Clinicopathological Characteristics and Prognostic Factors in Ovarian Metastases from Right- and Left-Sided Colorectal Cancer

Round 1

Reviewer 1 Report

To be accepted

Author Response

Not applicable

Reviewer 2 Report

The manuscript has been much improved and vague points in the original manuscript are now well explained. Still, some minor points seem to be revised.

1. The authors seem to overinterpret the cited literature greatly, which deteriorate the discussion.
   • “Other factors known to be associated with peritoneal spread, including mucinous histology, the presence of BRAF activating mutations, and poor differentiation are more frequently found in right-sided tumors [33].” (page 10, line19)
     This statement is tricky. It means that (1) mucinous histology, the presence of BRAF activating mutations, and poor differentiation are associated peritoneal spread and (2) These factors are more frequently found in right-sided tumors. Although (2) is described in the reference [33], this article says nothing about (1). The authors should cite the articles that show (1) or revise the expression.
   • “In this regard, the downregulation of E-cadherin was shown to promote ovarian-specific metastases in patients with CRC [42].” (page 10, line 29)
     This study [42] is largely based on animal model-based experiments and does not show the evidence described above. Moreover, the topic of this paragraph is the difference in colon cancer sidedness, not the general tendency of ovarian metastasis.
   • “Therefore, it could be hypothesized that the CMS1 and CMS4 subtypes might be more prone to develop ovarian metastases.” (page10, line 35)
     Again, the topic of this paragraph is the difference in colon cancer sidedness, not the tendency of ovarian metastasis. According to this article [33], the proportion of CMS4 does not differ between right-sided and left-sided colon cancers.
2. The definitions of extraovarian spread, peritoneal carcinomatosis and distant metastases are now clearly described. However, only two of these are listed and one is lacking on Table 1.
3. Now parameters are separated by lines on the tables. A line seems lacking (above “Chronology”) on Table 1.
4. The last part of the title of Table 2 is missing.
5. The citation of the table seems wrong (page 6, the last paragraph). Probably it is not Table 1 but Table 3.

 

Author Response

Thank you for your revision. We appreciate the constructive criticism

1. The authors seem to overinterpret the cited literature greatly, which deteriorates the discussion.

“Other factors known to be associated with peritoneal spread, including mucinous histology, the presence of BRAF activating mutations, and poor differentiation are more frequently found in right-sided tumors [33].” (page 10, line19)

This statement is tricky. It means that (1) mucinous histology, the presence of BRAF activating mutations, and poor differentiation are associated with peritoneal spread and (2) These factors are more frequently found in right-sided tumors.

Although (2) is described in the reference [33], this article says nothing about (1). The authors should cite the articles that show (1) or revise the expression.

Citations regarding the association between mucinous histology, presence of BRAF activating mutation, and poor differentiation and peritoneal spread were added to the manuscript:

Mucinous histology: Hugen, N.; van de Velde, C.J.H.; de Wilt, J.H.W.; Nagtegaal, I.D. Metastatic pattern in colorectal cancer is strongly influenced by histological subtype. Annals of oncology : official journal of the European Society for Medical Oncology / ESMO 2014, 25, 651-657, doi:10.1093/annonc/mdt591.

Presence of BRAF activating mutation: Franko, J.; Shi, Q.; Meyers, J.P.; Maughan, T.S.; Adams, R.A.; Seymour, M.T.; Saltz, L.; Punt, C.J.A.; Koopman, M.; Tournigand, C., et al. Prognosis of patients with peritoneal metastatic colorectal cancer given systemic therapy: an analysis of individual patient data from prospective randomised trials from the Analysis and Research in Cancers of the Digestive System (ARCAD) database. The Lancet. Oncology 2016, 17, 1709-1719, doi:10.1016/s1470-2045(16)30500-9.

Poor differentiation: Shelygin, Y.A.; Pospekhova, N.I.; Shubin, V.P.; Kashnikov, V.N.; Frolov, S.A.; Sushkov, O.I.; Achkasov, S.I.; Tsukanov, A.S. Epithelialmesenchymal transition and somatic alteration in colorectal cancer with and without peritoneal carcinomatosis. BioMed research international 2014, 2014, 629496, doi:10.1155/2014/629496.

“In this regard, the downregulation of E-cadherin was shown to promote ovarianspecific metastases in patients with CRC [42].” (page 10, line 29) This study [42] is largely based on animal model-based experiments and does not show the evidence described above. Moreover, the topic of this paragraph is the difference in colon cancer sidedness, not the general tendency of ovarian metastasis.

“Therefore, it could be hypothesized that the CMS1 and CMS4 subtypes might be more prone to develop ovarian metastases.” (page10, line 35) Again, the topic of this paragraph is the difference in colon cancer sidedness, not the tendency of ovarian metastasis. According to this article [33], the proportion of CMS4 does not differ between right-sided and left-sided colon cancers.

We agree that CMS subtypes and metastatic organotropism of CRC represent a different topic from the paragraph dedicated to the primary tumor sidedness. We are aware that the data to support the possible association between CRC ovarian metastases and specific CMS subtypes is rather limited. Therefore, the

hypothesis regarding the association between STO of colorectal origin and specific CMS subtypes was deleted from the manuscript.

2. The definitions of extraovarian spread, peritoneal carcinomatosis and distant metastases are now clearly described. However, only two of these are listed and one is lacking on Table 1.

The missing information regarding distant metastases was added to Tab. 2. The definition of “Extraovarian metastases” and “Distant metastases” was added to the table captions.

3. Now parameters are separated by lines on the tables. A line seems lacking (above “Chronology”) on Table 1.

The missing line was added.

4. The last part of the title of Table 2 is missing

The title was edited.

5. The citation of the table seems wrong (page 6, the last paragraph). Probably it is not Table 1 but Table 3.

The citation was corrected.

Reviewer 3 Report

Ready to be published.

Author Response

Not applicable

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Dear Authors,

I have evaluated your "Clinicopathological characteristics and prognostic factors in ovarian metastases from right- and left-sided colorectal cancer" entitled article, and here are my comments:

1. Abstract and introduction sections; page 1, line 17 and 44: “Secondary tumors of the ovary (STOs) account for 10–25% of all ovarian malignancies”. Please explain if the metastases from gynecological origins were included in this proportion (10-25%) of these secondary tumors or not.
2. Introduction; page 2 line 57: using the term secondary aim instead of another aim is more appropriate, please consider revising.
3. Results; page 5 line 167: “are listed in….” the remaining of the sentence is absent. May you forgot to write Table 2.
4. Results section: the laterality of the tumor is presented in Table 1 and Table 2. I suggest keeping it in only one of them, preferably Table 2.
5. Conclusion; page 10 line 346-8: “Further, a resection of both ovaries should be performed even in a case of unilateral mass when diagnosed through imaging methods because occult micrometastases in the contralateral ovary are frequently overlooked by these methods”. Such a conclusion couldn’t be derived from this study. Therefore, revision of this interpretation should be considered. Bilateral adnexectomy can be considered in postmenopausal women even tumor was detected only in one of the ovaries due to the possible micrometastases.

 

Author Response

1. Abstract and introduction sections; page 1, line 17 and 44: “Secondary tumors
of the ovary (STOs) account for 10–25% of all ovarian malignancies”. Please
explain if the metastases from gynecological origins were included in this
proportion (10-25%) of these secondary tumors or not.
Metastases from primary gynecological tumors were included in this
proportion, which is now stated in both the abstract and the introduction
section.
2. Introduction; page 2 line 57: using the term secondary aim instead of another
aim is more appropriate, please consider revising.
We do agree that the term “secondary” is more appropriate than “another”,
which was edited in the manuscript.
3. Results; page 5 line 167: “are listed in….” the remaining of the sentence is
absent. May you forgot to write Table 2.
The missing expression “Table 2” was added to the manuscript.
4. Results section: the laterality of the tumor is presented in Table 1 and Table 2. I
suggest keeping it in only one of them, preferably Table 2.
The duplicate information was removed from table 1.
5. Conclusion; page 10 line 346-8: “Further, a resection of both ovaries should be
performed even in a case of unilateral mass when diagnosed through imaging
methods because occult micrometastases in the contralateral ovary are
frequently overlooked by these methods”. Such a conclusion couldn’t be derived
from this study. Therefore, revision of this interpretation should be considered.
Bilateral adnexectomy can be considered in postmenopausal women even tumor
was detected only in one of the ovaries due to the possible micrometastases.
We revised this topic in the discussion section. Considering that bilateral
adnexectomy represents no additional morbidity over unilateral adnexectomy
and resulting hormone depletion is not an issue in postmenopausal patients,
we believe that bilateral adnexectomy should be performed in all
postmenopausal women. In premenopausal patients, the possibility of
unilateral adnexectomy should be discussed with the patient on individual
basis, especially if the fertility needs to be preserved. The patient, however,
needs to be fully informed on the risk of occult micrometastases in the
contralateral ovary and the possibility of developing metachronous metastases
in the preserved ovary. The development of metachronous metastases in the
preserved ovary was observed in one of our patients and had already been
described in the literature:
Erroi, F.; Scarpa, M.; Angriman, I.; Cecchetto, A.; Pasetto, L.; Mollica, E.; Bettiol,
M.; Ruffolo, C.; Polese, L.; Cillo, U., et al. Ovarian metastasis from colorectal
cancer: prognostic value of radical oophorectomy. Journal of surgical oncology
2007, 96, 113-117, doi:10.1002/jso.20803.

Reviewer 2 Report

The authors have performed a chart review-based retrospective analysis of CRC patients who presented with ovarian metastasis to evaluate if tumor sidedness influenced the natural history. This analysis concluded that ovarian metastases originating from the right-sided primary tumors were more frequently bilateral, associated with peritoneal carcinomatosis, had ovarian surface involvement by the tumor, and contained a mucinous component. Furthermore, the authors have provided data illustrating that the imaging methods are inaccurate to determine the laterality of ovarian metastases correctly compared to histopathological examination. The authors have also postulated that transperitoneal spread is the predominant metastatic pathway for the right-sided tumors, and hematogenous/lymphatic spread is the dominant pathway for the left-sided tumors. The data presented in this manuscript are clinically relevant and deserve attention. However, the manuscript has several significant flaws, as listed below:

1. The meaning of a sentence is often unclear. For example:

Line 22-24- ‘This retrospective, observational chart review study was conducted between January 2000 and December 2019 and included patients with histologically confirmed STOs of CRC origin.’

I think the authors wanted to say- ‘This retrospective chart review-based study included patients with histologically confirmed STOs of CRC origin treated between January 2000 and December 2019.’ I doubt that the study took 20 years to complete!

2. The manuscript is hard to follow because authors have jumped from one topic to another abruptly. For example,(line 29 to 31), the authors describe the unique characteristics of STOs of right-sided origin and in the following line, jumps to “ Median overall survival was 20.5 months (range 31 0.1–183.5).’- are they talking about the median survival of the whole group or the patients with right-sided tumors?
3. Incomplete information- for example, lines 32 and 33, - ‘Presence of macroscopic residual disease, ….’- are they referring to residual disease after cytoreductive surgery?
4. Lack of focus: Abstract is not presented in a concise manner. I think the abstract should focus on the unique characteristics of the right vs. left-sided tumors and that bilateral adnexectomy should be advised for patients who present with unilateral disease as radiologic studies often fail to detect the contralateral occult disease. Rest can go to the result section.

 

5. Line 128- The mean age of all the patients with CRC metastatic to the ovary was 59.0 ± 12.7 years- not sure what that means. I would suggest including median age with range here.
6. ‘we propose transperitoneal spread as the predominant metastatic pathways for STOs derived from right-sided primary CRC and hematogenous/lymphogenous metastatic pathway in left-sided primary tumors.’- this statement is overreaching given the small patient number.
7. Needs extensive language editing for clarity.

Author Response

The authors have performed a chart review-based retrospective analysis of CRC
patients who presented with ovarian metastasis to evaluate if tumor sidedness
influenced the natural history. This analysis concluded that ovarian metastases
originating from the right-sided primary tumors were more frequently bilateral,
associated with peritoneal carcinomatosis, had ovarian surface involvement by
the tumor, and contained a mucinous component. Furthermore, the authors have provided data illustrating that the imaging methods are inaccurate to determine the laterality of ovarian metastases correctly compared to histopathological examination. The authors have also postulated that transperitoneal spread is the predominant metastatic pathway for the right-sided tumors, and hematogenous/lymphatic spread is the dominant pathway for the left-sided tumors. The data presented in this manuscript are clinically relevant and deserve attention.
However, the manuscript has several significant flaws, as listed below:
1. The meaning of a sentence is often unclear. For example:
Line 22-24- ‘This retrospective, observational chart review study was conducted
between January 2000 and December 2019 and included patients with
histologically confirmed STOs of CRC origin.’
I think the authors wanted to say- ‘This retrospective chart review-based study
included patients with histologically confirmed STOs of CRC origin treated
between January 2000 and December 2019.’ I doubt that the study took 20
years to complete!

The inaccurate statement regarding the study information was revised. The
study included patients with histologically confirmed STOs of CRC origin
diagnosed between January 2000 and December 2019.

2. The manuscript is hard to follow because authors have jumped from one topic
to another abruptly. For example,(line 29 to 31), the authors describe the unique
characteristics of STOs of right-sided origin and in the following line, jumps to “
Median overall survival was 20.5 months (range 31 0.1–183.5).’- are they talking
about the median survival of the whole group or the patients with right-sided
tumors?
The abstract and manuscript were edited to provide intelligible information.
Academic paper reading services (Scribendy) were used to improve language
quality.

3. Incomplete information- for example, lines 32 and 33, - ‘Presence of
macroscopic residual disease, ….’- are they referring to residual disease after
cytoreductive surgery?
The macroscopic residual disease is referring to macroscopic residual disease
after cytoreductive surgery (i.e., R2 resection). This information was added to
the manuscript.

4. Lack of focus: Abstract is not presented in a concise manner. I think the
abstract should focus on the unique characteristics of the right vs. left-sided
tumors and that bilateral adnexectomy should be advised for patients who
present with unilateral disease as radiologic studies often fail to detect the
contralateral occult disease. Rest can go to the result section.
The abstract was extensively edited to increase clarity. However, we would
prefer to keep the information regarding prognostic factors in the abstract as
this was one of our main objectives.

5. Line 128- The mean age of all the patients with CRC metastatic to the ovary
was 59.0 ± 12.7 years- not sure what that means. I would suggest including
median age with range here.
We do agree that the median age (together with range) provides more
comprehensible information on the population age distribution than mean age
+ SD. Therefore, we included this information in the manuscript. However, we
preserved the mean age + SD of the whole study population and subgroups of
patients with right-sided and left-sided tumors in Table 1. With regards to the
result of D'Agostino-Pearson normality test proving normality of age
distribution in both subgroups (right-sided and left-sided primary tumors), we
believe that using a two sample t test together with mean age represents an
appropriate method to compare the age distribution between the two
subgroups.

6. ‘we propose transperitoneal spread as the predominant metastatic pathways
for STOs derived from right-sided primary CRC and hematogenous/
lymphogenous metastatic pathway in left-sided primary tumors.’- this statement
is overreaching given the small patient number.
We do agree that such strong conclusions cannot be drawn from the
retrospective study with limited number of individuals included. Therefore we
reformulated the conclusion as follows: “The results of our study suggest that,
while hematogenous and/or lymphogenous spread to the ovaries seems to be
the preferred metastatic pathway from left-sided primary tumors, peritoneal
spread might be the more common pathway of right-sided primary tumors.
However, more extensive research is clearly needed to support this hypothesis.”

7. Needs extensive language editing for clarity.
The manuscript was edited to increase intelligibility. Editing and proofreading
services (Scribendi) were used to increase the language quality of the
manuscript.

Reviewer 3 Report

This manuscript describes clinicopathologic characteristics of metastatic ovarian tumors of colorectal origin, especially focusing on the location of the primary tumors. It is a retrospective study with a rather small sample size and its impact on practical medicine may be limited, however, the authors’ viewpoint is quite unique and the results provide us with new insights on the way of colorectal cancer metastasis. The manuscript is well written in a concise manner, still it needs a considerable revision to be published.

 

Major

1. According to Table S1, twelve patients did not have bilateral adnexectomy, i. e. one ovary had not been extracted. Although laterality of STO (unilateral vs. bilateral) is given as a histological information (line 78), how was the histological presence of tumor determined for these unresected ovaries?
2. The authors recommend bilateral adnexectomy as the treatment of STO because imaging studies often fail to find one of bilateral tumors. However, if intraoperative inspection is accurate enough to detect all STO, bilateral adnexectomy is not always necessary. It might be better to add information on intraoperative findings and whether tumor is macroscopically visible or not. The reason why imaging studies failed diagnosis should be discussed.
3. Readers may want to know which side of the ovary (left of right) is involved for unilateral STO cases.
4. 95% CI or p value of the multivariable analysis are not shown.
5. To discuss the possibility of peritoneal spread of right-sided colon cancer, the information on whether the primary tumor had penetrated to peritoneum (pT4) is important.
6. The values of each factor on Table S2, except tumor location, could be presented separately according to tumor sidedness, just like Table 1 and Table 2.
7. Against the title of the study, little space is given to sidedness of primary tumor in the discussion. Comparison with previous studies (novelty of this study), differences in molecular pathogenesis between right- and left-sided colon cancer and the role of mucinous component might be included.

 

Minor

1. The values of “extraovarian metastases” on Table1 and “distant metastases” in the text (line 155) don't match. Does “extraovarian metastases” include peritoneal disease (“distant metastases” + “peritoneal carcinomatoses” = “extraovarian metastases”)? The authors’ definition should be described clearly to avoid such confusion.
2. The description of the relationship between radiologic assessment and histological diagnosis is hard to understand. A new table would be a help for readers.
3. The term “no pathological findings” on imaging studies is quite confusing in the context of this manuscript because it reminds readers histopathological findings (lines 180 and 260). They should be replaced with other words.
4. The last part of a sentence is missing (line 167).
5. The conclusion in the abstract sounds greatly over-interpreted: “we propose transperitoneal spread as the predominant metastatic pathways for STOs derived from 36 right-sided primary CRC and hematogenous/lymphogenous metastatic pathway in left-sided primary tumors.” The expression would better be softened.

Author Response

This manuscript describes clinicopathologic characteristics of metastatic ovarian
tumors of colorectal origin, especially focusing on the location of the primary
tumors. It is a retrospective study with a rather small sample size and its impact
on practical medicine may be limited, however, the authors’ viewpoint is quite
unique and the results provide us with new insights on the way of colorectal
cancer metastasis. The manuscript is well written in a concise manner, still it
needs a considerable revision to be published.
MAJOR:
1. According to Table S1, twelve patients did not have bilateral adnexectomy, i. e.
one ovary had not been extracted. Although laterality of STO (unilateral vs.
bilateral) is given as a histological information (line 78), how was the histological
presence of tumor determined for these unresected ovaries?
In those cases where unilateral adnexectomy/ovariectomy was performed, the
information regarding laterality of STOs was derived from intraoperative
findings. However, considering that neither intraoperative findings provided
reliable evaluation of the laterality of STO, we closely analyzed the data from
follow-up investigations. Early documented relapse in the preserved ovary
within 6 moths after surgery would be considered as initially bilateral STO.
However, no such case was documented. Only one patient developed relapse
in the preserved ovary 8.1 months after the initial surgery. Considering the
length of relapse-free survival, the STO was considered as metachronous in this
particular case. The information on methodology was added to the Materials
and Methods section.

2. The authors recommend bilateral adnexectomy as the treatment of STO
because imaging studies often fail to find one of bilateral tumors. However, if
intraoperative inspection is accurate enough to detect all STO, bilateral
adnexectomy is not always necessary. It might be better to add information on
intraoperative findings and whether tumor is macroscopically visible or not.
The intraoperative findings were compared to histopathological reports. In case
of unilateral adnexectomy, the preserved ovary was considered free of STO if it
appeared normal during the surgery and no relapse was documented in this
ovary within a 6-month follow-up. The discordance was found in 20% of cases –
bilateral STOs were misdiagnosed as unilateral in 5/50 (10%) patients and
unilateral STO were misdiagnosed as bilateral in 4/50 (8%) patients. One case
of bilateral STO with normal gross appearance was documented.
Despite intraoperative findings provided more accurate information regarding
the presence of STOs than imaging methods in our study, they cannot be
considered a reliable method to exclude ovarian spread.
The information on intraoperative findings and their ability to assess the
presence of STO was added to the manuscript.
The reason why imaging studies failed diagnosis should be discussed.
The reason for low accuracy of imaging methods was discussed in detail in the
manuscript.

3. Readers may want to know which side of the ovary (left of right) is involved for unilateral STO cases.
We added the information on the laterality (left vs. right ovary) of unilateral STO
to the Results section of the manuscript. Furthermore, the observation that
there was no association between the primary tumor sidedness and laterality of
STO (i.e. right-sided primary tumors showed no propensity to form unilateral
STO in the right ovary and likewise in the left-sided tumors) was noted. Such
findings have been reported previously by several authors:
Blamey, S.; McDermott, F.; Pihl, E.; Price, A.B.; Milne, B.J.; Hughes, E. Ovarian
involvement in adenocarcinoma of the colon and rectum. Surgery, gynecology
& obstetrics 1981, 153, 42-44.
Wright, J.D.; Powell, M.A.; Mutch, D.G.; Rader, J.S.; Gibb, R.K.; Huettner, P.C.;
Herzog, T.J. Synchronous ovarian metastases at the time of laparotomy for
colon cancer. Gynecologic oncology 2004, 92, 851-855, doi:10.1016/j.ygyno.
2003.12.017.

4. 95% CI or p value of the multivariable analysis are not shown.
We acknowledge that it is a common practice to show also confidence
intervals and p-values for estimates of parameters in “classical” Cox regression,
however, we had to opt for an alternative which does not allow meaningful
calculation of these values. These are our reasons:
Our data set was rather small and contained a high number of variables
relative to the number of patients, which could easily lead to overfitting if we
used “classical” Cox regression based on the ordinary least squares method
(OLS). (“Overfitting” means that the model provides a good fit to the set of
data on which it was built, however, as soon as it is applied to another set of
similar data, it performs badly, because it is too tailor-made for the concrete
data set. And if there are few records (patients) the danger of overfitting is
high.) To solve the problem of relatively high dimensionality (i.e. number of
potential predictors) with regards to the small sample size, we opted for
penalized Cox regression. Due to existence of collinearity we chose the elastic
net type of penalization, which can provide a sparse solution (i.e. reduce the
number of predictors).
The method we used for multivariable analysis (i.e. Cox regression with elastic
net regularization) has many advantages, however, to our knowledge there is
not yet any consensus about suitable methods for calculating confidence
intervals and other quantities based on the mean-square error of coefficient
estimates, though there has been intensive research in this area of statistics in
the past few years. To the problem itself: It is possible to calculate meansquare
errors for the coefficients (like it is in traditional Cox regression), however, they would not be very meaningful and could even be misleading (and that we endeavoured to avoid). The reason is that the penalized regression introduces a bias into the estimates of the parameters. Their estimates are set lower (biased) to decrease the risk of overestimating the influence of the tested parameters (and thus to decrease the risk of overfitting). That results in a somewhat worse fit within the sample used for the model building, but this seemingly “worse” fit is more robust, i.e. it provides better predictions if used in the future on other similar sets of data. However, given the current state of the art, a problem of this approach is to get a sufficiently precise estimate of the bias. This is a hurdle for any meaningful calculation of mean-square errors, confidence intervals, and significance tests.
The elastic net Cox regression (using a mixture of ridge and lasso restrictions)
sets parameters with a weak influence completely to zero, and “classical”
inference can be at least partly replaced by cross validation. The cross
validation process is based on dividing the sample into several subsamples
(we used three), taking data from all but one subsample, fitting the model to
the data, and testing the quality of its predictions on the remaining subsample.
This is repeated for each combination of subsamples, and average values of
the estimated parameters across all these (three) fits are calculated. This
procedure was used for various degrees of penalization in order to choose the
penalty leading to the model with the best prediction abilities.
To conclude, according to the model, the parameters with non-zero
coefficients (see Table 3) contribute to the quality of the predictions that can
be obtained from this model, whereas the parameters with a coefficient equal
to zero did not exhibit a relevant influence on the predictions. However, it
should be borne in mind that the data sample was small. An analysis of a
larger sample might be able to reveal also an impact of some parameters for
which we got coefficients equal to zero.
For reference cf. L1 and L2 Penalized Regression Models by J. Goeman, R.
Meijer and N. Chaturvedi
https://cran.r-project.org/web/packages/penalized/vignettes/penalized.pdf
(page 18-19)

5. To discuss the possibility of peritoneal spread of right-sided colon cancer, the
information on whether the primary tumor had penetrated to peritoneum (pT4) is important.
The information on T4 tumors proportion (in both right-sided and left-sided
primary tumors) was included in Table S2. Notably, we could not observe any
statistically meaningful difference in the proportion of T4 tumors between rightsided and left-sided primary CRCs (p = 0.764), which was further discussed.

6. The values of each factor on Table S2, except tumor location, could be
presented separately according to tumor sidedness, just like Table 1 and Table 2.
Table S2 was rearranged to include information on the right-sided and leftsided
primary tumors separately (with the exception of primary tumor site).
Furthermore, the information regarding primary tumor stage (T4 vs. ≤ T3) was
added.

7. Against the title of the study, little space is given to sidedness of primary tumor in the discussion. Comparison with previous studies (novelty of this study), differences in molecular pathogenesis between right- and left-sided colon cancer and the role of mucinous component might be included.
The topic of primary tumor sidedness was addressed more broadly in the
discussion section, including the role of mucinous histology in the
pathogenesis of peritoneal spread. Moreover, the hypothesis that the CMS1
and CMS4 subtypes of CRC are prone to form ovarian metastases was formed.
MINOR:
1. The values of “extraovarian metastases” on Table1 and “distant metastases” in
the text (line 155) don't match. Does “extraovarian metastases” include
peritoneal disease (“distant metastases” + “peritoneal carcinomatoses” =
“extraovarian metastases”)? The authors’ definition should be described clearly
to avoid such confusion.
We agree that this section might be confusing. Therefore we stated that
“extraovarian metastases” mean the presence of peritoneal carcinomatosis
and/or the presence of distant metastases while the term “distant metastases”
stands for the presence of extraovarian disease excluding peritoneal spread.

2. The description of the relationship between radiologic assessment and
histological diagnosis is hard to understand. A new table would be a help for
readers.
A supplementary table 3 was created to include information regarding
imaging methods and their comparison to histopathological findings.

3. The term “no pathological findings” on imaging studies is quite confusing in
the context of this manuscript because it reminds readers histopathological
findings (lines 180 and 260). They should be replaced with other words.
We agree that the term “no pathological findings” might be misleading. It was
therefore replaced by the term “no visible adnexal mass”.

4. The last part of a sentence is missing (line 167).
The missing expression “Table 2” was added to the manuscript.

5. The conclusion in the abstract sounds greatly over-interpreted: “we propose
transperitoneal spread as the predominant metastatic pathways for STOs derived from 36 right-sided primary CRC and hematogenous/lymphogenous metastatic pathway in left-sided primary tumors.” The expression would better be softened.
We do agree that such strong conclusions cannot be drawn from the
retrospective study with limited number of individuals included. Therefore we
reformulated the conclusion as follows: “The results of our study suggest that,
while hematogenous and/or lymphogenous spread to the ovaries seems to be
the preferred metastatic pathway from left-sided primary tumors, peritoneal
spread might be the more common pathway of right-sided primary tumors.
However, more extensive research is clearly needed to support this hypothesis.”