Home
This site is intended for healthcare professionals
Advertisement

Research Skills Medical Student Journal Club - November (Cancer Clinical Trials)

Share
Advertisement
Advertisement
 
 
 

Summary

This on-demand teaching session provides an in-depth look at the principles and applications of clinical trials relevant to medical professionals, particularly those dealing with cancer research. Topics include the significance of randomization, the implications of single and double-blinding techniques, the potential sources of bias, and various outcome measures typically used in cancer clinical trials. The speaker also highlights the 'intention to treat principle' and outlines the limitations of clinical trials, arguing for their necessity despite the high costs and time-consuming nature. This session aims to equip professionals with the knowledge to critically analyze clinical trial reports, illustrated through a case study on a phase three clinical trial exploring treatments for triple negative breast cancer.

Generated by MedBot

Description

Looking to sharpen your critical appraisal skills in a fun and supportive space? Our monthly journal club is the perfect place for you - no research experience required.

Theme of the month: Clinical trials in oncology

Paper this month: Combining pembrolizumab and chemotherapy vs placebo and chemotherapy for first time treatment of advanced triple-negative breast cancer: KEYNOTE-355 trial results

Session structure:

  • 20 minute presentation of a clinical trial paper by David Withey, 4th year medical student with a PhD in Cancer Sciences
  • Group discussion of the paper
  • Introduction to critical appraisal and randomised controlled trials

Learning objectives

  1. Understand the principles of randomization in clinical trials, including the purposes, methods, and significance in minimizing bias and ensuring balance of covariate influences.

  2. Explain the concept of blinding in clinical trials, differentiate between single and double-blinding, and discuss the potential impact on validity and bias in the trials.

  3. Recognize and interpret key outcome measures used in cancer clinical trials, such as overall response rate, disease control rate, and survival outcomes.

  4. Understand the concept of the intention-to-treat principle and its implications for trial results and real-world clinical practice.

  5. Discuss the main limitations of clinical trials, including the costs, time commitment, effect on external validity and long-term uncertainty.

Generated by MedBot

Similar communities

View all

Similar events and on demand videos

Advertisement
 
 
 
                
                

Computer generated transcript

Warning!
The following transcript was generated automatically from the content and has not been checked or corrected manually.

Treatment and control groups. Um So, yeah, this allows researchers to control the balan uh control and balance the influence of certain covariate such as age sex or disease severity. Um participants are placed in predefined subgroups and then randomized accordingly as I was saying. And then um it's used, uh it's also used widely in uh cancer research um which often has quite a highly heterogeneous um participants anyway. Um So that's uh those are the main principles of randomization uh that you should understand. Um Yeah, another key influence of uh clinical trials is the aspect of binding blinding. So, um this is important for enhancing the internal validity of the clinical trial. You have two options. Um You have single blinding where one party, normally the um the research themselves where one party is blinded to their clinical trial. This is normally the patient who's blinded. Um And then you have double blinding where b uh both parties um are blinded. So the participant uh doesn't know what um treatment they're receiving and the doctor or the researcher doesn't know what treatment they're giving. Um So, in terms of uh single blinding, um the, the main issue with this one is that um there's a huge potential for detection bias. So, if the, er, researcher or the doctor knows he's giving, um let's say the experimental treatment um subconsciously or even consciously might um might end up detecting, um it s signs that um may not be significant if the patient was in the placebo group, but he may think the signs are significant if he's in the experimental group. Um for example, um uh uh For example, if you've got a treatment for uh if you've got a new antidepressant that you're um that you're trying to test, then it uh then he may um end up um uh try. Uh he, he may end up thinking that uh the patient's doing better when. Uh but that's just because he knows he's giving the treatment. Um They, these are normally used when the double blinding isn't possible. So you don't commonly see it within um cancer um research anyway, it's normally for things like surgical intervention, uh psychological experiments where it's actually quite impossible to um uh to give a placebo. Um And the key thing to me uh to mention here is that any outcomes measured in the single blinding should just be objective. Um So it should be, you know, um examples of how much a a tumor has shrunk for example, or um or key um or, or just things you can actively measure qualitative data um really doesn't work well with uh single blinding um in terms of double blinding. Um like I was saying, it reduces the risk of detection bias because the uh the doctor um doesn't know what treatment he's giving. So he's got no impetus either way to um uh to uh subconsciously um think that some signs are more significant than others. Um It tends to be a lot easier in drug trials and um it is more costly and more difficult to implement, which is like I was saying why single blinding is uh sometimes used. Um Just a quick note to say on um some of the uh outcome measures uh that you'll often see in clinical trials. I'm focusing here on cancer. So uh other clinical trials might be um uh might have different um uh measures, but I've just got a list um here of things that um are commonly uh of definitions that are commonly seen within cancer clinical trials. So I split them into uh five different um categories um that uh the ones that are looking at efficacy, one's looking at survival, one's looking at disease control, uh those that are looking at uh toxicity or safety outcomes and those which um assess quality of life. So in terms of efficacy outcomes, you have things like overall response rate where the percentage of patients who achieve a partial or complete response to a treatment uh that's very commonly used. Uh Complete response is the disappearance of all signs of cancer in response to treatment. Uh Partial response is a reduction in the size of the tumor or extent of cancer but not uh the total disappearance. Uh Then some definitions, you've got stable disease, a situation where the cancer neither shrinks nor grows significantly. And then you have a progressive disease where the tumor growth um uh whether it's tumor growth or spread. Uh despite treatment, um some of the survival outcomes that are looked at is things like overall survival, which is the length of time uh from randomization or treatment start until uh death from any cause uh progression free survival. This is um the time uh from treatment starts until disease progression or death. Uh whichever one comes first uh time to progression. That's uh quite um that's normally used quite a lot. Um which is the time from the start of treatment until disease progression, uh excluding death. And then you've got disease free survival, which is the average length of time from the start of treatment that the patient remains uh cancer free. Um In terms of disease outcomes, you have things like disease control rate, which is the percentage of patients who achieve a partial or complete response to patients. Uh time to progression, which I mentioned earlier, apologies, uh duration of response, uh the time between the initial response and disease progression or relapse after the treatment. Uh patient reported outcomes which are data collected directly from the patients about their symptoms function and wellbeing. Uh, these are commonly used, uh, for quality of life, um, for the patients and there's normally a questionnaire that, um, they'll, uh, that they'll fill in, uh, toxicity. Uh, this is, um, very common, uh, to assess, uh, the, so these can be, uh, incidents of, um, adverse, uh, events. So the frequency of any side effects or complications observed during the trial, um, then you've got the grade of adverse events. So, w which assesses the variety of adverse events. These can be categorized into grades uh commonly, uh these are from 1 to 51 being mild, five being fatal. And then uh special attention is always drawn to the serious adverse events which are life threatening or disabling side effects or events that uh require hospitalization. Um Another key point um about clinical trials is the intention to treat principle, uh which is a strategy where all of the patients are analyzed in a group uh into the group to which they were originally assigned, regardless of whether they completed the treatment adhered to the protocol or withdrew from the study. So the um what they commonly say is once treated, always, always analyzed. Um The, the reason why people do this is it preserves the original randomization that the patients originally were given. Um It also reflects real world conditions because um um in the real world, um patients are going to miss doses, they're going to discontinue treatment, they're going to uh there's gonna be so many things in life that are going to get in the way that a lot of the time, um which is the main cri one of the main criticisms of a clinical trial because it's, it's just there to assess the medication when it's been given perfectly. It doesn't necessarily represent real life in a lot of cases. Um And it also gives a conservative estimate of the treatment effect. So, if the clinical trial is um still successful, when a number of the patients have dropped out of the study, um uh missed doses and it's still producing an effect, then you can be more confident in that effect. Um uh that it's um that is going to happen in real life rather than in a trial which everyone is taking the medication um as uh as prescribed. So it, in certain respects, it can sometimes be useful when patients drop out even though it gives researchers a, a lot of headache when it happens. Um, just some uh key limitations uh for clinical trials that you guys might want to be aware of is, as I was mentioned, there's obviously a high financial cost to these. Um, they're time consuming, um, time consuming as well, even though those two are needed and it gives you a lot of evidence. It can, um, um, it can be, it's often the reason why um there'll be an amazing drug looked at in, er, labs for example, uh with very promising results and then it's, you know, 10 years later that you finally see it um being given to patients. Um they're also weaker at confirming external validity. And uh what I mean by that is um the point that I was mentioning earlier um clinical trials. Um uh most of the um it's it, it doesn't account for people missing doses, it doesn't account for people missing appointments. Um It just assumes that everyone will take the medication perfectly. Um So that's a common criticism. Um Not so much in cancer, but as I was mentioning, there's challenges to do with blinding. Um And clinical trials tend to focus on the short and medium term data. So um uh a medication might be safe um within five years but uh scientists uh simply don't know um what the effect that the medication is gonna be like in 30 years time, you need real world data for that. So it's, it's unavoidable, but it's uh but it's a limitation and um there's also as we'll discuss a bit later, limited outcome measures. So most clinical trials will end up focusing on one or two key um um key measures like overall survival, for example, or uh or uh disease progression, they may not um be focused on patient quality of life um or patient or um um or how patients, yeah, how patients live with the medication. So, um again, it's real world data that you know those that, that conglomerate of different data normally comes together then which can uh can be a bit of a, um which can be a, a certain limitation for the trials. Um So, yeah, I've, I've talked to you through um some of the key aspects of uh clinical trials. Uh There's uh there's uh more stuff that you might want to read up online, but I think those are the main basics that should be covered. Um So now we're just gonna focus on um a critical analysis of a real paper where um uh where p uh pembrolizumab uh plus chemotherapy versus uh versus placebo plus chemotherapy for previously untreated locally recurrent inoperable or metastatic triple negative breast cancer. Um A randomized placebo controlled double blind phase three clinical trial. Um Now I'm very aware that this title sounds pretty horrific. But if you take a step back and actually look at um what the title is saying, we can already pick up a lot of this stuff already. So, uh we know what randomization is the patients are allocated to the treatment groups in a random manner, uh Placebo controlled. So, um um we, we know that the placebo plus the chemotherapy is the control and the pembrolizumab plus the chemotherapy is the experiment doesn't matter yet what the medications are. We just know what groups that the patients are being um assessed in double blind. So we know that both the researchers and the um and the clinicians are being blinded to the treatment. And then the phase three clinical trial is mainly just looking at, um, it's not so much looking at the safety of the medication anymore. Even though that is assessed, it's looking at how well the medication is doing at treating the cancer itself. So, first of all, I'm just going to, uh, give you guys a quick, uh, introduction, uh to, um, the, the basics of the paper before we begin it. So we're not lost before we start. Um So with 2.3 million cases being diag um diagnosed each year from 670,000 deaths, uh breast cancer is remains um the most common cancer among women worldwide and it's a leading cause of female mortality. Uh One very interest. Uh um 90% of these deaths uh won't occur from the primary tumor itself. It will occur through uh metastatic spread. So, the most common areas for breast cancer to metastasize to uh the brain, the lung, the liver and the bone. Um and then each of these will have their own media, uh their own uh median survival rate with the brain uh often having um the lowest median overall survival rate. Um when breast cancer uh colonizes a different organ, um it will uh I'm focusing on the brain here just because that's where my research was. There can be any organ. Um The uh there'll be consistent interaction between the tumor cells and the microenvironment. So, within the brain, for example, you'll have your microglia, your neurons, your astro astrocytes, and then you'll also have your lymphocytes, which is what we're focusing on today. Um So initially, uh the body will um uh develop an immune response against the cancers uh through these lymphocytes. Um where um what you'll commonly see is that the uh T cell receptor um will bind to the cancer antigen. Um and then it will uh initiate an immune response uh through um infiltration of things uh uh for signaling for macrophages, for example, which will phagocytose the cancer cells. One problem with this is that the tumor cells um very commonly start uh overexpressing uh the um A A ligand called PDL one. And um you guys might be very aware of this but um when the PD L1 from the tumor cell uh binds to the PD one receptor from the T cell, uh this actually inhibits the activation um of the T cell. So, Pembrolizumab which is a APD one inhibitor. Um It's uh it's an antibody which will bind to the PD one and it will stop this binding between P one and PDL one which um stops the um uh which allows the T cells to be activated again, allowing the T cells to launch an immune response against the tumor cells which will um cause um tumor death. This is uh uh a cancer cell death. So, this is now being heavily explored in solid tumors. Um It's um showing um some great results uh which is why um you'll see a lot of papers now focusing on different um different immunotherapies, Pember being one of them. So um the way that I'd recommend for all of you to analyze a to critically assess a randomized clinical controlled trial is the CSP checklist. So it's basically 11 questions designed by the Critical Skills appraisal program, which is designed to test the validity of the study in question. So it's useful to use for a structure for a structure and th approach when you're going through these papers. Um Remember though that this is used as a metric for the study, reliability and applicability. It doesn't give an indication of how important the research is. Um or how much impact it's having you, you could have the most, the most reliable paper in the world. Um But if it's not showing, you know, any real data or any real effect, then this isn't going to be picked up by the CSP checklist. It's just for the reliability of the paper. Um I've got uh you can find the checklist anywhere online but um I've got a link here um with the uh site I was using for it. So um I don't know how you want. Uh has, has everyone got the paper in front of them or um because I'm thinking it might be a good idea if um if people could, um, try and go, uh go through these questions, um, as we're doing it. So it's more of a discussion rather than a lecture. Um, but I'm happy to give you um these answers. Now. It, it, it's, it's really up to you guys to be honest. OK. So I've not getting a, um, not getting any reply. So, um, what, what I'm gonna do for, um, for here, then I'm just going to go through what the answers would be for this uh this paper. But um when I'm going through it, if uh you guys would uh want to try and find these answers yourself and then we can discuss each one as they're coming along. Um Please, um just, just let me know. Um Yeah, I think um yeah, I got um linked to the study there in the chat if um if you guys want to have a look at it. So, um the first question is, um does the study address a clearly um formulated research question? So, um in order to answer this, you just need to look at the um the population being studied, the intervention being given the comparator chosen and the outcomes um that are um that are measured. So, um in terms of, um and there's something, oh, so I guess is it um there's an issue um with my slide I've added, um I've added some um um some transitions but obviously because I um put them into the PDF. Um, they're not going, so I'm, I'm just going to go back to my slides and take the transitions off and then, um, reupload them. We should be ok for, um, for time. Uh, it'll give you all a chance to, um, get the paper up if, um, um, if you haven't got it there already. Um, my apologies for that. Sure. Mhm. Mhm. Yes. Yeah. Mhm. Yeah. Mhm. Oh, questions? Ok. Ok. The um, slides are just, um, uploading now. So again it might take a, um, mm, might take a minute or two. But, um, so far is everyone happy with the, um, uh, with what we've discussed so far with the clinical trial? Does anyone have any questions about, um, um, about what, uh, what's going on? Oh, so why we do interim analyses? So basically, um, the clinical trials will, um, occur over um, number of years. So I think this one's been, um, uh, this trial now has finished, um, in 2024 and it started in 2018, but there's a lot of data um, that you can get, um, that you can gain, um, after a couple of years, um, particularly, um, with respect to things like, um, and uh, a tumor response, for example, or the safety of, um, the medications. So it's often quite possible to, um, uh, to do, um, uh, to analyze for certain parameters, uh, a couple of years into the clinical trial before it's reached its conclusion. Um If that makes sense and it also um uh also allows um for people to um uh the interim analyses sometimes are when uh clinical trials will be stopped. For example, if they're having, um if they look like they're having no effect um on the uh on the patient themselves uh on the tumor itself, um or if it's having an unacceptable safety profile as well. So there's also a lot of, um, ethics behind it as well in terms of, um, yeah, when to keep a clinical trial growing glo uh uh going or just, um, stop flogging a dead horse, basically almost uploaded. Ah, here we go. So, yeah, the idea was, was I was going to, um, hide the answers uh from all of you and then, um, you'd, uh, guess, um, not guess you'd, um, try and find the answers as we were going. But, um, unfortunately, um, we're gonna have to, um, uh, look at, um, I'm just gonna have to tell you about, um, what we found with the paper. So, um, the first question to ask clearly is, does the study address a clearly formulated research questions? So this will be things like the population studied, uh the intervention given uh the comparative chosen and the outcomes um that are measured. So, um, a lot of the time guys, you can get this from the title. Um So if we look at the title, um of the paper that we've got here. A lot of these questions are already um uh uh are already answered. But um if you want to have a more comprehensive view from the introduction, you can see that they clearly state this. We aim to compare the efficacy and safety of pembrolizumab plus chemotherapy with placebo plus chemotherapy in patients with previously untreated locally recurrent inoperable or metastatic uh triple negative breast cancer. So in that one sentence, they cover um they cover everything there. Um Now, um the the next ones uh has a bit more meat to it but was the assignment of participants to interventions um randomized. So for this, you can have to consider things like um how was the randomization carried out? Was the method appropriate er was randomization sufficient to eliminate systematic bias? And was the allocation sequence concealed from investigators and participants? Ie was it double blinded? Um So we can see here that all of the patients were assigned in a 2 to 1 ratio to receive the pembrolizumab plus the chemotherapy or the placebo plus chemotherapy by means of block randomization, which we discussed earlier, it is often quite common to see as well that um in the later clinical trials, when there's um less um concerns to do with safety and um the and more patients might actually be really um would benefit from the uh the new medication. In in this case, Pembrolizumab for that to be randomized in a 2 to 1 ratio. Um So um the 22 patients will receive the treatment for everyone receiving the placebo. Um So don't be surprised if you see that a lot in uh particularly within the cancer uh clinical trials, uh they mentioned about the um stratification factors. Um So we'll see um in a second that they do um put the patient, they do stratify the patients. Um according to their um uh according to their uh express how much the tumor is expressed in PD one PD L1, which is known to affect the um the efficacy of Pember lozap treatment uh that, that's previously been established. And um the last one is them just talking about how they have um um how they have um um um how they have double blinded for the clinician and the um uh and the um uh and the patient as well, by the way, guys, these are all direct quotes from the um from the paper. So, uh if you do look, there will be um word for word within the paper um are the results of the trial valid. Uh So all the participants who went to the study accounted for at its conclusion. Um So, um we, we can see this in figure one, which I believe is coming up next. Yeah. So um a lot of the time um you can see this within the trial profile that all um of the um clinical trials should have. And um in terms of um the uh and they also mentioned as well, the efficacy was assessed in the intention to treat population, which included all patients randomly assigned to part two. This experiment was done in two parts, but there's no need to. Uh the first part was more about safety. The second part was more about the um efficacy, but we don't need to worry too much about the first part. Um So as you can see here, it's really small. So I do apologize but um we can see how they've randomly allocated the patients. So uh what's quite common in um im um immunotherapy in cancer, they've um randomly allocated the patients uh first. So they've uh signed 566 to the pembrolizumab chemotherapy and then 281 to the placebo chemotherapy. They've told us how they've stratified the patients uh in terms of their PD L1 expression. So they've got two different groups of this. Uh one where um the score is less than 10 and one where the score um one where the score is greater than 10. Sorry, it's very difficult to read and one where the score is just greater than one. And as we can see, um there's 566 in the intention to treat population and then uh 566 are included in the intention to treat analysis and it's the same for the control groups. They've accounted for all of the patients that are in this study, uh which is a big plus. Um The section B is a selection of questions that ba that ask was the um study methodologically sound. So, were the participants blind to the event um to the intervention they were giving? Um were the investigators blind to the intervention? Um they uh they were giving to the participants or was it double blinded? Were the people um assessing the, um, or analyzing the outcomes blinded? Um, so these are just things to consider here for it. Um, as well. Um, so as we can, what they've mentioned in the paper, uh, everyone was um, uh, was masked. Er, the only unmasked population were the pharmacists who provided the mask, uh study site staff with the ready to use identity packaged premal loz saline infusion solutions for the administration at the schedule infusion visits. So, everyone who was, who was seeing the patient or analyzed and the results was blinded, it's unavoidable that you'd have to get a pharmacist um, or someone else to um correctly give the medication but everyone else who was involved in the study was, um, was analyzed and the pharmacists weren't giving the medication, they were simply supplying them to the doctors without telling them what the medication was. Um, the next question is, were the groups similar at the start of the randomized controlled trial? So, were the baseline characteristics of each study group such as the age sex, socioeconomic group clearly set out and were there any differences between the study groups that could affect the outcomes? Um So I don't think I've got the table here, but um they, you should always have a baseline characteristic table in every clinical trial to look at. So you can see for yourself. Um But uh and you can um see it on the paper if you've got it in front of you. Um But what they mentioned here is that the baseline characteristic uh of the patient were as expected and similar between the two treatment groups. Um So that's uh that's a massive positive as well. Um particularly within um breast cancer guys. So um there's certain populations um I think it's um uh ii could be wrong but I think um black um I think uh black populations tend to fare worse with some of the uh chemotherapies in particular. So it's important that um you have a good, you have a good balance um particularly with age ethnicity. Um uh in order to ensure that the treatment um is fair. There's also a huge discrepancy in the treatment response from postmenopausal and premenopausal women um and age as well independent of whether they've been through the menopause. There's all things I'm sure it's the same in other cancers, other diseases, but you have to look at the patient characteristics um as well. So that's very important to look at. Um now was the uh apart from the experimental intervention, did each study group receive the same level of care that is, were they treated differently? Um So here you've got to consider, was there a clearly defined study protocol? Um if any additional interventions were given, were they similar between the study groups? And were the follow up intervals? Um the same for study group? Um So if you look in the methods, you can see that there's no deviation um from um uh from the treatment group or from the placebo group. And you can see um as well that the trial was done in accordance with the standards of good, good clinical practice and the declaration of Helsinki as well, which is all positive. Um Just to say as well guys, before we go on to the results, um you shouldn't be struggling um uh unduly to find a lot of this information. So a go cinical trial will uh will state all of these things and it shouldn't be difficult to find if it's difficult to find any of these questions. And you, you find yourself scouring through the paper and you're just not able to find it whatsoever. Then that's, that, that's a red flag in and of itself, all of this information should be uh should be given in a clear and presentable manner. So that's something to keep in mind. Uh This is probably gonna be the biggest section. So, um this is just talking about the results. So we're just gonna consider things like power calculations. Uh What the outcome, what outcomes were measured? How were the results expressed? Uh Were the results reported for each outcome in each study group at the follow up interval? Uh Was there any missing or I or incomplete data? Uh Was there differential dropout between the study groups that could affect the results? Uh All potential sources of bias identified and which stats statistical tests we use? And were P values reported again, really important things that you should always be asking. Um So in terms of the power calculations, uh just a quick run through with these ones, um these are made um before the study to calculate basically how many patients are needed in the intervention group and how many people are needed in the controlled group. Um So uh pronounced effect for. So if you have a study, which in the uh which in the earlier stages was showing a very pronounced effect from the intervention, then you'd need a smaller sample size because um you just need less people to prove your point basically. However, if there's um a smaller effect from the intervention given at the start, uh then larger sample, then a larger sample size will be required, potentially needing 100s, maybe thousands of people to actually confirm your data, uh uh confirm your results. I mean, um you should always uh see a power calculation reported. For any clinical trial you're studying. If you don't see it, then it's um that's, that's a huge red flag. Um A power of 80 to 90% is standard. Uh Keep in mind as that, this is a completely arbitrary number. So if you see one which is, I don't know, 7677 I it's chances are, it's fine. Um And if too many people uh drop out and not enough recruited, um then the study can be underpowered. Um And it can make it much more difficult um to make a conclusion from. So from the paper, we can see that the trial has an overall 86% power for the analysis of progression free survival in patients with CPS of 10 or more. Um The full statistical analysis plan is in the protocol. Uh One criticism I have of this paper is that I uh is that they mention the 86% power um for one of the stratified PD one groups. So these are the patients with the highest number of PD L1. However, they don't mention the uh they don't mention in the paper itself about the patients with a lower PDL one score. Um I'm um I'm sure they'll have it um uh in their own uh statistical analysis plan, but in my opinion, that information should be given um um in the paper itself, not in any supplementary figures or anything. So that's, that's one criticism of of the paper, um the subgroup analysis. Um And, and they do mention that in one of um when they stratified for chemotherapy, which we haven't shower, which I haven't shown here, but there was um some underpower of the result. Um So, which is probably why they didn't show it in any of their uh their figures as well. So it's just to keep in mind that if an ex, if one party is underpowered, then you'll start to think how much can I trust the um how much conclusions can I draw from the data? Um in terms of the outcome and uh outcomes measured, um they uh do a good job. Uh They mentioned that there's two primary outcomes. Um So you've got progression free survival and overall survival in patients uh with a high um PD L1 score or um or medium PDL one score uh um or a lower PD L1 score, I mean, and they also mentioned that they've got some secondary endpoints as well, such as uh objective response rate, duration of response and disease control rate. Um So you often have um secondary characteristics that the um that the uh that the researchers will be looking at as well. However, the primary analysis is the main um uh is the main. Um so uh is the main focus of the study um in the results, we can see that they're presented as Kaplan Mayer plots. Um So it's just looking at the percentage of patients um who are um who, who survive over time. So we can see that in the patients with the highest PD L1 score. Um So where uh the treatment is likely to have the most effect because there's more, just pembrolizumab inhibits PDL one, there's more PDL one for the um of the pembrolizumab to inhibit. Uh So we could see here that the patients on pembrolizumab survived 9.7 months and the patients on the placebo uh survive for 5.6. Uh when we look at the uh combined positive score of greater than one. So the low PDL one score, we can see a bit less of an impact. So again, 5.6 months for the control and then 7.6 months of survival for the um uh for the patients with the uh pembeum. And then when we look at the intention to treat population, so all of the populations combined together, um we can see a slightly reduced survival where you have 7.5 months for those treated with the pembrolizumab about 5.6 months for the um uh for the um uh for the control group. So we can see that even in the intention to treat population with people who may not have been taking the medication as regularly as they should have or may have dropped out early, um, w we can, um, we can still see that there's a good um uh there's still a good, well, a couple of month survival benefit from the pembrolizumab treatment um in terms of the other results. Um So they look at, um, basically the hazard ratio, um, w when they're looking at other, um, when they're looking at other characteristics or when you can split, when you split for things like age geographical region, um uh what chemotherapy they're on, uh the disease free interval. I won't go through all this now because there's not, there's not enough time. But if you want to read about it in uh your own time, then this is a uh an example of um how the treatment has worked on uh patients with different characteristics as well after they've gathered all of the data. Um and then they also look at the adverse effects. So you can see here that um the um that a lot of the patients um uh experience things like um anemia, neutropenia, um nausea. But um what's quite promising with this paper is if you compare the uh percentage of people um with the pembrolizumab to the placebo chemotherapy group that most of the adverse events um are uh have shared uh occur at a similar level of incidents between the two groups indicating that's probably the chemotherapy that are giving the patient the most amount of bother, not the pembrolizumab. We, we'll go into this in a bit more detail in a second. And was there a differential dropout which could affect the results. Uh, we've discussed, we've discussed this now, a lot of the patients, um, have dropped out of, uh, this medica, uh, uh, of this clinical control trial. I think it's over 90% but it's very common to see this in cancer. Uh, after all these are metastatic patients, a lot of them are gonna be very ill. Um, they're going to, um, either have their disease progress. Um, a lot of them are going to, these medications were, were planned to be given over a couple of years. So a huge amount of them are going to um pass away in that time. Um And then also, uh they might um the clinicians as well at certain points, um uh will just change their medication, for example, change their chemotherapy. So you've no longer got um a good control as well. So they'll have to drop out because of that. Um And then, so if we look at the actual withdrawal of consent or um uh or adverse effects, there's a lot less of those are a lot less of a reason for people to drop out rather than um um things that you normally expect like progression of disease or uh clinical um or, or, or clinical decisions, for example, um any potential sources um for bias. Um So they do mention um uh in here that the results should be interpreted with caution, uh particularly with the subgroups of the chemotherapy. We've mentioned this earlier um as um some of the results are underpowered. But the fact that they mention that um that is a very good sign of the paper because they're aware of the potential pitfalls of their experiment. Um And as long as they mention it, that's um that's uh that's fine. And then it's up to us then to think about how um what uh how much conclusions we can draw from the paper. Um in terms of uh statistical tests as with every clinical trial, there'll be millions of them. But uh they mention this uh in their um statistical analysis part. So you can see it's all here, uh looks like very standard, uh looks like very standard tests. So um II can't see any issues there and the paper does mention P values. Uh You can see, um I think I have um posted the wrong um paper on here. So, apologies for that uh uh the, the wrong section here. But um if you go towards the end, if you look at the discussion, they'll mention the P values, they'll mention that their work um is uh significant. Um is I'm assuming everyone's OK with um P values. Basically, if you've got ap value of less than naught point, naught five, then that, then scientists generally consider that as being um uh as that event not occurring due to chance um but occurring due to an effect of the treatment. So the smaller the P value the better. Basically, um we're on to the last couple of them now guys. So uh was the precision of the estimate of the intervention treatment um effect reported. So this is basically referring to the confidence intervals, what they reported. And you can see throughout the paper that they are, I got, I got an example there. I won't uh won't read it all out to you. Um But in terms of confidence intervals, um these are um outcome. Um basically the thing to remember with c clinical trials is that the outcome measurements are based on a sample of the population. So therefore, there's always going to be some level of uncertainty regarding the truthfulness of the results. So how, how do we know that these results hypothetically are just working on this sample of the population? And then when you give them to the 1000 millions of other people that are suffering from cancer, like whether is that going to work? Um the way that you can test for this is with confidence intervals. So these help us assess how reliable the results are. So if they are presented as a 95% confidence interval, then we can be 95% sure that the true effect of the size um of the effect lies within that range. Uh usually the lowest value is presented at the left and the highest value is presented on the right. And these are normally presented as a forest plot. So if we go back to the um results um here, um these are the confidence intervals. Um So basically, when we're looking at the, um whether or not the um Pembrolizumab is working for, um let's say a population of less than 65 years old. Um The So the data point here is the um um is the average of the results and the bar here is looking at the um at the actual confidence interval itself. Um So if that bar um is touching the uh is touching the line a known as line of no effect, then um then we classify that as um as, as not being um a, a true effect. Um If you see what I mean. So, um if we look at the um pembrolizumab um in the PD L1 in a high PDL one score, it's um in B um we can see that it's having an effect on the under 65 population. Um But uh because the confidence intervals are passing the line of no effect, it's not protecting anyone who's greater than uh uh who's older than 65. So it's um that's just how to um analyze uh the confidence intervals. So, um yeah, um apologies, guys are gone 10 minutes over. So, thanks for everyone who's um uh who stayed on. Does anyone have um any uh uh that's the end of my talk? So, um does anyone have any questions uh for me at all. Um I'm happy to answer them. No, nothing's coming through. Um Yeah, if you've guys got any other questions um that you might want to email me. Um uh My email is standard one David dot with the uh student.manchester.ac.uk, feel free to um uh to pop me an email at any time. Um uh If you guys need any um more assistance with looking at the clinical trials papers but um if not, I think um that might be me. So yeah, thanks guys. Um I'll um Yeah. Cheers. Yes.