Nervous System Tests and SEND: what’s going on?

There have been rumors of a proposed Nervous System domain (NV) for many years, yet still nothing has been published and it’s not listed in the content for SEND 4.0. So, just what is going on?

At the recent joint FDA and CDISC SEND meeting, the full scope of SEND 4.0 was presented, and so in my last post I described that content (read it here), and I mentioned my frustration that we still won’t have a dedicated domain for representing CNS data such as Functional Operation Battery (FOB) and Irwin Tests.

Following on from that post, I thought it would be a good idea to discuss some of the checkered history of SEND and the elusive NV domain. It was originally created for inclusion in SEND 3.1, which was published back in 2016. It then came close to inclusion in SENDIG-DART 1.2 but again was pulled late in the process. SEND 4.0 is planned to be published in 2023, yet NV still didn’t make the cut.

At the moment, the intention is for it to be included in a dedicated Safety Pharmacology Implementation Guide (SENDIG-SP), due for publication in 2025. Assuming that all goes to plan, that’s nearly a full decade later than originally expected when it was written.

So, why has it been continually delayed? The main issue cited is disagreement on how to represent the full list of available values. Clearly, in order to make use of the data, the scores and values themselves are not sufficient, we need the context. We need to know what values were available for the assessment. Personally, I’ve always advocated that the Define-XML file should be used for this. Many don’t like using the Define file due to incompatibility with Excel. There were suggestions for inclusion in the nonclinical Study Data Reviewer’s Guide (nSDRG), but the counter arguments were overwhelming.

So, some have driven ahead with a solution to have a dedicated domain in SEND called Scoring Scales (SX), housing lists of all available values for such assessments. NV was removed from SENDIG-DART 1.2 for juvenile studies because it needed the SX domain for the scoring scales. With SX now included in SEND 4.0, why do we still not have the NV domain?

The answer given is that nervous system tests are predominantly used in Safety Pharmacology, and as such belong in their own IG (SENDIG-SP) and not the main SEND IG, which is now supposed to just focus on general toxicology. So, “what about the domains for ECG, cardiovascular and respiratory?”  I hear you ask. Good point. They will still be in SEND 4.0, but without NV.

Having dredged over the history and issues, I come back to the statement in the FDA’s Technical Conformance Guide “Overall, the expectation of SEND datasets for nonclinical studies is linked to the pharmacological and toxicological information required to provide FDA with the data needed to assess and support the safety of the proposed clinical investigations” which then goes on to talk about how SEND is needed to “…assess the risks to human subjects…”.

As the nervous system data can be vital to the safety assessment, shouldn’t we include it in SEND? If we needed to do that, without the NV domain, what other options do we have? Maybe that becomes the topic of a future blog.

I know feelings vary on this topic, so if you’d like to continue this conversation with me, just drop me a note at marc.ellison@instem.com

‘til next time

Marc

SEND 4.0 – Just how big is this change going to be?

It was going to be called SEND 3.2 but now it’s going to be called SEND 4.0. Why the change and does it really matter? As you may have guessed, the change is due to the scope of the updates. The addition of several new domains and data types mean this is seen as a more significant update to SEND than previously imagined. Not only that, but we will see the removal of a couple of domains too.

So, let’s break down the scope of changes for SEND 4.0.

We’ll start with the removal of a couple of domains. Firstly, Bodyweight Gains (BG) is being dropped. With the currently available tools offering such flexibility in the presentation of the bodyweight data in BW, there’s been little need for the BG for quite some time. The FDA’s Technical Conformance Guide was updated a while ago to state this, so it’s only natural that its days in the SEND IG are numbered.

The tumor findings (TF) domain is also going away, but this one is a little more complex. It is disappearing as a separate domain, but the crucial piece of data it holds, that’s the time in days to detection of tumor, is going to move to the Microscopic Findings (MIDETECT will replace TFDETECT). I think we can all agree that this is a simpler and more elegant solution.

So, BG and TF are leaving us, but we are getting 5 new domains:

  • Pharmacokinetic Input (PI) is going to give us a place to present the data that were used as the input for TK analysis.
  • Scoring Scales (SX) will provide a domain to list out all available values for any test that uses a scoring scale. It paves the way for future nervous system data, while in this version, providing useful context for certain lab tests.
  • Cell Phenotyping (CP) and Immunogenicity Specimen Assessments (IS) are going to be introduced.
  • Also, there is a new domain for Ophthalmic Examinations (OE). This OE domain will take ophthalmology data that are currently presented in the Clinical Signs (CL) domain, and instead give them a more appropriate home in their own domain.

After the publication of SEND 4.0, there will be subsequent guides for both safety pharmacology, and dermal and ocular studies. The OE and SX domains are both intended for use in those future implementation guides.

So, there we have an overview of many years of work by a vast number of volunteers, all described in around 500 words. I hardly feel that I’ve done it justice. As we near publication and eventual regulatory requirement, I’ll dig into each of these in more detail and give you my own thoughts and opinions. Well it wouldn’t be a blog without that, would it?

And on that note, I’ll leave you my one controversial, parting thought: having discussed what is ‘in’ SEND 4.0, it still pains me that the Nervous System (NV) domain didn’t make the cut. After all these years, we still will not have a proper home for our Functional Observation Battery (FOB) data. Maybe that particular rant can fill another blog some other time.

‘til next time

Marc

How good is good enough?

There’s an argument I occasionally hear that absolutely infuriates me. It’s usually some variation on the line, “The FDA doesn’t care about the quality of SEND, so why should I?” I know I’m paraphrasing, and probably stating this a little too bluntly, but it’s based on the fact that the FDA’s Technical Rejection Criteria (TRC) sets the quality bar rather low and so there’s an assumption that so long as the TRC is satisfied, that’s ‘good enough’.

As a committed long-term CDISC volunteer developing the SEND standard, I’m highly motivated by the fact that the agency uses SEND datasets to make critical safety decisions about experimental new drugs, and this is the last checkpoint before they are administered to the first human being. Therefore, it is vital that the SEND datasets are a complete and accurate representation of the study data.

The TRC requires little more than the inclusion of a demographics file accompanied by the correct study name and study start date in a Trial Summary file with an appropriate define.xml. Providing those data alone will not allow for the review and analysis required to determine the safety implications for the participants of clinical trials.

This question about passing TRC was again raised in the recent FDA SBIA webcast:

Question: Once the data passes the Technical Rejection Criteria, can it be rejected later during review?

FDA Answer: …If the data comes in and the reviewer finds that it’s not clear enough for them to proceed with reviewing or doing analysis, that could still be rejected by the review division.

(The webcast is here and the quote is taken from 50:30 into the video)

The webcast then goes on to implore the listeners to provide “really good quality data” and to do “…whatever you can do to help reviewers to review the data [and conduct] analysis of the data.” Clearly the agency knows the importance of providing high-quality SEND datasets.

I know that some organizations still view the SEND datasets as an unwelcome burden and expense. I know some would like to keep the cost and time of producing SEND to a minimum. I know that this question about quality and the TRC will continue to occasionally crop up, and I’ll continue to get annoyed and maybe I’ll continue to use this blog as a way of venting my frustration. However, I think that across the industry, there needs to be an acknowledgement that the TRC is just the first automated gate. Passing TRC is not an assurance of the quality or usability of the SEND datasets. That said, at the recent joint CDISC and FDA public webinar it was shocking to learn that in the past year, over 500 studies submitted to the agency failed the TRC (between 15th September 2021 and 15th September 2022). Even such a low bar is a little too high for some.

As you can see, this is a subject I feel very strongly about and so I’d love to hear your feelings on the topic. As usual you can let me know your thoughts at marc.ellison@instem.com

‘til next time

Marc

Virtual Control Groups and more – Update from the CDISC Fall meeting

Last week was the Fall 2022 CDISC meeting. As usual, the highlight of the meeting was the public joint presentation between CDISC and FDA. With each event, cross study analysis has slowly become more of a focus. This time, there were multiple presentations referring to this topic and the opportunity to use SEND beyond just FDA compliance.

The meeting began with a presentation on behalf of Japan Pharmaceutical Manufacturers Association (JPMA). While many interesting points were raised, probably the most thought-provoking topic was the discussion around the use of virtual control groups. While this is an idea that has occasionally been mooted in nonclinical safety assessment circles, it was fascinating to see it being stated in relation to SEND. JPMA stated quite clearly that they understand this is a highly controversial suggestion. According to the presentation, this was partly driven by the 3Rs, but also noted that there is difficulty in obtaining certain types of subjects. They also stated that “In clinical field, there is a movement to replace the placebo control group with historical control and/or real world data”. Carrying this principle into nonclinical will certainly provoke some interesting discussions!

Regarding the development and expansion of the SEND standard, the single biggest piece of news to drop from CDISC was the fact that the next version of the SEND Implementation Guide will be called version 4.0 and not version 3.2 as previously communicated. What is the significance of this? CDISC leadership stated that they wanted to communicate that this release includes quite substantial changes. There will be 5 new domains available:

  • Pharmacokinetic Input (PI)
  • Scoring Scales (SX)
  • Cell Phenotyping (CP)
  • Immunogenicity Specimen Assessments (IS)
  • Ophthalmic Examinations (OE)

This version will also drive a new version of the underlying SDTM model as 18 new variables are being added across the guide.

In addition to SEND 4.0, work continues with both the SENDIG-DART 1.2 for juvenile toxicology and SENDIG-GT 1.0 for genetic toxicology. Both standards have made good progress towards their 2023 publication dates. It’s also worth mentioning that good progress is also being made by the Tumor Combinations working group which is creating, publishing and maintaining recommendations for combining tumors for analysis in nonclinical using SEND Controlled Terminology.

Finally, the event concluded with an excellent presentation from the BioCelerate – FDA CDER SEND Harmonization/Cross Study Analysis Working Group who are looking to use SEND for cross study analysis in relation to the Target, and specifically “Understanding off-target toxicity”. The presentation included several novel visualizations that they have developed specifically for this application of cross study analysis.

While this joint presentation was the highlight of the week, significant progress was being made by all the CDISC SEND workstreams. I think that many involved will agree that it was an exhausting but very productive week.

‘til next time

Marc

Go Together

If you want to go fast, go alone but if you want to go far then go together.Proverb

I don’t like moving slowly. I like to feel that things are moving quickly, and the end is in sight. I can become frustrated when things do not move quick enough. Yet, I continually see the value of the slower pace of going together.

While something that can be applied to many areas of life, it’s particularly true of standards development. I oversaw the development and eventual publication of SEND 3.1.1. Despite being an important yet relatively minor change, it literally took several years to go from idea to eventual publication. Many committees needed to approve, and many key stakeholders needed to be on board. The consultation was extensive, soliciting feedback across both nonclinical and clinical.  Much time was spent debating the possible use of ADaM (clinical analysis datasets) instead of SEND. Though it was slow, there is certain rigor which comes with going together.

SENDIG-DART v1.2 is just completing its 3-month public review. It feels like it’s taken an age to complete. Even getting to public review felt slow and arduous. We are currently scheduled for publication early in 2023 and even then, we’ll be years away from it being required by the Data Standard Catalogue.

Again, though it has felt slow, the truth is that the reviews and scrutiny result in a significantly improved standard. The multiple reviews and reviewers have ultimately been a huge benefit to the implementation guide. CDISC has a more rigorous process now that demands an example ‘proof of concept’ study along with conformance rules. Previously such things would be produced after publication. Now they are required before public review, far in advance of publication. The result is a better product, but at the expense of speed.

The eventual publication of SEND 3.2 will have taken significantly longer than the publication of SEND 3.1, but again with the benefit of having a far wider range of contribution and a more rigorous review process.

So, yes, the standards development process seems slower than ever and seems to require more effort now than it ever has before. But that’s the cost of going together.

‘til next time

Marc

SENDIG-DART is more tricky than regular SEND

I write this as I arrive at the European Teratology Society in Antwerp, Belgium to present a poster relating to the SENDIG-DART. This feels significant for a couple of reasons. Firstly, and personally, this is my first in-person conference or meeting since the COVID lockdown. To be travelling for business really feels like things are getting back to normal again.

Secondly, and more relevant for this blog, is the fact my poster, “Preparation for the regulatory requirement for SEND for developmental and reproductive toxicology studies” is being shown with just months to go before the requirement date of March 15th, 2023 kicks in.

It describes the SENDIG-DART in terms of the scope of studies covered, and how the electronic data are to be standardized for tabulation. Put simply, a very brief introduction for the uninitiated.

However, the main focus of it is to provide advice and feedback from the learnings from the FDA’s Fit-For-Use (FFU) pilot. Instem prepared the SEND datasets for one of the studies that took part in the pilot. The lessons we learnt through that process were invaluable in preparing us ahead of the expected demand for conversions. Though we had our Submit™ software ready well in advance and we trained our team in the new standard, there’s nothing like working with a real study to see just how ready we are. Real data always throws up unexpected complications and challenges.

As well as describing the learnings from Instem’s participation, I also describe the main takeaways from the FFU. My personal summary is simply: SENDIG-DART is more tricky than regular SEND. I would say that though SENDIG-DART is quite focused and specific, it certainly shouldn’t be underestimated. Some of the organizations taking part in the FFU hit a few pitfalls. When the CDISC team took a look at the FFU, it was obvious that the Implementation Guide wasn’t clear enough in certain areas. While I wouldn’t go as far as saying it was incorrect, or even misleading, re-reading it through, we could see where some of the guidance may have been a little too subtle. For SENDIG-DART v1.2, much of the text was addressed in order to make the meaning far clearer. The meaning hasn’t changed, but hopefully now it’s far less likely to be misinterpreted.

I’m sure I’ve mentioned this before, but version 1.2 is currently out for public review, and only has a week or two left, so anyone intending to review it who has not done so yet, would be well advised not to delay.

My personal opinion when it comes to the SENDIG-DART, is that any organizations with studies needing converting, unless they have a high volume of studies, I’d advise that they use a specialist SEND vendor as the complexity and cost of implementing SENDIG-DART simply wouldn’t be worth it for only a small number of studies per year. Those with a higher volume would be well advised to engage with a SEND specialist vendor with experience with SENDIG DART for specialist training and consultancy.

‘til next time

Marc

Reflecting on the last 10 years of SEND

Spoiler Warning: This week’s blog post is even more self-indulgent than usual.

I’m recognizing two very significant anniversaries in my professional career. This month marks 25 years of working in the nonclinical space. September 1997 saw me take a programming job at a preclinical software vendor, fully in the expectation that it would be a temporary position for no more than two years, just to pay the bills until I became a rock star. I’m not being flippant. Through a combination of naivety and arrogance, I genuinely believed I was destined to a be a guitar hero. 25 years later, I’m still here so you can see how well that worked out for me.

In the context of SensibleSEND, the more significant anniversary is marking 10 years since I joined CDISC and was first exposed to SEND. So, it seems a good time to take a moment to see how much has been accomplished.

Thinking back to 10 years ago, while the FDA were making all the right noises about their enthusiasm for SEND, much of our industry never really believed SEND would be mandated for submission. Today, the inclusion of SEND datasets for submitted studies has just become business as usual, having been a requirement since 2016 for NDAs and 2017 for INDs.

Initially just covering general toxicology and carcinogenicity studies, we’ve seen SEND expand to cover cardio and respiratory safety pharmacology studies with the introduction of SEND 3.1.

My own involvement with SEND was triggered by the move to cover Developmental and Reproductive studies. It may have taken 10 years, but we are now only months away from these studies requiring SEND datasets for submission too, and I’m sure you are well aware of the upcoming requirement for SEND for CBER submissions.

So, for me it’s been 10 years. In any given year, we seem to make less progress than I would like. However, looking back over a decade, I see just how far we have come. SEND is now my full-time job. Even 10 years ago, I’m not sure how many of us would have imagined that SEND could be anyone’s full time job. Yet, today within nonclinical, it’s a serious career path. My organization, like many others, has entire departments dedicated to SEND. We have SEND experts, consultants, trainers and other service providers, for whom SEND is their full-time job.

I’ve often quoted the line “SEND has been the biggest change to our industry since the introductions of GLP”. Someone once said this to me, and looking back at my 25-year anniversary, this certainly rings true. So much has changed, so much is still changing and so here’s to the next 10 years.

‘til next time

Marc

New standards on review

The CDISC SEND team have several new standards edging their way towards publication.

The SENDIG-Genetox is the CDISC SEND standard for the representation of in vivo genetic toxicology studies. This has completed an internal CDISC review and should soon be heading to a full public review. For those unaware, this first version of SENDIG-Genetox focuses on in vivo Micronucleus and in vivo Comet assays.

Currently in internal CDISC review, and also on the path towards its public review, is version 1.0 of the SEND Tumor Combinations. While not an Implementation Guide, it has been produced at the request of the FDA.  It was created by a working group of biopharmaceutical experts from the Society of Toxicologic Pathology, FDA, and members of the CDISC SEND team. Its primary goal is to assist pharmacology/toxicology reviewers and biostatisticians in statistical analysis of nonclinical tumor data. It is a spreadsheet to provide a user-friendly, high level hierarchy of tumor types or categories correlating the tumor names from the INHAND publications with those available in the NEOPLASM CDISC Controlled Terminology code list in SEND.

Further along in its publication process we have the SENDIG-DART v1.2. Regular readers will be well aware that I lead the CDISC team that produced this next version of the SEND standard for DART studies. This version was created in direct response to the FDA’s ‘Scope of SEND’ section of their study data Technical Conformance Guide. That addition by the agency let the industry know that juvenile toxicology studies are in scope of the SEND requirements. However, for studies where the data need to be analyzed by the postnatal day (i.e. analyzed by age rather than by dose), there are no examples included in any of the published SEND IGs.

SENDIG-DART v1.2 adds one new domain, that is the DP domain for development milestones. While the domain is new, there are no new variables added to the Implementation Guide (IG). The update shows how the existing concepts introduced in v1.1 can now be applied to juvenile studies.

As well as this addition, some existing sections were updated to add clarification, particularly in light of the results of the FDA’s Fit-For-Use (FFU) pilot. Feedback from the FFU pilot made it apparent that there was confusion around the representation of the study day, the dose day and the reproductive phase day. A new section has been added which describes how these are independent concepts and how each should be represented. I hope that this addition really clears things up for those implementing SENDIG-DART.

SENDIG-DART 1.2 is currently out for public review, so please take the time to review and comment on the document. Once complete with public review, and assuming that no significant issues are found, the document is due for publication in early 2023. For the public review, the document is accompanied by an updated version of the SEND Conformance Rules and a proof of concept study. The CDISC SEND DART team has put a huge amount of work and many volunteer hours into producing this update and its supporting materials. I’m very thankful to them for all the effort and expertise they offer.

‘til next time

Marc

Questioning my own SEND vocabulary

This week I’ve been thinking about some of the verbiage that we use to describe how we work with SEND. I’m worried that sometimes this may be misleading, misused or misunderstood.

An obvious example that springs to my mind is the use of the word ‘dataset’. The correct use of this term in the context of SEND is to describe a single xpt file of data. However, the word is often misused to describe the entire SEND package, which itself could contain 20 or 30 datasets, plus the define.xml and nonclinical study data reviewer’s guide.

Another word which causes me some discomfort in the context of SEND, is ‘converting’ as in ‘converting the study data to SEND’. I must admit that I try to avoid this and instead I try to use the phrase ‘representing the study data in SEND’.  The data are not converted. They are the same data whether they are shown on a PDF table in the appendices of study report, or if they are shown in a SEND xpt file. These are just 2 different representations of the same data. To me the most accurate description would be ‘to render data in SEND’, however this is not a phrase I’ve heard anyone else ever use.

I am fully aware that I’m likely being overly fussy here. While I strive for a level of accuracy in how I describe the way we work with SEND, I understand that language evolves, and I am a strong believer in the idea that language is correct by usage. So, I accept that most people understand and are comfortable with the phrase ‘converting data to SEND’. And so, from time to time I find myself using it too.

This week was one of those times as I’ve largely been focused on discussing the comprehensive range of services we are offering regarding SEND. Like other organizations that offer services around SEND, our mainstay has always been our ability to ‘convert data to SEND’. That term has become a useful shorthand to describe our main service. So, despite my reservations with the term, I’ve found myself using it quite a lot this week with the various industry presentations I have been involved with.

In case you are interested in watching one of those recent presentations, you can check it out here: https://go.instem.com/webcast_send_advantage_services

So, yes, spending a week talking about the services we offer has got me questioning the vocabulary we use to describe the work.

‘till next time

Marc

The headache of clinical signs

How do you create a standardized model for data that drastically vary from lab to lab? Of course, I’m talking about clinical signs. When the clinical signs domain (CL) was first defined in SEND, it was clear to the modellers that every lab collects and reports clinical signs differently. So, the answer was to have a super-flexible domain with very little controlled terminology. No matter how data are collected or reported, they can somehow be fit into this domain. Which means that any study can be represented. However, it also means that it’s near impossible to perform any kind of meaningful cross-study analysis.

Several brave souls banded together to finally address the headache of clinical signs and solve the unsolvable. Trying to impose a level of standardization which would mean that the clinical signs data could be used in cross-study analysis. To attempt impose rules on the unruliest of domains.

And so, there is currently a highly controversial proposal on the table. To take the permissible and often ignored Result Category (CLRESCAT), add controlled terminology, and suddenly elevate its importance. Data can still be collected as they currently are. Symptoms, descriptors and modifiers do not need standardization; we simply assign each result to a category from a controlled list, which can then be used in cross-study analysis. It sounds like a simple and elegant solution to an impossible problem.

However, a robust critique found this to be one of the most controversial proposals I’ve seen put forward to the CDISC SEND team. The concerns are valid. We all know that this domain really needs true standardization of the CLSTRESC variable (that’s the Clinical Signs Standardized Result in Character format for those not yet fully fluent in SEND-ese), however nobody is really prepared to attempt to boil that particular ocean.

It has given voice to the question that some may have thought, but never uttered out loud: What’s the point of cross-study analysis of clinical signs? I mean, the data are so subjective. One person’s mild symptom is another person’s moderate. How much value can we really put on these data?

I really admire the effort and ambition of those trying to add such standardization to clinical signs, yet I also question the value of it.

Such thoughts have then led to the wider discussions to question the value of certain nonclinical data. Food and water data for example. In hushed circles, in a surreptitious discussion, there was the comical scenario described, of group-housed subjects ensuring that the food is not spilled and is equally shared out to give an accurate measure of consumption per subject. We know this is not the case and therefore the data are not accurate, so why do we place such value on the data?

It reminds me of the words a senior scientist told me early in my career: You know, for nonclinical data, the only things that really matter are pathology and clinical pathology data. Everything else is too subjective.

Surely clinical signs are the most subjective and variable of the nonclinical data. Trying to add any truly useful standardization is admirable, but is it an impossible task?

I’d love to hear your thoughts. You can contact me at the usual address marc.ellison@instem.com

‘till next time

Marc