may be a higher risk of cancer in patients with RCC who did not give up smoking after treatment, or in patients who smoked frequently over a long period prior to RCC diagnosis. However, these are hypotheses that have yet to be con rmed.
Furthermore, we analyzed the survival function for patients with SPM. We found that the occurrence of SPM, and an early time interval to SPM, could signi cantly shorten the OS when compared with patients who did not have SPM. However, the same site of SPM was not signi cantly associated with outcomes; for example, SPM of thyroid cancer, ipsilateral RCC, and prostate cancer, all showed an excellent 5-year OS after the diagnosis of SPM was made. After adjusting for other risk factors, we also found that SPM of contralateral RCC, cecum/small intestine, female breast, thyroid, and prostate cancer, had a similar or signi cantly better OS when compared with ipsilateral RCC. A worse OS was associated with SPM of the brain, liver, gallbladder/bile duct, lungs/bronchi, pancreas, and esophagus/stomach; furthermore, these sites of SPM also had a higher SIR compared with the general population. Another interesting nding was that the stage of SPM was also signi cantly associated with OS following the diagnosis of SPM; an SPM of a local disease had a better OS than regional and distant SPM. In addition, patients with SPM also could bene t from surgical treatment. These results clearly suggested that SPM can in uence the OS. However, early detection and surgical intervention for solid SPM tumors is of great signi cance and can in uence outcomes; this emphasizes the need to establish lifelong follow-up strategies rather than performing surveillance over the short-term.
This study had same limitations that need to be considered. First, all analyses were carried out using a registry-based dataset with inherent limitations. For example, we did not have access to any detailed clinical information such as patient comorbidities or a poor performance that lead to mortality within a period of follow-up time. This represents a vital competing risk for the occurrence of SPM and could not be adjusted in our multivariate analysis for SPM prediction. In addition, when analyzing the prediction of SPM for ipsilateral and contralateral kidney cancer, we had no access to information relating to hereditary RCC, such as von Hippel Lindau (VHL), hereditary papillary renal carcinoma, Birt Hogg Dube´ syndrome, and hereditary leiomyomatosis RCC [25]; these factors have all been identi ed previously as signi cant predictors for the metachronous de novo development of RCC over a long-term follow-up period. We did not have access to information relating to environment exposure, lifestyle, family history, and genetic mutation, all of which are known to act as risk factors for SPM. Second, patients with primary RCC or other diseases would generally pay more attention to routine cancer screening or surveillance than the general population, thus increasing the chances of identifying the occurrence of SPM in RCC survivors. Therefore, it is possible that surveillance bias may exist in our study. Third, after the rst primary diagnosis of RCC, it is possible that the preexisting or concomitant malignancy, metastasis diseases, or relapse in the patients with RCC, may have confounded the subsequent detection of SPM. To control for confounding factors, we only included patients with SPM that had been diagnosed at least 6 months after the rst primary RCC diagnosis as our study cohort. When using the SEER dataset, we maintained quality assurance by performing systematic and standardized data collection procedures. Finally, due to a lack of information relating to local recurrence, we included patients with ipsilateral kidney cancer; these cases may have just been recurrences rather than SPM. However, more than 90% of ipsilateral kidney tumors were diagnosed more than 3 years after RCC diagnosis; 50% were diagnosed after more than 5 years. Furthermore, 35.6% of patients with RCC showed histological changes between the rst and second occurrence of ipsilateral kidney cancer.

Conclusion
Our analyses demonstrated a higher incidence of SPM among RCC patients surviving over the long-term when compared with the general population. We also found that age at the diagnosis of rst primary RCC, race, gender, economic status, histological type of RCC, and tumor stage, were all signi cantly associated with the occurrence of SPM. Tumor stage and the site of SPM showed particularly strong associations with OS. Lifetime follow-up and cancer screening is therefore recommended for RCC survivors. Although the occurrence of SPM can threaten long-term survival, patients with low grade/early-stage SPM could bene t from aggressive surgical treatment for solid tumors. According to the patient age, and the time interval after RCC diagnosis, it is also worthwhile monitoring high-risk RCC patients by site-and time-speci c surveillance strategies; however, needs of balancing cost-effectiveness and Lifelong surveillance.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.