Pre-Treatment C-Reactive Protein Predicts Survival in Small Cell Lung Cancer Patients

Round 1

Reviewer 1 Report

This is a study of OS and pre-treatment CRP in the Danish Cancer Cohort. Overall, this is a novel and thorough investigation of one of the largest cohorts of SCLC and OS. The methods and results of this paper need some additional clarification prior to publication for full transparency and interpretation. 

The definition of AIC is fine but it's confusing to have both the nmol/L and mg/L in the text. What is the additional benefit of including both? I couldn't determine if the converted all patients for both measurements and which was used as a covariate in the Cox HR predictor analysis and AIC. Please clarify this. Also there is no mention of the use of quartiles in the methods only the binary cutpoint so this needs to be added and clarified as well.

Given the strong association between CRP and smoking and the low number of non-smokers or patients with an unknown smoking status it would have made sense to do a sensitivity analysis in this group alone to determine how prediction using CRP could improve. I think this would be a valuable addition to the paper.

It is listed in the paper that a DAG was used to select for confounders but was this based on the association between CRP alone (Table 1) or were additional analysis done? More information needs to be included on this since the paper mentions Chi-square and the rank sum test but why were these tests included? Were these a preliminary step for confounder selection for the DAG and were other covariates considered? More information on this is critical for interpretation of results and transparency. Why wasn't performance status included in the Cox PH analysis? Table 2 is confusing because statistically it doesn't make sense to include CRP in both forms (as binary and in quartiles). Were these separate models? Were Cox PH models used as a selection for AIC because if so why is PS included in those but not in the Cox PH models? All of this is very confusing. 

For follow-up time I assume the error in the text that lists all patients had a follow-up time of at least 2 days and not 2 years. Please correct this since all of the overall survival is listed in months and the median is listed as less than 1 year. 

There is no supplementary table 1 even though it is mentioned in the text. Figure 1 has an error and lists 'sodium' measurement and not CRP. There are also a number of typos throughout the text that should be cleaned up prior to acceptance.

I believe the discussion section is not supported by what is currently in the results section. The C-index in table 3 is not improved by very much with the inclusion of CRP (and again is this quartiles or binary?). I feel the findings are very overstated unless some of the results were not included with submission. The first paragraph of the discussion has results that should be moved to the results section and do not belong in the discussion "Yet, when exploring the different levels of CRP, we found 191
that the association was only significant for the highest CRP levels (3
rd quartile: 18 - 59 192 mg/L / 221 - 560 nmol/L and 4th quartile: >59 mg/L / >560 nmol/L)." Where is this information in the results? It makes it very difficult to follow the findings.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

Submitted manuscript titled “Pre-Treatment C-Reactive Protein Predicts Survival in Small Cell Lung Cancer Patients”, by Stensvold et al investigated the clinical benefit of CRP in small cell lung cancer. The article was generally well written, there are some clinically important findings.

1. CRP is known to be elevated in several inflammatory conditions including bacterial, viral and autoimmune diseases. Were these patients excluded from the study
2. In figure 1, why is sodium mentioned?
3. Please explain the criteria for setting the CRP cut-off to 8, and add the cut-off information of previous studies to the discussion section.
4. Usually CRP and albumin have been studied together as cancer blood biomarkers. It is expected that better results will be obtained if the contents of albumin are added.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Thank you for the clarification and updates. My concerns have been addressed but I think it would be useful to add to the methods that statistical tests weren't used to select for confounders, it was based on the literature and placed into a DAG. This is a less typical approach and needs to be clear for the readers. I don't understand the use of statistical tests (Chi-square rank sum) if the weren't used to select confounders for adjustment. They are confusing and don't add to the overall message of the paper-it is sufficient to describe the populations using descriptive statistics. I think a reference from the literature supporting the relevance of using a DAG for selection of confounders would also be warranted as it's a less common method for selection and the reader may have further questions. 

Author Response

We agree that further clarification of the method section would improve the manuscript. Therefore, it has now been added that confounders included in the DAG are based on the literature, just as a reference for the DAG model has been added. Page 3, line 113.

Furthermore, the use of chi square test and rank sum as part of the descriptive statistics has been precised. Page 3, line 103.    

Reviewer 2 Report

The authors responded appropriately and this paper is considered suitable for publication.

Author Response

Thank you for the constructive comments which have been addressed in the previous response.