Advancing Academic Cancer Clinical Trials Recruitment in Canada

Round 1

Reviewer 1 Report

Unfortunately, this work is not sufficiently novel or generate added value information to warrant publication at this time.  With respect to weakness of the study the authors have not commented on inherent biases with respect to before-after analysis when looking at the years of pre-CTN baseline with post implementation.  The lack of trial level data, as stated by authors, did limit their ability for more in-depth analysis.  Unfortunately the authors -- through no fault of their own -- were unable to demonstrate any predictors of recruitment achievement by the centers with the data available nor were they able to demonstrate any association between funding and increased recruitment. 

Author Response

Please see the attachment. Thanks.

Author Response File: Author Response.docx

Reviewer 2 Report

Dear authors,

Thank you for writing the manuscript entitled "Advancing academic clinical trials recruitment in Canada". You describing an important and successful initiative to enable expanded accrual to academic clinical trials.  Overall the paper is well written and will likely be of interest to Canadian and international readers. Suggestions below are provided for your consideration, with the intention of easing readability for a wider audience. 

Minor suggestions:

1. A simple figure demonstrating the structure of 3CTN, to accompany text presented in Background section 1 could help understand its design at a glance.

2. Tables: I found the tables dense and it took me a few reads to understand the data contained within.  Consider expanding all abbreviations in footnote at bottom of each table to ease reader experience; e.g. NACC so readers do not have to refer back to text, and Canadian provinces so makes sense to international readers who may be interested in the approach but not familiar with the geography.  Consider adding a comma or some means to separate Median, Range when these are presented in the same row. In Table 2, it appears the footnote symbol may better relate to the column 3 heading, rather than the subsequent "portfolio trial composition" subheading. Was there a trial with a sample size of 164,946 (Table 2)? Maybe correct, a typographical error, or my misunderstanding.   

3. Core funding. The Background section 1 states "NCCs received core funding to support academic cancer clinical trial (ACCT) recruitment based on population served."  Discussion states "Core funding, which was provided by provincial funders to support activities at centres within their jurisdiction was variable." Is this the same core funding or two different sources?  If different, to which does core funding in results tables refer to? If the same, suggest briefly expanding to make the relationship and flow of funds between provincial cancer agencies, clinical trial sites and 3CTN clear.  

4. I enjoyed reading the informative discussion with appropriate Canadian and international references.  https://www.mdpi.com/1718-7729/28/2/143, published in Current Oncology in April 2021, demonstrated a high reliance on industry sponsorship and funding of Canadian cancer clinical trials and could also be referenced if desired. 

Author Response

Please see the attachment. Thanks. 

Author Response File: Author Response.docx

Reviewer 3 Report

Summary

From the abstract: "The Canadian Cancer Clinical Trials Network (3CTN) was established in 2014 to address 13 the decline in academic cancer clinical trials (ACCT) activity." 

The goal of the article is two-fold: describe to what extend 3CTN managed to achieve its goal of increasing trial participation and finding predictive factors for this increase. As it turns out the 3CTN is very successful, surpassing its goal of 50% increase in accrual compared to pre-CTN levels. The quest of the authors to identify predictive factors for increase in accrual is less successful. By their own account among the factors considered only time showed predictive value for accrual. (But see also major comment 1 below)

Overall comment:

From the perspective of the reader the most interesting fact is that the 3CTN apparently worked in getting more patients into trials. Accademic cancer trials are struggling with accrual everywhere in the world so if Canada (seemingly following an example from the UK) found a way to do something about this, that is great news that deserves to be well known.

The reader will be primarily be interested in the question: can we implement this, or parts of this, in our own country? Identifying successfactors is useful for answering this question but first and foremost we want in sections 1 and 2 a much more detailed description of what the 3CTN did (and or a reference to a place where it has been written down before): how was this funding used, what concrete measures were taken to increase accrual etc. Part of this does show up in the article eventually in the discussion but it good to have it earlier so that we better understand the 3CTN before we go into the results. 

Even if you couldn't find predictive factors among the limited number of factors you explicitly put into the model, there must have been something that was succesful among all the other details of the 3CTN (given the overall success) so at least as readers we want to know about those details.

I recommend restructuring the article to focus more on the description of the 3CTN method, unless that has already been done in a previous publication (but then make that more clear). All in all readers will be delighted that there is finally some good news about accrual in academic trials and want to know more about how you achieved that.

Major more detailed comments:

1. The description of the GEE model could be more explicit. The way I understand it now is that at each quarter and each site you computed the accrual expressed as a percentage increase from baseline for that site and regressed these numbers on the factors (charactersitics of the site) listed in table 4 as well as time point. This was not yet the major comment. I just typed this in order to type the following (which IS the major comment):

If this reading is correct, then it seems you are looking for factors predictive of success in the following meaning of the word: factors that (would) explain why some sites are more succesful than others.

However: in light of what I type under 'overall comment' and the overall success of the 3CTN program I would have expected (and as a reader would be more interested in) factors predictive of succes in the following sense: what factors introduced by the 3CTN explain why now there is more accrual than when the 3CTN started. In other words: I would be more interested in comparing 3CTN period vs baseline rather than comparing sites within the 3CTN period. This is a different type of analysis, but if you have the relevant data, please add this analysis!

2. There seems to be a mismatch between the p-values in table 4 and the description of these same p-values in the text. Please check what is going on here.

Minor comments:

1. Please please please do not use the word 'trend' to indicate p-values between 0.05 and 0.1. I will probably not be the first person complaining about this. Only use 'trend' for things that change (in a given direction) over time. In fact in the current article you ARE looking at changes over time so maybe this is what you meant and p-values are actually shrinking over time. In that case you should just clarify this a bit more in the wording but in reality I think that wasn't what was meant.

2.  In figure 1 accrual plumets and then goes up again. Do you have an explanation for this innitial dive in accrual? If so please add it.

Author Response

Please see the attachment. Thanks. 

Author Response File: Author Response.docx

Reviewer 4 Report

Review of Manuscript – Current Oncology

Advancing Academic Cancer Clinical Trials Recruitment in Canada

 

This is an important manuscript detailing the progress of the 3CTN in advancing academic clinical trials recruitment following a decline in this activity for a variety of reasons, such as complexity of ethical and regulatory processes, costs and limited support.  The manuscript is primarily (and appropriately) descriptive, although they did attempt to identify factors associated with success.

 

Some of the background material is light on detail.  For example the goals and objectives as listed are reasonably clear for the first three, since the improvement is measureable, but how do they plan to demonstrate the impact of the Network and academic trials on the Canadian Health Care system?  This objective is very broad and non-specific, and would benefit from some information as to how they plan to do that.  I don’t believe it is addressed in the manuscript beyond that first mention, so another option is to remove it from the list of goals.

 

Similarly in the following paragraph, the NCCs were “encouraged to share a portion” but it isn’t clear how, or if, that was done.  Was this measured?  If sharing is optional, it’s been my experience that it often doesn’t happen when it comes to funding.

 

For Figure 1, the axis on the left and on the right should be labelled for ease of interpretation.

 

I am concerned about the potentially misleading 4-month period (October 2014-March 2015) which is at least visually compared to the one-year blocks that follow.  I think it is artificially low because it is only 4 months, and because of that, it looks like there was a substantial drop after the Pre-3CTN period.

 

For Table 4, there is either a typo in the table or in the narrative but it indicates that number of baseline trials was not statistically significant, yet in the table it has a p=0.005.  I suspect that the table is incorrect, and should be p=0.05.

 

This is a minor point but Figure 2 needs to have better discriminatory ability between the lines.  In black and white, it is not easy to tell the West and Ontario lines apart (at least not in the version that I have).

 

I could not find it expressly stated, but I believe that the number of sites is 49?  My concern about the multivariable model is that there are really too many predictors for the number of outcomes.  It states that 25 (51%) met the year 4 recruitment target, so that means (using the rule of thumb of 10 events per predictor), they should really only enter 2 or 3 variables, maximum.  Because this is exploratory work, that rule can be relaxed a bit but 7 is a lot to enter for a sample size of 49, especially when some of the variables have multiple levels. 

 

The other concern is that of collinearity. In table 4, site type and site size are highly significant in the simple models, but not even close in the multivariable model, which is a classic sign that something is going on behind the scenes.  I suspect that some of these variables are highly significantly associated with each other – in particular, core funding, site type and, size.  In this case, it is important to identify the variable(s) that are of the most interest, as it is never a good idea to put highly associated variables in the same model.  They may functionally cancel each other out, which appears to be what happened here.  Testing for collinearity will also allow you to reduce the number of covariates going into the models.

 

For Table 5, the authors state that since no factors were significant in the univariate models, no multivariable modelling was done.  However, it’s been my experience that when you control for other factors, a borderline variable may actually become statistically significant, so it might be worth a try.  There are two variables that are very close to statistical significance.  If you do try it, the same comment above about collinearity applies here.

 

It is also important to address these two borderline findings – as the number of baseline trials, and number of baseline trials staff increase, there is a drop in the likelihood of meeting year 4 accrual targets – is this simply a celling effect?  If the numbers are already high, they are less likely to increase?  This should be addressed in the discussion.  With a sample size of 49, these “borderline” variables suggest that you may be underpowered, rather than that these factors are not relevant.

 

Author Response

Please see the attachment. Thanks.

Author Response File: Author Response.docx