Increased burden of familial-associated early-onset cancer risk among minority Americans compared to non-Latino Whites
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
This is a large population based study examining the familial risks of cancer in a California population limiting the analysis to cancers occurring under the age of 30. The work has found that risk of cancer is increased if a person has a first degree relative with cancer. Increased familial risk of cancer to first (and second) degree relatives is long established, but this is a large source of data and it is nonetheless valuable to have this published. They have been able to look specifically at Latino risks as this is a common ethnic group in California.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
The role of race/ethnicity in genetic predisposition of early-onset cancers can be estimated by comparing family-based cancer concordance rates among ethnic groups.
Methods:
We used linked California health registries to evaluate the relative cancer risks for first-degree relatives of patients diagnosed between ages 0 and 26, and the relative risks of developing distinct second primary malignancies (SPMs). From 1989 to 2015, we identified 29,631 cancer patients and 62,863 healthy family members. We calculated the standardized incident ratios (SIRs) of early-onset primary cancers diagnosed in proband siblings and mothers, as well as SPMs detected among early-onset patients. Analyses were stratified by self-identified race/ethnicity.
Results:
Given probands with cancer, there were increased relative risks of any cancer for siblings and mothers (SIR = 3.32; 95% confidence interval [CI]: 2.85–3.85) and of SPMs (SIR = 7.27; 95% CI: 6.56–8.03). Given a proband with solid cancer, both Latinos (SIR = 4.98; 95% CI: 3.82–6.39) and non-Latino Blacks (SIR = 7.35; 95% CI: 3.36–13.95) exhibited significantly higher relative risk of any cancer in siblings and mothers when compared to non-Latino White subjects (SIR = 3.02; 95% CI: 2.12–4.16). For hematologic cancers, higher familial risk was evident for Asian/Pacific Islanders (SIR = 7.56; 95% CI: 3.26–14.90) compared to non-Latino whites (SIR = 2.69; 95% CI: 1.62–4.20).
Conclusions:
The data support a need for increased attention to the genetics of early-onset cancer predisposition and environmental factors in race/ethnic minority families in the United States.
Funding:
This work was supported by the V Foundation for funding this work (Grant FP067172).
Article activity feed
-
-
Author Response:
We thank you for the careful review and the opportunity to resubmit this manuscript. We particularly acknowledge the reviewer who helped to clarify the statistical arguments and stimulated our re-analysis of all results. This re-analysis has helped to change the focus of the work to identify significantly variable (higher) familial cancer risks in several race/ethnically described minority groups in the US, which we feel has broadened the message stimulating a word change in the title.
Reviewer #1 (Public Review):
This is a very well written and comprehensive paper that is a valuable contribution to the literature of childhood cancers. It shows that some childhood cancers have an inherited component and the risk could be to the mother or to the siblings. Although the relative risks are significant, childhood cancer is …
Author Response:
We thank you for the careful review and the opportunity to resubmit this manuscript. We particularly acknowledge the reviewer who helped to clarify the statistical arguments and stimulated our re-analysis of all results. This re-analysis has helped to change the focus of the work to identify significantly variable (higher) familial cancer risks in several race/ethnically described minority groups in the US, which we feel has broadened the message stimulating a word change in the title.
Reviewer #1 (Public Review):
This is a very well written and comprehensive paper that is a valuable contribution to the literature of childhood cancers. It shows that some childhood cancers have an inherited component and the risk could be to the mother or to the siblings. Although the relative risks are significant, childhood cancer is fortunately rare and the actual risk to the siblings is small.
Can we assume this is less than one percent? i think it would be helpful to provide some absolute risk numbers for the siblings so that parents could be reassured that the risk to other children is small.
Response: We appreciate this comment on absolute risk. It is true that the actual risk is very small given the rarity of childhood cancers. We calculated the overall absolute risk for mothers and siblings of a proband and compared it with the general population. It now reads “Moreover, due to the rarity of childhood cancers, the absolute risk is very small, but still higher among young siblings and mothers in the current study (0.074%) compared to general population (0.023%) of the same age group” in line 316 of the Discussion section.
Do the authors have a suggestion on what genetic tests should be done on children with cancer? Do you have recommendations to make? i assume that the authors do not recommend screening of siblings for cancer except in rare cases. It would be useful to see what the authors recommend.
Response: In this manuscript we do not provide clinical recommendations as we feel that is out of the scope of this research. Instead, we are making several points:
That conventional US-based birth and cancer registries can be utilized to study familial-based cancer risks.
That different ethnic groups appear to have different familial risks for some cancer subtypes.
Early onset parental cancers can add information about familial-based risks.
Second primary malignancies are enriched in families that exhibit familial risks (line 260 of the Results section). These characteristics will provide useful information for genetic counselors who need to advise families on their own decisions about genetic testing and family planning. At the present time the genetic counseling clinical discipline is tasked to make specific recommendations to families about screening siblings for cancer and presence of cancer predisposition alleles, such advice is stimulated by examining family history of cancer. Our work suggests that Latino families may have a higher risk of familial alleles in solid tumors overall, which may promote more attention or scrutiny of families by ethnicity.
Are there some sites where the risk to siblings is there but not to parents which might suggest recessive inheritance?
Response: this is an interesting question, but there are two reasons why our study may not be adequate to assess this. First, our sample size may not be large enough to adequately study this point. The risk to cancer in the general population is higher in children than it is in young adults – and therefore the low numbers of cancer in mothers that we see is largely a reflection of the low risk of cancer in young adults, since we cut off our observational age at 26 (due to the extent of follow-up on our young population). There is a lack of cancer at many of the ICC-03 defined childhood cancer sites among our parents, making it impossible to estimate cancer risk in the adults. Second, childhood cancers are biologically distinct from adults, so the risk imparted for childhood cancer from predisposition alleles that affect those cancers may not always have any effect on young adult cancers. Additionally, the progenitor cells at risk from childhood cancer may have differentiated, leading to no cells “at risk” of transformation after adolescence and the effect of childhood cancer predisposition alleles on those adult cancers not a meaningful comparison. Of course, there are exceptions to this such as TP53 alleles which affect cancer risk of many subtypes at any age.
If the childhood cancer is rare and fatal one might not see it in the parents because of loss or reproductive fitness. Please comment.
Response: We appreciate this comment a lot and have the same concern that patients with cancer that have a strong genetic cancer predisposition may not be capable to reproduce (even if the patient survives). We added a comment in the discussion section, and it now reads “Furthermore, it is likely that the low number of mothers with cancer is a result of bias against some very strong cancer predisposition alleles, so the patients could not survive long enough or be healthy enough to reproduce” on line 408.
Should we assume that the higher risks for Latino children are purely due to genetic influences? Could there be environmental factors at play as well?
Response: We appreciate this comment and totally agree that environmental factors also play a role. Not only genetic factors, but also the environmental factors, and the interaction between genetic and environmental factors would contribute to the variation in relative risks. We have addressed this point in lines 341 (“This familial concordance is likely due to both shared genetic and environmental…”) and 419 (“Second, the comparative attributable fraction of familial risk based on environmental risk factors interacting…”) of the discussion section. We believe that this point should stimulate further research, and we are constructing our own future studies to explore environmental factors along with genetics.
Reviewer #2 (Public Review):
[...] Although the authors comment that the results from the Chi-Sq test are not consistent with the specific group SIRs and 95%CIs, they do not explain how these results can be so different.
I am concerned that there is either an error in the calculations or an error in the assumptions. It is not acceptable to have such contradictory results between the two distinct methods.
For example, for hematological cancers the 95% CI for Latinos is entirely contained within the 95%CI for Non-Latino white, while this gives a p less than 0.05. The authors need to explore why these methods are giving very different answers and be clear that the low p-values are not simply an artifact of poor assumptions.
Response: We sincerely appreciate the comments from Reviewer 2. And we want to thank Reviewer 2 for pushing on the inconsistency between confidence intervals and p-value comparing the SIRs between race/ethic groups. While overlapping CI’s do not necessarily indicate a lack of significance in the effect sizes, the apparent contrast in these statistical measures was too extreme to be believable and indeed there was an error.
We reconstructed our data from scratch and recalculated all statistical comparisons with our statistician, Dr. W. J. Gauderman, and found a recurrent mistake in the calculation of p-value comparing the SIRs between race/ethic groups. We have corrected this mistake throughout the manuscript. Please refer to the new Figure 1, 3, and supplementary materials for the corrected numbers. The p values are now somewhat attenuated, and significant differences between Latinos and NL whites persist for solid tumors. In addition, Asians have significantly increased familial risk for hematologic cancers, and non-Latino Blacks have significantly increased risk of solid tumors when compared to non-Latino whites. Because of this broader enhanced risk evident in minority groups (with the corrected statistical comparisons), the focus of the manuscript was changed slightly emphasizing higher risks among minority groups in respective hematologic and solid tumor categories. There were also SIR differences suggested between many individual types of cancer, while not reaching formal statistical significance.
-
Reviewer #2 (Public Review):
California health registries were linked to evaluate the relative cancer risks for first degree relatives of patients diagnosed between ages 0-26, Over the period 1989-2015, 29,631 cancer patients were identified and 62,863 healthy family members. The strengths are that this is a large population based study and that the relative cancer risk for specific ethnic groups could be investigated. The analyses were limited to mothers and siblings of children or adults with cancer. Increased relative risk of cancer is well established and these data add to this evidence base.
The major finding that the authors emphasise is increased relative risk of cancer for Latinos (compared to non Latinos). I am not convinced that they have much evidence for this as the Forest plots in the manuscript do not slow large …
Reviewer #2 (Public Review):
California health registries were linked to evaluate the relative cancer risks for first degree relatives of patients diagnosed between ages 0-26, Over the period 1989-2015, 29,631 cancer patients were identified and 62,863 healthy family members. The strengths are that this is a large population based study and that the relative cancer risk for specific ethnic groups could be investigated. The analyses were limited to mothers and siblings of children or adults with cancer. Increased relative risk of cancer is well established and these data add to this evidence base.
The major finding that the authors emphasise is increased relative risk of cancer for Latinos (compared to non Latinos). I am not convinced that they have much evidence for this as the Forest plots in the manuscript do not slow large differences between the two ethnic classifications. Their evidence for these differences comes from a separate comparison of the SIRs using an approximate Chi-squared test.
Although the authors comment that the results from the Chi-Sq test are not consistent with the specific group SIRs and 95%CIs, they do not explain how these results can be so different.
I am concerned that there is either an error in the calculations or an error in the assumptions. It is not acceptable to have such contradictory results between the two distinct methods.
For example, for hematological cancers the 95% CI for Latinos is entirely contained within the 95%CI for Non-Latino white, while this gives a p less than 0.05. The authors need to explore why these methods are giving very different answers and be clear that the low p-values are not simply an artifact of poor assumptions.
-
Reviewer #1 (Public Review):
This is a very well written and comprehensive paper that is a valuable contribution to the literature of childhood cancers. It shows that some childhood cancers have an inherited component and the risk could be to the mother or to the siblings. Although the relative risks are significant, childhood cancer is fortunately rare and the actual risk to the siblings is small.
Can we assume this is less than one percent? i think it would be helpful to provide some absolute risk numbers for the siblings so that parents could be reassured that the risk to other children is small.
Do the authors have a suggestion on what genetic tests should be done on children with cancer? Do you have recommendations to make? i assume that the authors do not recommend screening of siblings for cancer except in rare cases. It would be …
Reviewer #1 (Public Review):
This is a very well written and comprehensive paper that is a valuable contribution to the literature of childhood cancers. It shows that some childhood cancers have an inherited component and the risk could be to the mother or to the siblings. Although the relative risks are significant, childhood cancer is fortunately rare and the actual risk to the siblings is small.
Can we assume this is less than one percent? i think it would be helpful to provide some absolute risk numbers for the siblings so that parents could be reassured that the risk to other children is small.
Do the authors have a suggestion on what genetic tests should be done on children with cancer? Do you have recommendations to make? i assume that the authors do not recommend screening of siblings for cancer except in rare cases. It would be useful to see what the authors recommend.
Are there some sites where the risk to siblings is there but not to parents which might suggest recessive inheritance?
If the childhood cancer is rare and fatal one might not see it in the parents because of loss or reproductive fitness. Please comment.
Should we assume that the higher risks for Latino children are purely due to genetic influences? Could there be environmental factors at play as well?
-
Evaluation Summary:
This is a large population based study examining the familial risks of cancer in a California population limiting the analysis to cancers occurring under the age of 30. The work has found that risk of cancer is increased if a person has a first degree relative with cancer. Increased familial risk of cancer to first (and second) degree relatives is long established, but this is a large source of data and it is nonetheless valuable to have this published. They have been able to look specifically at Latino risks as this is a common ethnic group in California.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)
-
-