Cognitive Trajectories in Older Patients with Cancer Undergoing Radiotherapy—A Prospective Observational Study

Round 1

Reviewer 1 Report

 

The purpose of this study was to assess cognitive trajectories in older patients undergoing radiotherapy for cancer based on the cognitive screening test MoCA and associated factors.

According to the lack of studies focused on older cancer patients, this study is interesting. Furthermore, the study methodology is good and the manuscript well-written. I have only minor comments.

 

-          Abstract:

o   Line 39: the authors should add in the abstract than the decline of MoCA score in the very poor group is transient.

 

-          Results:

o   Patient characteristics included the number of daily medications. Is psychotropic medication use is known? (Psychotropic medication use is associated to cognitive impairment, especially in older people).

o   3.2. Lines 230-232: description should start with the most impaired cognitive domain: working memory.

o   Lines 234-237: The description of the distribution of MoCA z-scores according to Figure 2 is my main concern about this paper. Is not easy to find 62.1% of patients with score <1SD below the normative mean and even more 37.9 ≥1SD. According to the Figure 2, 62.1% of patients had score >1SD of the normative mean (normal range) and 37.9%? (I cannot find this percentage on the graph) had MoCA scores ≥-1SD for the normative mean (cognitive impairment).

 

-          Discussion:

o   Line 362: the authors should add “for frail patients” at the end of the sentence.

o   Paragraph on the study limitations: the authors should add that only the version 7.1. of the MoCA was used and not the parallel version. A practice effect could exist, especially since multiple assessments are very close together: 8 weeks apart.

 

-          Conclusion:

o   Line 384: Instead of “high prevalence” the percentage of cognitive impairment should be added.

o   This paragraph should also include a sentence of the deleterious effect of the radiotherapy only for the very poor MoCA group and the notion of a very transient effect (improvement at 16 weeks).

 

-          Figures:

o   Figure 2: The graph is not easy to understand. The authors should add if these are cumulative percentages if they are.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The authors describe an observational study considering the cognitive trajectories in older patients with cancer undergoing radiotherapy assessed by means of the Montreal Cognitive Assessment (MoCA) at T0 (baseline - before RT), T1 (immediately after RT), T2 (8 weeks after RT) and T3 (16 weeks after RT). They found that cognitive impairment is quite frequent in elderly people undergoing RT but that RT had a negative cognitive impact only in few of the cases (i.e. those more compromised and fragile).

Although non particularly innovative, the study appears linear and "straightforward", and has the merit of longitudinally assessing and following a population of patients, showing that RT can be regarded as relatively "safe" also on an elderly population. My main concern is however related to the lack of a strong rationale and background hypothesis to assess.

Herewith my comments:

INTRODUCTION:

1) in general I would suggest a double-check for the English language

ASSESSMENT:

2) p.3 line 122: MoCA does not assess "executive functions". Some of the measures do, but in general it is an overall cognition measure. I guess the authors meant "cognitive functions" instead of "executive functions".

3)p.3 line 129: the authors claim that "per protocol, T1 assessment was omitted for patients receiving <9RT fractions". Could the authors specify why was this done?

STATISTICAL APPROACH:

4) p.4, line 155: why did the authors transform the education level measure (which is a continuous variable) into an ordinal scale?

5) p.4, line 159: "Spearman’s rho was calculated among all predefined variables (data not shown), however, no multicollinearity issues were identified." It would have been good to have the possibility to look at the correlation matrices. Could the author prepare a table (maybe in supplementary materials) showing these data? For transparency.

RESULTS:

6) Results and analyses are largely descriptive. The main statistical comparison is that performed by means of growth mixture model identifying the 4 groups of MoCA performance and showing that only for the "Very poor" T0 group a decline was found through T1 and T2, followed by a recovery. However, as also acknowledged by the authors, the result is a bit misleading since at the different stages the number of patients contributing the mean score changes, due to attrition (which is partcularly heavy for the "Very poor" group). Basically, at T3 all the more compromised patients had died, thus pushing the mean score artificially higher. Thus, the claim that they face a "transient decline" (p.10, line 290) turns out to be questionable. Indeed one could argue that the clinical decline progressed through T3, leading to the death of the patients. It would have been better in my opinion to perform the analysis on those patients assessed in all conditions and times, to get a more reliable picture of the situation. or at least to present both analyses (maybe one in the supplementary material).

 

7) Table 1 (p.4-5) summarizes the clinical data according to the MoCA group classification. However at that point in the paper the analysis was not yet performed, so it would be better for the reader to have Table 1  placed at p.9, where the MoCA scores trajectories are discussed.

8) p.9, line 273: the authors refer to a Table 5 which however is not present anywhere in the paper.

DISCUSSION:

9) It basically seems that the "Very poor" MoCA group is composed by the most frail people with the higher number of comorbidities and negative prognostic factors (including undergoing RT for "palliative" approach). I think that, particularly the patients undergoing palliative RT should be treated separately, since their general clinical picture is likely to be quite different already at baseline T0.

10) related to this: it is not very clear to me what is the final point in showing that the most frail group is also the one more compromised from a cognitive point of view. What I mean is that, with the current analysis, MoCA scores could be treated as one of the many clinical variables considered (equally important), since they do not seem to be more predictive or more reliable than other factors. What does the MoCA administration add in the evaluation of these patients? Particularly since the authors (in the discussion: p. 10, lines 310-311) stated that already many past studies showed that the more frail patients also show the most severe cognitive deficits. This is what I mean by saying that I fail to find a strong rationale and background experimental hypothesis behind the study.

It would have been very informative for example to attempt to "rank" all the clinical factors considered to see what is the role (or the weight) of cognition in the model. Or to see which of them had the higher predictive value over RT tolerance. Or over cognition.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

I am satisfied with most of the answers and revisions provided by the authors. 
I am afraid however to have to insist into just one point, and I would not insist if the explanations would not potentially be opposite. And this relates to the account for the "recovery" of MoCA scores at T3.

I agree that both approaches (including vs. excluding the drop-off patients) have limitations and lead to potential biases. That is why I was suggesting to show both results. In order for the reader not to being misleaded into thinking that RT has a transient negative effect on cognition even in the most frail group. Unfortunately I do not think that just rewording the sentence substituting “experienced” with “was registered” is sufficient. I am sorry to insist on this point, but this is since I, myself, was misleaded in the interpretation of the results (when I first read the paper) looking at the graph and was quite disappointed in realizing that the “improved” T3 scores did not consider the patient who had died. And I would not insist if the potential accounts for T3 score levels would not be  “opposite” (i.e. the patients recovered vs. the patients worsened at the point that many of them had died  thus not contributing to the final score). Is there any particular reason making it hard to perform the same analysis considering alternatively only the completers or the whole sample? This would make the results much clearer and informative to the reader. And I think that (whatever the result) it would not diminish the solidity of the work, which is only explorative and not testing a particular hypothesis.

Author Response

Please see attachment.

Author Response File: Author Response.docx