Better cancer treatment and early detection have increased survival rates among patients with cancer. But some cancer survivors can develop a second cancer called a second primary cancer. Second primary cancers may occur months or years after successful treatment of the primary cancer. They are not caused by the spread of the original tumor like a cancer metastasis. Instead, they appear to occur independently in another location or tissue.

Scientists are trying to understand what causes second primary cancers. Genetics, lifestyle, the environment, treatments used for the initial tumor, or other factors may all contribute to individuals developing a second cancer. Learning more about who is at risk of developing a second cancer and why, may lead to new prevention, treatment or screening strategies.

Ruan, Huang et al. found that people with some primary cancers have an increased risk of secondary primary cancers in specific tissues. The researchers first looked at the Surveillance, Epidemiology, and End Results (SEER) database that tracks US cancer patients to see if different types of cancers were more likely to lead to a second primary cancer. Then, the team conducted a comprehensive analysis for a causal relationship in a second extensive health database, the UK Biobank, to determine if the primary cancers may have caused the second primary cancer. The study showed that patients diagnosed with mouth or throat cancers were at increased risk of later developing a lymph node cancer called non-Hodgkin lymphoma. Patients diagnosed with ovarian cancer were at increased risk of later developing cancer in one of the body's soft tissues. Kidney cancer is likely the cause of later lung cancers and a type of blood cancer called myeloma.

Understanding the relationships between an initial and later cancer diagnosis is essential to improve cancer survivors' care. It is especially important for patients diagnosed early in life. More studies are needed to confirm the links Ruan, Huang et al. identified and to understand the mechanism. If more studies confirm the associations, physicians may want to screen survivors for specific cancers. Scientists may also be able to use the information to develop new strategies to help prevent or treat secondary primary cancers.

Cancer incidence is rapidly growing worldwide in the past decades. The reasons are complex including the aging of the population, the application of screening, environmental risk factors and genetic risk factors (Sung et al., 2021). In 2020, there were an estimated 19.3 million new cases and 10.0 million deaths of different types of cancers globally (Sung et al., 2021). In the US, there would be about 1.9 million new cases of cancers in 2023,and the most common type of cancers in the male and the female were prostate cancer (29%) and breast cancer (31%), respectively (Siegel et al., 2022; Siegel et al., 2023). Despite the rapidly increased incidence of cancers, the survival of (most types of) cancers in the US has vastly improved since the mid-1990s with medical advances and technical developments (Siegel et al., 2022; Siegel et al., 2023). For example, the 5-year relative survival rates of prostate cancer, melanoma, and female breast cancer, during 2011–2017 in the US were 98%, 93%, and 90%, respectively (Siegel et al., 2022). Such disparity may lead to an increase in the prevalence and the tumor burden in US society. More importantly, prolonged survival makes it possible for individuals to be diagnosed with a second primary cancer (SPC) after the first primary cancer (FPC) during the follow-up.

According to the Italian (1976–2010; AIRTUM Working Group, 2013), the Swiss (1981–2009; Feller et al., 2020), and the Swedish (1990–2015; Zheng et al., 2020) cancer registration data, increased risks of SPCs were observed in many types of cancer as the FPCs. Patients with oral cavity & pharynx, larynx, and esophagus as FPC were found to have a significantly elevated risk of any SPCs in both Italy and Switzerland (Feller et al., 2020; AIRTUM Working Group, 2013). In Sweden, liver cancers, as well as nasal and oral cancers, were found to be associated with SPCs (Zheng et al., 2020). In addition, several studies focused on certain cancer also suggested a potential relationship between FPC and SPC (Chattopadhyay et al., 2018). For example, increased FPC risk of colorectal cancer, kidney cancer, and melanoma were observed following the diagnosis of non-Hodgkin lymphoma. Despite the strong association observed in these studies, whether there is any underlying causal relationship is unknown, or the association observed is due to the potential confounders such as aging.

In the present study, our objectives are to perform a pan-cancer association study in the Surveillance, Epidemiology, and End Results (SEER) and to interpret the underlying causal relationship via Mendelian Randomization approaches using genetic variants in a large population cohort (UK Biobank, UKB). The study may help us understand critical questions in clinical practice about who should be more careful of the second primaries. In addition, precision screenings against certain cancers should also be considered among those patients with increased risk of second primaries in addition to the regular follow-up evaluations. In addition, many cancers are known to have a multifactorial etiology and some cancer treatments are known to be carcinogenic. It is essential to illustrate the potential relationship between FPCs and SPCs which mary optimize the treatment for patients’ prognosis and survivorship.

We set out to identify the observational association between FPCs and SPCs among 22 solid tumors and 4 hematological malignancies based on the SEER database (Figure 1). A total of 264 significant standardized incidence ratio (SIR) existed among them when compared to a standard population (Supplementary file 1a-1q). Hierarchical clustering analysis (heatmap) is shown in Figure 2 which illustrates the comparison between the cancer incidence in patients with a certain type of FPC and the incidence in the population average level in the SEER dataset. Second primary thyroid cancer, small bowel cancer, or cancers of endocrine system were strongly and significantly associated with most of the FPCs. Cancers of the digestive system, cervix uteri, as well as lung cancer were clustered together. They are closely correlated with an increased risk of several types of cancers such as bladder cancer, kidney cancer, etc. Our subgroup analysis suggested that patients with prostate cancer who received radiation therapy were at an increased risk of being diagnosed with another type of primary cancer during the follow-up including small intestine, soft tissue, and leukemia, compared with those without such therapy (Supplementary file 1m). And SPC risk after breast cancer was inconsistent among males and females. Men with breast cancer had a higher risk for thyroid cancer and prostate cancer (SIR_SEER = 1.30; 95% CI, 1.13–1.49; SIR_SEER = 2.33, 1.12–4.29; Supplementary file 1j), but with no significant risk change for other cancers.

Figure 2

Download asset Open asset

Details of the included exposure-associated SNPs in European ancestry were shown in Table 1 and Supplementary file 2a. The number of SNPs ranged from 5 (esophagus) to 104 (female breast), and the proportion of variance explained by SNPs (R²) ranged from 0.10% (thyroid) to 8.46% (bladder). F-statistics for all 13 cancers exceeded 10, suggesting no weak instrument bias here. However, some problems, including too few SNPs, no related GWAS and disease heterogeneity, affected the comprehensive MR analysis (Table 1). Due to the low incidence of cancers in UKB, the power to detect a significant effective size (0.8/1.2) was relatively low, except for colorectal cancer, lung cancer, female breast cancer and prostate cancer (Supplementary file 2b).

Results from MR analyses are presented in Supplementary file 2c-2p. A total of 23 significant association was detected (16 positive causality and 7 negative causality). Concordant significant results and unconcordant results between MR and the SEER SIR analyses are shown in Table 2 and Supplementary file 2c, respectively. The concordant results suggested that patients diagnosed with primary oral and pharynx cancer would cause a significantly increased risk of second primary non-Hodgkin lymphoma (SIR_SEER = 1.18, 95% CI: 1.05–1.31; IVW-MR P=8.96 × 10^–4). After a primary diagnosis of pancreatic cancer, SPC risks were increased for small intestine (SIR_SEER = 4.37, 95% CI: 2.85–6.40; MR-Egger p=0.04). It also indicated that female patients initially diagnosed with melanoma would cause a mild but significantly increased risk of cancers in the breast (SIR_SEER = 1.17, 95% CI: 1.12–1.23; IVW-MR p=0.04), as well as ovary cancer on soft tissue related cancer (SIR_SEER = 1.72, 95% CI: 1.08–2.60; IVW-MR p=8.39 × 10^–5). The greatest number of casual relationships were observed in kidney cancer as FPC. A primary kidney cancer might cause an elevated risk of cancers of lung and bronchus (SIR_SEER = 1.28, 95% CI: 1.22–1.35; IVW-MR p=0.01), non-Hodgkin lymphoma (SIR_SEER = 1.19, 95% CI: 1.08–1.31; IVW-MR p=3.64 × 10^–3), myeloma (SIR_SEER = 1.54, 95% CI: 1.33–1.78; IVW-MR p=3.94 × 10^–3). Meanwhile, some primary cancer site might give protective effect against another cancer (for instance, lung cancer vs. female breast cancer, Table 2). Scatter plot, Funnel plot, forest plot and leave-one-out analysis showed single SNP effective size in Figure 1—figure supplements 1–4, respectively.

More conservative analyses were performed to further confirm these causal relationships. We applied 2 outlier-detected methods with modified second order weights (radial IVW and radial MR-Egger) and CAUSE (Causal Analysis Using Summary Effect estimates) to each pair of phenotypes, with the rationale that robust relationships would exhibit consistent and statistically significant results across different methods. Additionally, CAUSE is the only method capable of distinguishing causality from both correlated and uncorrelated pleiotropy. The relationship with at least two significant results was treated as a robust causality (Table 3). We found consistent evidence for a causal effect of oral and pharynx cancer on non-Hodgkin lymphoma (P_radial-IVW=6.00 × 10^–3, P_cause = 8.90 × 10^–3), ovary cancer on soft tissue cancer (P_radial-IVW=1.07 × 10^–3, P_cause = 0.01), kidney cancer on lung and myeloma (P_radial-IVW=6.60 × 10^–3, P_cause = 0.05; P_radial-IVW=0.02, P_cause = 0.02).

With the expanded life expectancy and the prolonged survival of cancers, the incidence of SPCs has been rapidly growing in the past decades (Copur et al., 2019). Genetic factors or shared environmental factors are probably the major causes. In the previous association studies, individuals with a certain type of primary malignancy would have an increased risk of another malignancy (Feller et al., 2020; AIRTUM Working Group, 2013; Zheng et al., 2020). However, whether there are any causal effects within the associations is unclear. In the present study, via the association study based on the SEER database and the MR approach using the UKB genetic dataset, we were able to perform this comprehensive investigation across 26 different types of cancers. 13 out of 26 types of cancers with adequate GWAS data were able to be further investigated using MR analysis. We found that numbers of primary malignancies were associated with an increased risk of a second primary malignancy, however, only a small part of the associations would have a causal relationship (Table 2).

Many significant findings were observed in the SEER SIR analysis. SEER is one of the largest cancer registration-based datasets making itself the most proper data source to answer the study objectives; however, several advantages of this database and limitations of the results should be noted. First, the results from the SEER SIR analysis were associations rather than causal inferences. Many factors may influence the results of the associations. For example, the confounder of screening effects may exist. Patients diagnosed with primary cancer might have more frequent healthcare visits compared to those who did not have any cancers. Therefore, some indolent cancers such as thyroid cancer and low-risk prostate cancer could have been over-diagnosed due to the screening effects. Second, it is a cancer registration-based cohort rather than a population cohort, the standardized incidences calculated from the SEER database may not represent the situation in the general population. Third, treatment preferences and follow-up strategies may vary in different locations or institutions, which would also affect the occurrence/detection of the second primary cancer. For example, radiation therapy may increase the risk of cancer in nearby organs⁶. However, treatment and follow-up information are not completed in the SEER database due to the natural design of the cohort³¹. More importantly, lifestyles, comorbidities, and environmental factors were not included in the SEER database. These are important confounders of the associations between the first primary cancers and the second primary cancers.

The MR approaches in the current study revealed some interesting findings, but several non-concordant results between MR analyses and SEER SIR were also observed (Table S20). It does not indicate that the causal relationship does not exist. Some factors, such as the period of follow-up, may conceal and cause false negatives in the association study (SEER SIR). For example, pancreatic cancer might cause an increased risk of cancers in esophagus, colon and rectum, etc. based on the MR analysis in our study; however, no association was observed in SEER SIR analysis. The short and poor survival of pancreatic cancer could be the most critical reason for the failure of finding a positive association in the population data --- simply did not have enough time of follow-up to observe the outcomes. Therefore, the interpretation of these results should be more careful at this stage.

The lack of GWAS findings would be a major limitation of the MR approach for some diseases as in the present study. The MR approach may only represent part of the biological effects in the causal pathway between the exposure and the outcome. A final causal inference should always be established based on biological mechanisms. From the angle of organ location, some cancers (ovary cancer and soft tissue sarcoma) might share the same tumor-related or tumor-developing environment. Besides the outside therapeutic settings (radio- and chemotherapy, immuno-suppressive agents) and individual factors (smoking, hormone level, certain occupational settings, HIV or HPV infections, and family histories), FPC might also influence the iatrogenic immune by suppressing antitumor defense mechanisms via inflammation or other meditating effects (Shalapour and Karin, 2019). For instance, the increased risk of renal cell cancer, and non-Hodgkin lymphoma were observed in immunosuppressed patients in Denmark and Sweden (Hortlund et al., 2017). The immune factor might be the inner relation between FPC and SPC. In our assumption, the question of mechanism should be answered via cross-trait GWAS meta-analyses, searching for shared genetic architecture under high heritability or potential meditation factors with comprehensive database and analyses, functional experiments (tissue- or cell-specific findings) and final validation in a cohort of comorbidity patients. In terms of clinical implications, lifestyle, environment, treatment, host factors and other influences may contribute to multiple primary cancer. It required clinicians and researchers to explore a specific link between FPCs and SPCs. It would be of clinical importance to make a personalized screening plan for certain primary cancer patients (eg. oral pharyngeal cancer, ovary cancer and kidney cancer), especially for those young onsets. For example, patients with kidney cancer would have an increased risk of lung cancer and myeloma. Thus, a screening for lung cancer and myeloma should be recommended among those patients.

Finally, besides the limitations mentioned, the relatively small number of cases of some diseases in UKB may lower our statistical power. As UKB is a population-based prospective cohort, the relatively short follow-up period may not allow us to observe enough events (multiple cancers) at this stage. And it is expected to be independently replicated in another dataset. Regardless, these findings progress our understanding of the relationship underlying both FPCs and SPCs, and potentially provide points of exploration and intervention that may reduce second primary cancers.

Data used in this research are publicly available to qualified researchers on application to GWAS catalog (https://www.ebi.ac.uk/gwas/), the SEER database (https://seer.cancer.gov/) and UK Biobank (https://www.ukbiobank.ac.uk/). Source code for all analyses were in related software or R packages.

1. 1. AIRTUM Working Group

   (2013)

   Italian cancer figures, report 2013: Multiple tumours

   Epidemiologia E Prevenzione 37:1–152.

   • PubMed
   • Google Scholar
2. 1. Bowden J
   2. Spiller W
   3. Del Greco M F
   4. Sheehan N
   5. Thompson J
   6. Minelli C
   7. Davey Smith G

   (2018) Improving the visualization, interpretation and analysis of two-sample summary data Mendelian randomization via the Radial plot and Radial regression

   International Journal of Epidemiology 47:1264–1278.

   https://doi.org/10.1093/ije/dyy101

   • Google Scholar
3. 1. Bowden J
   2. Del Greco M F
   3. Minelli C
   4. Zhao Q
   5. Lawlor DA
   6. Sheehan NA
   7. Thompson J
   8. Davey Smith G

   (2019) Improving the accuracy of two-sample summary-data Mendelian randomization: moving beyond the NOME assumption

   International Journal of Epidemiology 48:728–742.

   https://doi.org/10.1093/ije/dyy258

   • Google Scholar
4. 1. Brion MJA
   2. Shakhbazov K
   3. Visscher PM

   (2013) Calculating statistical power in Mendelian randomization studies

   International Journal of Epidemiology 42:1497–1501.

   https://doi.org/10.1093/ije/dyt179

   • PubMed
   • Google Scholar
5. 1. Buniello A
   2. MacArthur JAL
   3. Cerezo M
   4. Harris LW
   5. Hayhurst J
   6. Malangone C
   7. McMahon A
   8. Morales J
   9. Mountjoy E
   10. Sollis E
   11. Suveges D
   12. Vrousgou O
   13. Whetzel PL
   14. Amode R
   15. Guillen JA
   16. Riat HS
   17. Trevanion SJ
   18. Hall P
   19. Junkins H
   20. Flicek P
   21. Burdett T
   22. Hindorff LA
   23. Cunningham F
   24. Parkinson H

   (2019) The NHGRI-EBI GWAS Catalog of published genome-wide association studies, targeted arrays and summary statistics 2019

   Nucleic Acids Research 47:D1005–D1012.

   https://doi.org/10.1093/nar/gky1120

   • PubMed
   • Google Scholar
6. 1. Burgess S
   2. Thompson SG

   (2017) Interpreting findings from Mendelian randomization using the MR-Egger method

   European Journal of Epidemiology 32:377–389.

   https://doi.org/10.1007/s10654-017-0255-x

   • PubMed
   • Google Scholar
7. 1. Burgess S
   2. Davey Smith G
   3. Davies NM
   4. Dudbridge F
   5. Gill D
   6. Glymour MM
   7. Hartwig FP
   8. Kutalik Z
   9. Holmes MV
   10. Minelli C
   11. Morrison JV
   12. Pan W
   13. Relton CL
   14. Theodoratou E

   (2019) Guidelines for performing Mendelian randomization investigations: update for summer 2023

   Wellcome Open Research 4:186.

   https://doi.org/10.12688/wellcomeopenres.15555.3

   • PubMed
   • Google Scholar
8. 1. Bycroft C
   2. Freeman C
   3. Petkova D
   4. Band G
   5. Elliott LT
   6. Sharp K
   7. Motyer A
   8. Vukcevic D
   9. Delaneau O
   10. O’Connell J
   11. Cortes A
   12. Welsh S
   13. Young A
   14. Effingham M
   15. McVean G
   16. Leslie S
   17. Allen N
   18. Donnelly P
   19. Marchini J

   (2018) The UK Biobank resource with deep phenotyping and genomic data

   Nature 562:203–209.

   https://doi.org/10.1038/s41586-018-0579-z

   • PubMed
   • Google Scholar
9. 1. Chattopadhyay S
   2. Sud A
   3. Zheng G
   4. Yu H
   5. Sundquist K
   6. Sundquist J
   7. Försti A
   8. Houlston R
   9. Hemminki A
   10. Hemminki K

   (2018) Second primary cancers in non-Hodgkin lymphoma: Bidirectional analyses suggesting role for immune dysfunction

   International Journal of Cancer 143:2449–2457.

   https://doi.org/10.1002/ijc.31801

   • PubMed
   • Google Scholar
10. 1. Copur MS
    2. Wedel W
    3. Jonglertham P
    4. Aprn CS
    5. Horn A

    (2019)

    Hypercalcemia in a patient with small lymphocytic lymphoma/chronic lymphocytic leukemia

    Oncology 33:688884.

    • PubMed
    • Google Scholar
11. 1. Davies NM
    2. Holmes MV
    3. Davey Smith G

    (2018) Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians

    BMJ 362:k601.

    https://doi.org/10.1136/bmj.k601

    • PubMed
    • Google Scholar
12. 1. Davis EJ
    2. Beebe-Dimmer JL
    3. Yee CL
    4. Cooney KA

    (2014) Risk of second primary tumors in men diagnosed with prostate cancer: a population-based cohort study

    Cancer 120:2735–2741.

    https://doi.org/10.1002/cncr.28769

    • PubMed
    • Google Scholar
13. 1. Feller A
    2. Matthes KL
    3. Bordoni A
    4. Bouchardy C
    5. Bulliard J-L
    6. Herrmann C
    7. Konzelmann I
    8. Maspoli M
    9. Mousavi M
    10. Rohrmann S
    11. Staehelin K
    12. Arndt V
    13. NICER Working Group

    (2020) The relative risk of second primary cancers in Switzerland: a population-based retrospective cohort study

    BMC Cancer 20:51.

    https://doi.org/10.1186/s12885-019-6452-0

    • PubMed
    • Google Scholar
14. 1. Hemani G
    2. Zheng J
    3. Elsworth B
    4. Wade KH
    5. Haberland V
    6. Baird D
    7. Laurin C
    8. Burgess S
    9. Bowden J
    10. Langdon R
    11. Tan VY
    12. Yarmolinsky J
    13. Shihab HA
    14. Timpson NJ
    15. Evans DM
    16. Relton C
    17. Martin RM
    18. Davey Smith G
    19. Gaunt TR
    20. Haycock PC

    (2018) The MR-Base platform supports systematic causal inference across the human phenome

    eLife 7:e34408.

    https://doi.org/10.7554/eLife.34408

    • PubMed
    • Google Scholar
15. 1. Hortlund M
    2. Arroyo Mühr LS
    3. Storm H
    4. Engholm G
    5. Dillner J
    6. Bzhalava D

    (2017) Cancer risks after solid organ transplantation and after long-term dialysis

    International Journal of Cancer 140:1091–1101.

    https://doi.org/10.1002/ijc.30531

    • PubMed
    • Google Scholar
16. 1. Kamat MA
    2. Blackshaw JA
    3. Young R
    4. Surendran P
    5. Burgess S
    6. Danesh J
    7. Butterworth AS
    8. Staley JR

    (2019) PhenoScanner V2: an expanded tool for searching human genotype-phenotype associations

    Bioinformatics 35:4851–4853.

    https://doi.org/10.1093/bioinformatics/btz469

    • PubMed
    • Google Scholar
17. 1. Morrison J
    2. Knoblauch N
    3. Marcus JH
    4. Stephens M
    5. He X

    (2020) Mendelian randomization accounting for correlated and uncorrelated pleiotropic effects using genome-wide summary statistics

    Nature Genetics 52:740–747.

    https://doi.org/10.1038/s41588-020-0631-4

    • PubMed
    • Google Scholar
18. Website

    1. National Cancer Institute

    (2020) Number of Persons by Race and Hispanic Ethnicity for SEER Participants - SEER Registries

    Accessed October 15, 2020.

    https://seer.cancer.gov/registries/data.html#a7
19. Website

    1. National Cancer Institute

    (2023) Multiple Primary - Standardized Incidence/Mortality Ratios (MP-SIR/SMR)

    Accessed July 17, 2023.

    https://seer.cancer.gov/seerstat/mp-sir.html
20. Software

    1. R Development Core Team

    (2021) R: A language and environment for statistical computing

    R Foundation for Statistical Computing, Vienna, Austria.

    https://www.R-project.org
21. 1. Shalapour S
    2. Karin M

    (2019) Pas de deux: control of anti-tumor immunity by cancer-associated inflammation

    Immunity 51:15–26.

    https://doi.org/10.1016/j.immuni.2019.06.021

    • PubMed
    • Google Scholar
22. 1. Siegel RL
    2. Miller KD
    3. Fuchs HE
    4. Jemal A

    (2022) Cancer statistics, 2022

    CA 72:7–33.

    https://doi.org/10.3322/caac.21708

    • PubMed
    • Google Scholar
23. 1. Siegel RL
    2. Miller KD
    3. Wagle NS
    4. Jemal A

    (2023) Cancer statistics, 2023

    CA 73:17–48.

    https://doi.org/10.3322/caac.21763

    • PubMed
    • Google Scholar
24. 1. Skrivankova VW
    2. Richmond RC
    3. Woolf BAR
    4. Yarmolinsky J
    5. Davies NM
    6. Swanson SA
    7. VanderWeele TJ
    8. Higgins JPT
    9. Timpson NJ
    10. Dimou N
    11. Langenberg C
    12. Golub RM
    13. Loder EW
    14. Gallo V
    15. Tybjaerg-Hansen A
    16. Davey Smith G
    17. Egger M
    18. Richards JB

    (2021) Strengthening the reporting of observational studies in epidemiology using mendelian randomization: the strobe-mr statement

    JAMA 326:1614–1621.

    https://doi.org/10.1001/jama.2021.18236

    • PubMed
    • Google Scholar
25. 1. Staley JR
    2. Blackshaw J
    3. Kamat MA
    4. Ellis S
    5. Surendran P
    6. Sun BB
    7. Paul DS
    8. Freitag D
    9. Burgess S
    10. Danesh J
    11. Young R
    12. Butterworth AS

    (2016) PhenoScanner: a database of human genotype-phenotype associations

    Bioinformatics 32:3207–3209.

    https://doi.org/10.1093/bioinformatics/btw373

    • PubMed
    • Google Scholar
26. 1. Sung H
    2. Ferlay J
    3. Siegel RL
    4. Laversanne M
    5. Soerjomataram I
    6. Jemal A
    7. Bray F

    (2021) Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries

    CA 71:209–249.

    https://doi.org/10.3322/caac.21660

    • PubMed
    • Google Scholar
27. 1. Zheng G
    2. Sundquist K
    3. Sundquist J
    4. Försti A
    5. Hemminki A
    6. Hemminki K

    (2020) Rate differences between first and second primary cancers may outline immune dysfunction as a key risk factor

    Cancer Medicine 9:8258–8265.

    https://doi.org/10.1002/cam4.3454

    • PubMed
    • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "Risk of Second Primary Cancers After a Diagnosis of First Primary Cancer: A Pan-Cancer Analysis and Mendelian Randomization Study" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Caigang Liu as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) First primary ovarian cancers can be classified into different subtypes to further analyze the relationship between subtypes and causal secondary primary cancers, such that secondary primary cancer prevention would be more specific and targeted.

2) Please consider providing a better rationale for why studying second primary cancers is crucial.

3) Please provide a discussion on the potential clinical implications of the observed associations between first and second primary cancers.

Reviewer #1 (Recommendations for the authors):

1. Four prominent groups of causal relationships mentioned in the research could be further explored. For instance, first primary ovary cancers can be classified into different subtypes to further analyze the relationship between subtypes and causal secondary primary cancers, so that secondary primary cancer prevention would be more specific and targeted.

2. The limited SNPs included in some cancer types mentioned in the research might lead to missed opportunities to discover other significant associations. Enlargement of the data source could have benefits.

3. SEER database used in stage 1 and UK Biobank database used in stage 2 are based on completely different populations, and this may cause confounding factors due to genetic heterogeneity, which should be considered.

Reviewer #2 (Recommendations for the authors):

The paper investigates the association and causal relationship of second primary cancers following a diagnosis of first primary cancer. The combination of a pan-cancer analysis and the use of Mendelian randomization adds novelty and value to the study. However, as the author state, the objective of the study is to investigate the causal relationship between FPCs and SPCs. This can be done based on the UKB data. it is not clear why use SEER data which only provides the association.

Essential revisions:

1) First primary ovarian cancers can be classified into different subtypes to further analyze the relationship between subtypes and causal secondary primary cancers, such that secondary primary cancer prevention would be more specific and targeted.

Thank you for your comments. We performed subgroup analyses in SEER Research Plus Data 18 Reg, Nov 2020 Sub (2000-2018), with additional 2-year follow-up data. We stratified the total patients with first primary ovarian cancers to four groups based on the pathological subtypes: serous (~69.5%), endometrioid, mucinous, and clear cell carcinomas. The high-grade serous carcinoma (HGSC), as the most common type of ovarian cancer, was also evaluated. The SIR analyses were repeated in the above subgroups. The additional result was listed in Supplementary file 1l. However, only some subtypes (e.g. serous, high-grade serous and endometrioid ovary cancer) conformed to the results in the relation between overall primary ovary cancer and several secondary primary cancers (lung and bronchus and non-Hodgkin lymphoma).

2) Please consider providing a better rationale for why studying second primary cancers is crucial.

We’ve added it on lines 90-97. The study may help us understand critical questions in clinical practice about who should be more carefully monitored for the second primaries. In addition, precision screenings against certain cancers should also be considered among those patients with increased risk of second primaries in addition to the regular follow-up evaluations. As known, many cancers are known to have a multifactorial etiology and some cancer treatments are known to be carcinogenic. It is essential to illustrate the potential relationship between FPCs and SPCs which may optimize the treatment for patients’ prognosis and survivorship.

3) Please provide a discussion on the potential clinical implications of the observed associations between first and second primary cancers.

In terms of clinical implications, lifestyle, environment, treatment, host factors and other influences may contribute to multiple primary cancer. It required clinicians and researchers to explore a specific link between FPCs and SPCs. It would be of clinical importance to make a personalized screening plan for certain primary cancer patients (eg. oral pharyngeal cancer, ovary cancer and kidney cancer), especially for those young onsets. For example, patients with kidney cancer would have an increased risk of lung cancer and myeloma. Thus, a screening for lung cancer and myeloma should be recommended among those patients. Please refer to line 223-231.

Reviewer #1 (Recommendations for the authors):

1. Four prominent groups of causal relationships mentioned in the research could be further explored. For instance, first primary ovary cancers can be classified into different subtypes to further analyze the relationship between subtypes and causal secondary primary cancers, so that secondary primary cancer prevention would be more specific and targeted.

Thank you for your comments. We performed subgroup analyses in SEER Research Plus Data 18 Reg, Nov 2020 Sub (2000-2018), with additional 2-year follow-up data. We stratified the total patients with first primary ovarian cancers to four groups based on the pathological subtypes: serous (~69.5%), endometrioid, mucinous, and clear cell carcinomas. The high-grade serous carcinoma (HGSC), as the most common type of ovarian cancer, was also evaluated. The SIR analyses were repeated in the above subgroups. The additional result was listed in supplementary file 1l. However, the number of patients with specific pathological type was too few to obtain a stable association result.

2. The limited SNPs included in some cancer types mentioned in the research might lead to missed opportunities to discover other significant associations. Enlargement of the data source could have benefits.

We totally agreed with you. Data sharing is critical to find significant associations. In particular, a complete genome-wide association study (GWAS) summary statistics that underpin the selection and weighting of genetic variants for a particular trait. Comprehensive databases of GWAS summary statistics, such as the pioneering NHGRI-EBI GWAS Catalog, are widely utilized by the community but still only a minority of published GWAS share their full summary statistics. The current study, however, provides the most up-to-date evidence in this topic. We also believe that it should be regularly updated with additional GWAS results in the future.

3. SEER database used in stage 1 and UK Biobank database used in stage 2 are based on completely different populations, and this may cause confounding factors due to genetic heterogeneity, which should be considered.

We mainly focus on the white people in SEER database (non-Hispanic whites) and UK Biobank (British, Irish and any other white background). As a matter of fact, since this study was mainly focus on the European Ancestry (aka, white or non-Hispanic or non-Nordic Caucasian), the genetic heterogeneity between these two populations was very low. [citation: https://handwiki.org/wiki/Social:Non-Hispanic_or_Latino_whites]

Reviewer #2 (Recommendations for the authors):

The paper investigates the association and causal relationship of second primary cancers following a diagnosis of first primary cancer. The combination of a pan-cancer analysis and the use of Mendelian randomization adds novelty and value to the study. However, as the author state, the objective of the study is to investigate the causal relationship between FPCs and SPCs. This can be done based on the UKB data. it is not clear why use SEER data which only provides the association.

Thank you for being so concerned. The SEER database is the most significant incidence dataset in the United States based on population cancer registration. It is very critical to find significant associations between FPCs and SPCs based on such a great number of cancer patients. While the UKB data is composed of ~500,000 people aged 40-69 yrs, the incidence of FPCs and SPCs is limited, which might considerably impede the statistical power. On the other hand, in order to interpret the MR results under a more stringent standard, we would like to perform the analyses in independent study populations given the low genetic heterogeneity between these two populations. We think it is rigorous to combine these two stage results to draw a relatively solid conclusion.

Author details

1. Xiaohao Ruan

   Department of Urology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China

   Contribution

   Software, Formal analysis, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Contributed equally with

   Da Huang

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-8373-5237

2. Da Huang

   Department of Urology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China

   Contribution

   Resources, Software, Formal analysis, Funding acquisition, Visualization

   Contributed equally with

   Xiaohao Ruan

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6203-9459

3. Yongle Zhan

   Division of Urology, Department of Surgery, Queen Mary Hospital, The University of Hong Kong, Hong Kong, China

   Contribution

   Software, Methodology

   Competing interests

   No competing interests declared

4. Jingyi Huang

   Department of Urology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China

   Contribution

   Data curation, Investigation

   Competing interests

   No competing interests declared

5. Jinlun Huang

   Department of Urology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China

   Contribution

   Data curation, Investigation

   Competing interests

   No competing interests declared

6. Ada Tsui-Lin Ng

   Division of Urology, Department of Surgery, Queen Mary Hospital, The University of Hong Kong, Hong Kong, China

   Contribution

   Project administration

   Competing interests

   No competing interests declared

7. James Hok-Leung Tsu

   Division of Urology, Department of Surgery, Queen Mary Hospital, The University of Hong Kong, Hong Kong, China

   Contribution

   Project administration

   Competing interests

   No competing interests declared

8. Rong Na

   Division of Urology, Department of Surgery, Queen Mary Hospital, The University of Hong Kong, Hong Kong, China

   Contribution

   Conceptualization, Resources, Data curation, Supervision

   For correspondence

   narong.hs@gmail.com

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-7470-5108

• 320

  Page views

• 81

  Downloads

• 0

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.