Clinical Characteristics, Treatment, and Short-Term Outcome in Patients with Heart Failure and Cancer

Round 1

Reviewer 1 Report

Thank you for the opportunity to read your work.

I understand that there are few patients who met the inclusion criteria of the study, however, divide a group of 70 patients in three sub-groups removes any statistical power of the analysis. The study is still interesting if you remove this analysis. You can only compare patients of Group A and B.

You need to clearly state what kind of study do you want to perform. Is it exploratory? Is it a correlational study? 

If you use ESC guidelines, you must characterize patients according to the three types of HF: HFpEF, HFmEF and HFrEF. 

Table 4: NYHA class can not be presented like this. You need to state how many patients are in NYHA II, III or IV. Also the heading of the table is too long and confusing 

In general, there are too many variables for a the number of patients. It is important to clarly define what is more important to analyse at this moment and to do a solid analysis and then, with a large sample, to seek for other conclusions. 

It is possible to compare treatment but there are a lot of confunders that can lead to this results.

Author Response

We wish to express our appreciation for the comments and suggestions for our manuscript entitled “Clinical characteristics, treatment and short-term outcome in patients with heart failure and cancer”. We have carefully revised the manuscript taking into consideration all the comments.

Our changes are in blue in the revised text

RESPONSES TO Reviewer # 1:

I understand that there are few patients who met the inclusion criteria of the study, however, divide a group of 70 patients in three sub-groups removes any statistical power of the analysis. The study is still interesting if you remove this analysis. You can only compare patients of Group A and B.

As stated in the text (116-118 lines) data on the patients treated with chemotherapy (A3) was scarce, not fully available and this is the cause of incomplete analysis. 

Finally, according to the reviewer’s suggestion we decided not to define chemotherapy patients as a separate subgroup, to delate its definition and the results from the analysis concerned chemotherapy group. We agree with the reviewer that the shortage of data in chemo- group significantly limits the validity of the results.

You need to clearly state what kind of study do you want to perform. Is it exploratory? Is it a correlational study? 

This is retrospective, analysis of medical records. The study is exploratory - differences in clinical cardiovascular status between of ca- and non-ca patients (what was explored) are not known and are not understood.

Second part of the analysis was dedicated to identify correlations between in-hospital death in ca (Group A) and non-ca patients (Group B) and analyzed parameters (correlational study).

If you use ESC guidelines, you must characterize patients according to the three types of HF: HFpEF, HFmEF and HFrEF. 

In 2016 ESC HF guidelines first time distinguished three types of HF according to EF. HF with mid-range EF was defined as EF within the range 40-49%. It is nonnatural division resulting from some practical and not biological and not pathophysiology reasons. EF is continuous parameter with normal distribution in population. In 2021 there was a great discussion whether to maintain any EF dependent HF classification. Finally, 2021 ESC HF guidelines renamed HF with mid-range to HF with mildly reduced (HFmrEF). Most trials divide HF patient into HFpEF and HFrEF and we did so as well.

The threshold of 40 % for EF is significant in terms of the treatment because all medications modifying prognosis have been proved and are recommended from this level. The comparison of the guideline directed treatment in the respect to the threshold EF = 40% was presented in the text (lines 138-142):

“However, 22 (82%) and 21 (78%) of 27 Ca patients, 39 (87%) and 35 (78%) of 45 non-Ca patients with EF < 40% were on beat-blockers and mineralocorticoid receptor antagonists, respectively which did not make statistically significant difference (p=0.74; p=1.0, respectively.”

However, according to 2021 ESC HF guidelines for HFmrEF patients the same treatment may be considered. Because the main aim of our study was to look at the clinical course, we decided to divide the patients into HFpEF ≥ 50 and HFrEF < 50%. The study group is too small to make up more than two subgroups. After division the study would be less clear, any statistical power might be lost and in it would add no more to analyses. 

Table 4: NYHA class cannot be presented like this. You need to state how many patients are in NYHA II, III or IV. Also, the heading of the table is too long and confusing 

NYHA stage was analysed by Mann-Whitney U test. If NYHA class distribution between the study groups is approached by the non-parametric test it may be reported as a mean +/- standard deviation. We decided to present the results of NYHA class as a mean +/- standard deviation because it is more „readable”.

The heading of the table 4 was changed.

In general, there are too many variables for a the number of patients. It is important to clearly define what is more important to analyse at this moment and to do a solid analysis and then, with a large sample, to seek for other conclusions. 

The study has retrospective nature. All parameters taken into account in the analysis are routinely used in clinical practice to characterize any particular patient. Having such a set of parameters we tried to identify the significant differences between ca- and ono-ca individuals in terms of them. We do not see any reason for which some of the parameters should be dropped from the comparison. We also have no idea what criteria of the selection for the parameters might be applied.

It is possible to compare treatment but there are a lot of confunders that can lead to this results.

We compared the rate of the use of each drug in Group 1 and Group 2. No conclusions on the results were made.

Mantel-Haenszel (M-H) estimator was used to remove the confounding effects in multivariate analysis.

When HgB; K+; AST; HPEF were removed from the analysis and only three parameters left the results are as follows:

The results did no change significantly.

Author Response File: Author Response.docx

Reviewer 2 Report

Authors addressed an intriguing issue concerning the HF management in patients with cancer. These data were interesting. However, the small number of patients enrolled in this study strongly limits the value of the results obtained.  Major points are:

• I don't feel qualified to judge about the English language and style but the abstract does not present the work in an interesting way
• I don't understand the meaning of the sub-group of patients who undergo chemotherapy. The authors performed only a marginally e partially showed analysis on this subgroup
• It would be interesting to see the echocardiographic data on the right heart sections
• It would be interesting to see the complete tables with univariate and multivariate analysis

Author Response

We wish to express our appreciation for the comments and suggestions for our manuscript entitled “Clinical characteristics, treatment and short-term outcome in patients with heart failure and cancer”. We have carefully revised the manuscript taking into consideration all the comments.

Our changes are in blue in the revised text

RESPONSES TO Reviewer # 2:

Authors addressed an intriguing issue concerning the HF management in patients with cancer. These data were interesting. However, the small number of patients enrolled in this study strongly limits the value of the results obtained.  Major pointa are:

• I don't feel qualified to judge about the English language and style but the abstract does not present the work in an interesting way

I don't understand the meaning of the sub-group of patients who undergo chemotherapy. The authors performed only a marginally e partially showed analysis on this subgroup.

As stated in the text (116-118 lines) data on the patients treated with chemotherapy (A3) was scarce, not fully available and this is the cause of incomplete analysis. 

Finally, according to the reviewer’s suggestion we decided not to define chemotherapy patients as a separate subgroup, to delate, its definition and the results resulting from the analysis concerned chemotherapy group. We agree with the reviewer that the shortage of data in chemo- group significantly limits the validity of the results.

It would be interesting to see the echocardiographic data on the right heart sections

Data on right heart (right ventricle and right atrium) are not available. We did not pay attention to these parameters as imaging of the right heart by echo- has not been well validated and has very limited practical value.

• It would be interesting to see the complete tables with univariate and multivariate analysis

We decided not to present nonmeaningful data.

Below univariate analysis by Mann-Whitney test and Ch2 test for categorical variables.

Tabela 2 – Test Manna-Whitneya (wartość p), w tabeli podano liczbę i procent osób z danym czynnikiem ryzyka. Grupa 1A – chorzy z nowotworem, którzy zmarli w trakcie hospitalizacji, grupa 1B – chorzy z nowotworem, którzy przeżyli w trakcie hospitalizacji.

Na żółto zaznaczono wartości p istotne statystycznie w analizie jednoczynnikowej.

The association of medication with higher mortality was skipped over as the association is obvious – more severely ill needs mor treatment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Authors addressed an intriguing issue concerning the HF management in patients with cancer. These data were interesting. However, the small number of patients enrolled in this study strongly limits the value of the results obtained.  Major pointa are:

• I don't feel qualified to judge about the English language and style but the abstract does not present the work in an interesting way

The authors did not respond to the comment. The abstract in this form is not acceptable for a scientific journal. The main problems are:

Background: Cancer and heart failure (HF) often coexist. These two pathologies interacts on numerous clinical platforms. These sentences do not introduce the purpose of the work, they are too general.

Our study aimed to look at the clinical characteristic, treatment and short term outcome of patients hospitalised due to heart failure with coexisting cancer.

Methods: Seventy one cancer patients hospitalised due to exacerbation of HF were identified retrospectively. Data on clinical characteristics were collected. Control group is not mentioned.

Results: Cancer patients presented with less advanced NYHA class had more frequently HFpEF, higher peak troponin T level and smaller left atrium size (…then non Cancer).

This group were less frequently treated with beta-blockers and mineralocorticoid receptor antagonists. As the authors reported in the new version of the manuscript, the relative % of patients on BB and MC antagonist with reduced ejection fraction (EF<40%) did not differ between cancer and no cancer group (line 141-145. Please, in this section change “beat” blockers with beta blockers). This absolute difference is probably due to the higher prevalence of HFpEF in cancer group. BB and MC antagonist are not mandatory treatments for patients with HFpEF. So, I don’t think that this is an essential information for the abstract.

In-hospital death was associated with: higher NYHA class, lower HgB level, worse renal function, higher K and AST levels, presence of diabetes mellitus and HFpEF.

The authors shown only the Mann-Whitney test, this is only a way to compare variables without a not normal distribution between two groups. With this analysis you can say only that these two groups have different median values. To investigate the dependence between two variables you have to perform a univariate regression analysis (see below in the comment on the not shown tables provided from the authors) (please change again the title of the table 4)

Data on length of hospital stay, in-hospital bleedings and in-hospital mortality between cancer and no cancer (line 128-129) are not mentioned but are essential. Please add this information before the analysis on the factors associated with death in cancer patients with HF.

Impaired renal function was the only independent predictor of in-hospital death.

The authors must list the variables of the multivariate model (see below in the comment on the not shown tables provided from the authors)

Conclusions: Cancer and the methods of its treatment may affect the course of HF. The conclusion is not appropriated. The authors addressed the clinical characteristics of patients with HF with and without cancer.

I don't understand the meaning of the sub-group of patients who undergo chemotherapy. The authors performed only a marginally e partially showed analysis on this subgroup.

As stated in the text (116-118 lines) data on the patients treated with chemotherapy (A3) was scarce, not fully available and this is the cause of incomplete analysis. 

Finally, according to the reviewer’s suggestion we decided not to define chemotherapy patients as a separate subgroup, to delate, its definition and the results resulting from the analysis concerned chemotherapy group. We agree with the reviewer that the shortage of data in chemo- group significantly limits the validity of the results.

Ok

It would be interesting to see the echocardiographic data on the right heart sections

Data on right heart (right ventricle and right atrium) are not available. We did not pay attention to these parameters as imaging of the right heart by echo- has not been well validated and has very limited practical value.

Ok, I can understand that you did not register these parameters. However, I don’t think that right echo parameters have not been validated and have limited practical value. Aren’t right atrium area, TAPSE, and peak velocity of tricuspid regurgitation used every day in clinical practice?

• It would be interesting to see the complete tables with univariate and multivariate analysis

We decided not to present nonmeaningful data.

Below univariate analysis by Mann-Whitney test and Ch2 test for categorical variables.

As mentioned above, Mann-Whitney test is not a univariate analysis. You have to perform a univariate regression analysis to provide an OR for every variable considered (Mann Whitney test can only help to choose the variables). Generally, the significant variables in the univariate analysis are inserted together in the multivariate model to investigate which ones are independently associated with the event considered.

The authors report on the manuscript (line 161-163): “By multivariate logistic regression analysis, impaired renal function was the only independent predictor of in-hospital death in Ca patients (OR- 1.15; CI 1.05; 1,27); p=0.017). The following covariates entered the regression: NYHA class, HgB, GFR, K+, AST, diabetes mellitus t.2, HFpEF”. This is the most interesting analysis of the manuscript; you have to shown the data about univariate and multivariate regression analysis. The OR for renal impairment how was calculated? This is an OR for every point in less of eGFR or for the presence of values below a specific cut off?

Comments for author File: Comments.docx

Author Response

We wish to express once more, our appreciation for the valuable comments and suggestions for our manuscript entitled “Clinical characteristics, treatment and short-term outcome in patients with heart failure and cancer”. We have carefully revised the manuscript taking into consideration all the comments. They will contribute to the significant improvement of our manuscript.

Our changes are on the yellow field in the revised text,

while our responses are in blue.

RESPONSES TO Reviewer # 2:

Ad 1.

Background: Cancer and heart failure (HF) often coexist. These two pathologies interact on numerous clinical platforms. These sentences do not introduce the purpose of the work, they are too general.

The above sentences were deleted.

The abstract was corrected and rewritten according to the reviewer’s suggestions.

Suggested definition of controls was added.

This group were less frequently treated with beta-blockers and mineralocorticoid receptor antagonists.

We fully agree with the reviewer’s opinion about insignificancy of the sentence “This group were less frequently treated with beta-blockers and mineralocorticoid receptor antagonists”.

This above sentence was deleted.

“Beat blockers” was changed to the proper form – “beta blockers” (line 145).

Data on length of hospital stay, in-hospital bleedings and in-hospital mortality between cancer and no cancer (line 128-129) are not mentioned but are essential. Please add this information before the analysis on the factors associated with death in cancer patients with HF.

Data on length of hospital stay, in-hospital bleedings and in-hospital mortality between cancer and no cancer (line 129-131) was supplemented.

There were no differences in terms of length of hospital stay (6,0±4,06 vs. 5,7±3,30 days; p=0,32), in-hospital bleedings (9,21%vs.5%; p=0,48) and in-hospital mortality (2,63% vs.2,5%; p=0,65) between non-ca and ca-patients, respectively.

 Also, the length of hospital stay (6,5±7,7vs.6,1±3,8 days; p=0,31) and in-hospital bleedings (17%vs.2%; p=0,08) did not differ in ca- patients who died and who survived, respectively (lines 156-158).

As this data did not differ between the study groups it was not added to further analysis.

Conclusions: Cancer and the methods of its treatment may affect the course of HF. The conclusion is not appropriated. The authors addressed the clinical characteristics of patients with HF with and without cancer.

According to the reviewer’s suggestion the conclusion was changed.

The clinical picture and course of heart failure in patients with and without cancer are different.

Ad2

The authors shown only the Mann-Whitney test, this is only a way to compare variables without a not normal distribution between two groups. With this analysis you can say only that these two groups have different median values. To investigate the dependence between two variables you have to perform a univariate regression analysis (see below in the comment on the not shown tables provided from the authors) (please change again the title of the table 4)

The title of table 4 was changed. It is as follows:

Statistical analysis (Mann-Whitney test and Ch2 test for categorical variables) between independent characteristics and death.

The reviewer is right that Mann-Whitney test and Ch2 test are not the equivalents of univariate analysis. Nevertheless, we think that these tests are precise enough for our study to select parameters different between the study to process them by multivariate analysis.

In future studies we will be using univariate regression for such selections  

Ad 3

“By multivariate logistic regression analysis, impaired renal function was the only independent predictor of in-hospital death in Ca patients (OR- 1.15; CI 1.05; 1,27); p=0.017). The following covariates entered the regression: NYHA class, HgB, GFR, K+, AST, diabetes mellitus t.2, HFpEF”. This is the most interesting analysis of the manuscript; you have to shown the data about univariate and multivariate regression analysis. The OR for renal impairment how was calculated? This is an OR for every point in less of eGFR or for the presence of values below a specific cut off?

The OR is for every point in less of eGFR. In other words the above sentence should be perceived:

By mulitivariate logistic regression analysis, lower value of GFR (the “more” impaired renal function)  was the only independent predictor of in-hospital death in Ca patients.

Author Response File: Author Response.docx

Round 3

Reviewer 2 Report

Thanks you for responding to the comments. You paper is now acceptable for publication