査読者への回答　その１

RESPONSE TO REVIEWER 1:

We wish to express our appreciation to the Reviewer for his or her insightful comments, which have helped us significantly improve the paper.

Comment 1: The manuscript may benefit from some additional analyses. First, please give the stage and grade distributions of cases. I would like to see the analyses stratified by grade (2-7, 8-10) and stage separately, using categories for missing as needed. The definition of advanced disease used in the analyses is a bit odd, as it appears that local stage cases with metastatic score 8-10 were classified as localized, but if stage were missing they were classified as advanced. I would accept a definition of metastatic score 8-10 or extraprostatic disease as advanced, but this would be in addition to seeing the data for grade and stage separately.

Response: We thank the Reviewer for this pertinent comment.

In this study, information on metastatic score or degree of differentiation was used only when cases could not be defined by information on local staging.

In accordance with the Reviewer's comment, we first divided cases using local stage only into the classifications of advanced cases (extraprostatic or metastatic cancer involving lymph nodes or other organs as regionally invasive or metastatic cases) and localized cases (cancer confined within prostate). There were 227 localized cases and 125 advanced, while 82 could not be defined as either group and were classified as undetermined cases.

Relative risk of prostate cancer according to consumption by local stage was as follows:

TABLE. Relative risk of prostate cancer according to consumption by local stage.

Moreover, we also divided cases for which local staging information was not available (75 undetermined cases) into localized and advanced cancer using information on metastatic score or degree of differentiation. We added 13 cases with a high metastatic score (8 to 10) or poor differentiation to the advanced cancer group. These criteria were selected to allow the identification of cases with a high likelihood of a poor prognosis. Further, we added 42 cases with a low metastatic score (≦7) or well or moderate differentiation to the localized cancer group. Finally, we confirmed 256 localized cases, 101 advanced cases and 27 (5% of total) undetermined cases.

Relative risk of prostate cancer according to consumption by local stage, metastatic score or histological differentiation was as follows:

TABLE . Relative risk of prostate cancer according to consumption by local stage, metastatic score or histological differentiation.

We compared these two tables, but they were not substantially different. We therefore divided the cases by using the information on local stage, metastatic score or histological differentiation (reference 25). Further, we did not change Table 2. However, as the reviewer noted, the possibility of misclassification exists, in that local stage cases with metastatic score 8-10 were classified as localized, but as advanced if stage was missing.

We have therefore added the following text as one of the limitations of the study (p. 18, lines 4-9):

“Finally, there is possibility of misclassification that local stage cases with high metastatic score (8 to 10) were classified as localized, but if stage information were missing they were classified as advanced. Unfortunately, we could not classify cases by metastatic score only, because we collected metastatic score as supplementary information, and the proportion of cases with information of metastatic score were low (23% of total cases).”

In accordance with Reviewer 1’s comment, we have changed the expression of definition of localized and advanced prostate cancer.

Moreover, in accordance with comment 4 of Reviewer 3, we have made a new section “Definition of localized and advanced prostate cancer” in Materials and Methods, and added the following text (p. 9, line 10- p. 9, line 18):

“Definition of localized and advanced prostate cancer For cancer registry in our study, local staging is required item, but metastatic score is supplementary information. Therefore, cases were classified as advanced cases (extraprostatic or metastatic cancer involving lymph nodes or other organs) and localized cases (cancer confined within prostate). There were 227 localized cases and 125 advanced. Of these cases, 82 cases (20% of total) could not be defined as either group (undetermined cases), because some prostate cancers which a prostatectomy was not done were not determined local stage by using local imaging only.. The stage distribution in our study was similar to those in Japan overall (24). Moreover, if the information of local staging was not available (75 undetermined cases), we added 11 cases with a high metastatic score (8 to 10) or poor differentiation to advanced cancer. These criteria were selected to allow the identification of cases with a high likelihood of poor prognosis. On the other hand, we added 42 cases with a low metastatic score (≦7) or well or moderate differentiation to localized cancer. Finally, we confirmed 256 localized cases, 101 advanced cases and 27 (5% of total) undetermined cases (25).”

Further, we have changed the following text from (p. 2, lines 12-14):

“During this time, 384 men were newly diagnosed with prostate cancer, of whom 101 had advanced cases, 256 were organ-localized and 27 were of an undetermined stage.”

to

“During this time, 384 men were newly diagnosed with prostate cancer, of whom 101 had advanced cases, 256 were localized and 27 were of an undetermined stage.”

We have also added the following reference.

“25. Kurahashi N, Iwasaki M, Sasazuki S, Otani T, Inoue M, Tsugane S. Soy product and isoflavone consumption in relation to prostate cancer in Japanese men. Cancer Epidemiol Biomarkers Prev 2007;16:538-45.”

Comment 2: Data in Table 1 should be adjusted for age. What are the p-values testing? Perhaps indicating which means differ across consumption categories would be more informative.

Response: In accordance with the Reviewer's comment, Table 1 has been adjusted for age. P difference values for characteristics between categories of consumption were calculated by analysis of variance and the chi-square test for homogeneity.

Thus, the following sentence has been inserted in the footnote in Table 1.

“* All variables except for age were standardized to the age distribution (categorized by 5-year intervals) of the entire cohort.” “‡ P difference values of characteristics between categories of consumption were calculated by analysis of variance and the chi-square test for homogeneity.”

The questionnaires used in Cohort I and Cohort II differed slightly with respect to food items, method of expression and frequency categories. Therefore, when we used covariates of fruits, green or yellow vegetables, dairy food, soy food and genistein, we calculated separate estimates for Cohort I and Cohort II, and then analyzed the combined result using a fixed-effects model. Thus, we could not calculate the mean of food item, and we showed that the percentages differ across consumption categories in the whole cohort.

Further we have changed the following text (p. 13, lines 4-7) from:

“Subject characteristics at baseline according to category of consumption are shown in Table 1. Persons with high consumption were older. The proportion of current smokers was high in the highest category of consumption but that among regular drinkers was low. Fewer men lived with their wife in the lowest category than in the other categories. The proportion of daily coffee or black tea drinkers decreased as consumption increased. Further, intake of miso soup, fruits, green or yellow vegetables, dairy food and soy food increased with intake.”

to

“Subject characteristics at baseline according to category of consumption are shown in Table 1. Participants with more consumption tend to be older, to smoke more, to have higher proportion of men living with their wife, to consume more miso soup, fruits, vegetables, and soy food, and to consume less coffee.”

Comment 3: The statement that risk of total and localized cancer did not differ by consumption is not altogether logical. The finding was no association for local disease and a reduced risk for advanced. The only reason that there was no association for total cancer was that the effect for advanced was diluted. Perhaps authors could find a more clear way to give these results.

Response: In accordance with the Reviewer's comment, we have changed the following text in the Introduction from (p. 1, line 18):

"was not associated with total or localized prostate cancer.”

to

“ was not associated with localized prostate cancer.”

Similarly, we have changed the following text in the Results (p.13, line 3-4)

“In contrast, no association was observed between consumption and total or localized prostate cancer.”

to:

“In contrast, no association was observed between consumption and localized prostate cancer.”

Comment 4: Please have the English reviewed by a medical editor. Figures: Please reconsider the need for the figures, some of which don’t add much to the manuscript. Also, can Tables 2 and 5 be combined into one table?

Response: We have had the manuscript rewritten by an experienced scientific editor, who has improved the grammar and stylistic expression of the paper

In accordance with the Reviewer’s comment, we have deleted Figures 3, 4 and 6 from the revised manuscript and have combined Tables 2 and 5 into a new Table 2

We wish to thank the Reviewer again for his or her valuable comments.