Increased burden of familial-associated early-onset cancer risk among minority Americans compared to non-Latino Whites

Reviewer #2 (Recommendations for the authors):

My major concern is with the methods used to compare the ethnic subgroups. This needs to be revisited.

It is distracting that the legend on the Forest plot is on the log-scale, but the numbers reported are on the original scale. Please change the numbers on the X axis to be on the original scale.

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.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there is a remaining issue that needs to be addressed, as outlined below:

Please revise the following statements in the abstract: "Higher relative risk of any cancer in siblings and mothers given a proband with solid cancer (P<0.05) was observed for both Latinos (SIR=4.98;95%CI:3.82-6.39) and for non-Latino Blacks (SIR=7.35;95%CI:3.36-13.95) compared to non-Latino White subjects (SIR=3.02;95%CI:2.12-4.16)." This statement needs to be rewritten as the single p value does not make any sense where it is placed.

Thank you for the issue brought up about the statement in the abstract. We agree that the placement of the p-value is unclear as to what it refers to. We therefore would like to change the sentence as follows:

"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 given a proband with solid cancer when compared to non-Latino White subjects (SIR=3.02;95%CI:2.12-4.16)."

In this sentence, we state that both Latinos and non-Latino Blacks have “significantly” higher risk, which assures the reader that the p value reaches a p=0.05 threshold. This significance refers to the fact that the Latino SIR (4.98) and non-Latino black SIR (7.35) were statistically significantly higher than white subject SIR (3.02). We feel that adding the actual p values to the sentence (within the quotes) will not make the point clearer, as readers may believe that P values would refer to the SIR being statistically different than the null (which they are, but that is not the point of the sentence).