The most important ESMO oral presentation on colorectal cancer treatment in 2014, “First-line targeted therapy for mCRC: anti-VEGF or anti-EGFR?”, has attracted a lot of attention due to the inconsistent and controversial results of the two most important studies, FIRE-3 and CALGB 80405. Because of the importance of the topic, the oral presentation was not followed by the usual question-and-answer session for the audience, but by 1-2 representative questions for each study, which the investigators were asked to answer on the spot.
After the question and answer session, the “conclusion” of this special session is also unique, perhaps because two such important studies have brought about a lot of controversies that cannot be explained so far. Perhaps because of the controversial nature of two such important studies, the conference was arranged for four leading oncologists from Europe to present their insights as a “conclusion”.
Although the author also attended the meeting on site, but considering the language problem and time limitation, the understanding on site is inevitably deviated. Now the full text is presented to the domestic peers for your reference.
The main reason for sending out this issue of the conference broadcast after 2 weeks is to consider the accurate and complete translation of these on-site materials, so if you have any objection to this, please feel free to discuss.
Questions and answers created by the conference.
RAS testing for CALGB 80405: Issues with testing methods and cutoff values
Chair: Professor Lenz, you mentioned that the RAS assay in the 80405 study used a new assay, the BEAMing method, and selected 1% as the new cut-off value. Does it affect survival outcomes? Will this become a new standard for molecular testing?
Prof. Lenz: That’s a good question, from a time saving point of view you can use only a 5% cutoff, why do we use 1%?
It’s because some of our recent results show that the best cut-off value is probably between 1% and 5% and uses the same assay (BEAMing) and the same cut-off period as the FIRE-3 study. Now I can also tell you that from our preliminary results, 1% and 5% did not make a difference, whereas 5% was used in the CRYSTAL and OPUS studies. So, I personally don’t think that this makes any difference in the results. Of course, I don’t think anyone has a standard answer to the question of what the ultimate optimal cutoff value would be yet.
I hope that we will also use the long-term follow-up results to see if different cutoff values end up having different survival outcomes, although at this point we haven’t seen what difference 1 percent and 5 percent makes.
Chairman: Other studies have used different methods, but it seems that the RAS mutation rates are about the same, 15.3% and 15.8%, how do you rate that?
Prof. Lenz: I would say that the cut-off values of 1% and 5% in fact only changed the results in a very small number of patients, not two dozen patients, maybe just a few patients, so I don’t think it makes a difference in the results.
Prof. Venook: In fact in the 80405 study we analyzed the results of the 1% and 5% as cut-off values and we will present them for you later and then you can make your own judgment.
The FIRE-3 study: the question of correlation
Chairman: ORR as the primary study endpoint, why there was no significant advantage in the Cet (cetuximab) group in the original ITT population analysis, but today’s results show that FOLFIRI+Cet has higher tumor remission rate and more early tumor regression, can you make some comments? Secondly, from the results it looks like ETS correlates with DpR and DpR correlates with OS (not very consistent), so the second point we want to discuss today is does the increase in DpR really explain well the prolongation of OS, maybe in other trials the situation is different?
Prof. Stintzing: Regarding the first question, the results I’m presenting today are for the “evaluable population”, which was prospectively defined in the study protocol before the original trial started, and in fact, as we knew from the ORR results published last year, even looking at the KRAS wild-type population, the FOLFIRI results were not significant. group, FOLFIRI+Cet also significantly increased the ORR, and the difference was significant at that time;
Secondly, with regard to tumor response to treatment, I think what we brought to this meeting was much more robust and compelling data, and when we look at those patients who went on to treatment, because of the longer duration of treatment, when another CT evaluation was performed, of course it became possible to see more tumor regression, or at least, we can now see some differences between the two treatments.
The second question, regarding early tumor regression (ETS) and depth of remission (DpR), I think both of these can be used as endpoint surrogates for the study, but further studies are still needed to validate their value, and at this point we can’t have a final conclusion. But the more interesting point is that, as I saw in the poster discussion section today, several studies have focused on this issue, suggesting that the prior RECIST evaluation system may not be sufficient to fully detect those real differences in treatment response that we see in patients.
The question of CALGB 80405 resection rates: How can we explain the reasons for the imbalance in resection rates between the two groups? Could this be the reason for the difference in OS?
Question for CALGB 80405 resection rates: How can you explain the reasons for the imbalance in resection rates between the two groups? Could this be the reason for the difference in OS?
Chairman: Professor Venook, as you showed in the results, the secondary surgical resection rates are not the same, at least the observed values are, and there were more surgical resections in the group treated with Cet, 30 more cases. Is it possible that this is the reason or at least part of the reason for the mild difference in final OS?
Prof. Venook: There is no doubt that there is definitely an imbalance in the resection rate between the two groups and there is a reason for it, but at this point I really don’t know what the exact reason is.
I believe that part of it may be due to selection bias at the time of treatment decision and some kind of patient concern. The doctors involved in the study know that patients treated with Bev (bevacizumab) who are going to undergo surgery have to stop Bev treatment at least 6 weeks before surgery and wait another 6 weeks after surgery to start Bev treatment, and I can say that this is part of the reason, but it certainly does not explain the full phenomenon, for example, the study are not double-blind, you always have to tell the patient the next step of treatment, so as much as I would like to but the fact is it is very difficult to explain clearly the surgical resection imbalance.
I think one of the more interesting things about the difference in outcomes in these two studies is to analyze the depth of tumor remission and early tumor regression, which I believe is actually very similar to what people usually say, which is “good patient outcomes means good work; bad patient outcomes means bad work” (I don’t know if that’s right or not). I believe the survival of patients in the Cet group in the 80405 study is very similar to that of the Cet group in the FIRE-3 study, numerically. It’s very similar, obviously;
I believe the underlying reason for this is that patients treated with Bev in the US and Canada had a better OS, why this is so I am not sure exactly, but I think it suggests that we should pay more attention to this phenomenon than to still dwell on why we don’t see a difference in outcomes, which may help change our clinical treatment paradigm. It may help to change our clinical treatment paradigm to reach a higher level of treatment, although I am not sure, but that is what I am trying to say.
Question for all the experts: How do you explain the difference in results between 80405 and FIRE-3?
Chairman: The last question for all the experts is this key question, do you think there are differences in the results of the CALGB 80405 and FIRE-3 studies? If so, what do you think is the hypothesis for these differences? Are the patient population characteristics different? Are the treatment characteristics different? Is it a difference in 2nd line and follow-up treatment? Or is it something else?
Prof. Lenz: Maybe the other side of the same question, which is let’s look at the similarities in the two studies, you’ll find actually more similarities in the results of the two studies versus differences. Let’s look at ORR, PFS and OS, and overall, the two studies matched well in terms of results except for the longer OS that occurred in the FOLFOX/Bev treatment group in the US study that Professor Venook just mentioned.
I think this raises a very important question, whether this is because of different molecular biological choices, different second-line and follow-up treatments or different patient prognostic characteristics in the patient populations in the two studies, which I don’t think we have a clear answer to today. The main difference between the Bev group in the FIRE-3 trial and the US 80405 trial that we can now know is the difference in the basal chemotherapy regimens, and when we look at the FOLFIRI/Bev subgroup, there are some differences in the results, but these are relatively small subgroups;
And the largest subgroup, FOLFOX/Bev, also had different results from the FOLFIRI/Bev subgroup, which may involve many reasons that are not yet clear, but I hope we can fully think about the following possible scenarios, including the different molecular subtypes, the clinical characteristics of the patients, and the subsequent treatment. I think this is perhaps the more important issue that we can learn from.
Prof. Venook: Let me interject a few words, I don’t really agree with my colleague from Los Angeles, Prof. Lenz. With respect to the differences between the FOLFIRI/Bev and FOLFOX/Bev subgroups, obviously we all expect to be well explained, and I agree with that. I also suspect that there is something in this that is worth exploring. I think we should combine specific information on these subgroup groups to be able to do exploratory analysis more smoothly.
One possible reason for this is that the average tumor load at enrollment may have been relatively low in the US and Canadian patients in our trial (here, the 80405 study), and I don’t know what the natural tumor load was in the German and Austrian patients (here, the FIRE-3 study).
But it is true that there were many patients in the study who were actually in a very low tumor load state when they were withdrawn from the study because of tumor progression, for example, from 2.0 cm to 2.4 cm, and such patients may in fact survive for 2-3 years after disease progression, so the most important message to me is that as long as we treat the patients properly in the post-progression follow-up, they will be able to survive for 2-3 years. So the most important message to me is that the choice of first-line therapy may not be so important as long as we are treating the patient properly in the post-progression follow-up.
Prof. Stintzing: A little bit of comment from our point of view. As Alan (referring to Prof. Venook) said earlier, I think there is still a need for more detailed information about first-line treatment, including the duration of treatment, the intensity of the dose of treatment, and also to compare the level of tumor load at the beginning of the study and at the end of the PFS endpoint event, which is important to really understand why there is such a big difference in survival after progression, after all, the progression-free survival in both studies is very close. After all, the progression-free survival in the two studies was very close.
The “summary” (statement) of the four invited experts.
Prof. A. Cervantes, Spain.
I would like to summarize this in two points. First, why are the OS results of these two important studies so inconsistent? To answer this question I think we still need detailed information about the second-line and follow-up treatment in both studies, especially for one particular group, the proportion of patients randomized to “chemotherapy + Bev” in the study and who did not receive subsequent Cet treatment, which I think could be quite important to interpret these inconsistent results. I think this could be quite important in interpreting these inconsistent results.
If more than 80% of patients received second-line therapy, this might suggest that the order of treatment is less important for overall survival; however, if a disproportionate number of patients did not receive Cet in second-line therapy, then the order of treatment might be quite important. The second point I would like to summarize is, what is the importance of the Cet group having a higher response rate for treatment? Based on the findings brought by our American colleagues, chemotherapy + Bev or chemotherapy + Cet are two effective initial treatment options for patients with all RAS wild-type advanced colorectal cancer.
So, how does the fact that “regimens containing Cet result in higher response rates” affect your treatment decisions when you are choosing initial therapy for your patients? Therefore, I think the information on response rates in the CALGB 80405 study should be further analyzed in terms of depth of remission and early tumor regression, which may be important information when we communicate with patients about their treatment choices.
Prof. A. Sobrero, Italy.
When each new data comes out, we generally interpret it through a process as follows: first, are they real? Second, do they have value? And finally, is it valuable enough to change clinical practice? Sometimes, however, it is not that simple and we may not have any standard process to choose from.
Imagine if this is about studying a new drug in five lines of therapy, you can of course judge whether the data are real and have clinical value, and then decide whether to use the new drug based on that value; but if you are faced with a decision about the layout of each line of therapy when you already have a lot of strong data, the situation is completely different. Therefore, for a topic like “RAS wild-type advanced colorectal cancer targeted therapy – anti-VEGF vs. anti-EGFR”, I will analyze the available new data from the following five aspects.
1. Regarding the use of Bev in first-line therapy, the data are strong and convincing. In terms of absolute median survival improvement, the available data include 4.7 months in the Hurwitz study (referring to AVF2107), 4.8 months in the Fuchs study (referring to BICC-C), 3.9 months in the Kabbinavar study, 3.9 months in the Cunningham study (referring to AVEX) and 1.4 months in the 16966 study, and second-line treatment with 2.1 months in the Giantonio study (referring to E3200) and 1.4 months in the TML study. Therefore, for aspect 1 I conclude that, except for the data from the 16966 study, I think the data for first-line use of Bev are strong and convincing.
2. regarding the third-line application of anti-EGFR therapy, the data are strong and convincing. data from the Karapetis study published in NEJM 5 or 6 years ago (referring to the Australian study published in 2008) showed an absolute median survival benefit of 4.7 months compared to best supportive care.
3. The data on first-line anti-EGFR therapy in the K-RAS wild-type population are relatively less convincing. These data (absolute survival benefit) include 3.5 months in the CRYSTAL study, 4.1 months in the PRIME study, -0.7 months in the COINC study, -0.7 months in the NORDIC study, and some second-line treatment data shown in parentheses, which appear to be inconsistent across studies and suggest that the data for absolute benefit are not as strong;
Also, there is the issue of skin toxicity associated with the use of anti-EGFR therapy in first-line treatment to consider, so my assessment of the strength and credibility of the evidence in this area is not as good as the other two areas mentioned above.
4. Data on first-line anti-EGFR therapy in the RAS all-wild-type population: compared to chemotherapy alone. The data in this area are strong and convincing. Including data from two large studies recently updated, the absolute survival benefit was 5.7 months in the PRIME study and 8.2 months in the CRYSTAL study. The very consistent data from these two large studies is very consistent and has led to EMEA and FDA mandating RAS testing prior to anti-EGFR therapy after reviewing these data, so to me, this is very compelling.
5. Data on first-line anti-EGFR therapy in the RAS all-wild-type population: comparison with Bev. That’s the last point I’m going to make today, is that it’s painful for me to draw conclusions when looking at data on first-line anti-EGFR therapy versus Bev therapy in all-RAS wild-type patients, but if you still believe in clinical study results, then we now have data from two well-established, convincing studies, one is the data update that we just heard from FIRE-3, ( anti-EGFR versus Bev) survival benefit differential improved from 5.7 months to 8.1 months.
The data from a well-established study showing a survival benefit of 8.1 months is very convincing. Finally, the PEAK study also showed a net survival benefit (for anti-EGFR versus Bev) of 12.4 months, but of course, this was only a phase II randomized controlled study with a sample size of 190 cases, which should be taken into account when analyzing the results.
The above five areas of data are the results that have been published so far, so now we have new data: the
New data from ESMO 2014: updated data on the study of FIRE-3, which is easy to interpret because as you can see from the results reported by the authors today, the consistency of the study results themselves has been strengthened, the plausibility of the interpretation of the results has been strengthened, and you can see from the investigator’s report that there was an additional 15% increase in treatment response rate, an additional 15% in early tumor regression increase, so those are good results. So, what about the data from the CALGB 80405 study? We know that this study spans 10 years, and as was highlighted in the discussion session, the results that are being published are very preliminary.
In conclusion, my conclusion is that only when more complete data are available will I come to challenge my already formed views on the above five aspects.
Prof. D. Arnold, Germany (session chair).
I would like to start with “Clinical decision making in first-line treatment of mCRC”, which depends on two aspects: clinical characteristics and treatment goals, and whether we have available biomarkers at this stage. Until 2012, the only “negative efficacy predictor” we had was KRAS, i.e. patients with KRAS mutations should not receive anti-EGFR therapy. However, by 2014, we knew that with full RAS testing, treatment could become more precise and the survival benefit more limited;
So, although it is still the same similar markers, there seems to be some nuance as to whether RAS status can evolve from a “negative efficacy predictive marker” for KRAS to a “positive efficacy predictive marker”, i.e., RAS wild-type tumors Should anti-EGFR monotherapy be preferred? This suggests a new way of thinking about treatment decisions: should we change from “clinical first” to “marker first”, or should the value of biomarkers be even more important in decision-making?
So, the central question now is, can RAS really be a “positive” marker of efficacy?
The PEAK study was a phase II RCT, and the limitations of the study were that there was no hypothesis to be tested and OS was a secondary endpoint, so the study design was not designed with a pre-test of efficacy and its validity was relatively weak, but nevertheless, the findings showed a very strong survival advantage with an HR of 0.63. The results of FIRE-3 as you have seen today, although there was a significant benefit in OS with a HR of 0.7, this study still had problems with OS being a secondary endpoint and the primary endpoint being negative.
The CALGB 80405 study is the only one that presupposes OS superiority as a validation hypothesis and the results show no statistically significant difference in OS (HR=0.9). However, if we come to the HR of the OS endpoint for all three studies, all of them were <1.0, so I'm sure everyone here would agree that there must be a story there, in the different degree of difference in OS that we can see, and perhaps, the FIRE-3 study over-presented this difference, while the CALGB 80405 study under-presented it.
In conclusion, to me, there is some consistency in all these data, so it is slightly better to prioritize anti-EGFR therapy for patients with full wild-type RAS, but at the same time, as discussed, there are still a lot of unanswered questions: that is, why these differences in results between studies? There is still a lack of detailed information on follow-up treatment, such as length of treatment, second-line treatment, etc. We are looking forward to this information;
We’re all looking forward to this information, but what I wonder is, even if we had this information, would it really change our clinical opinion about the results of these studies? Will it change the whole debate? I’m not optimistic about that.
So, first-line treatment decisions for mCRC based on biomarkers are going to become more and more complex. First of all, I think we are still missing information on both clinical features and molecular markers. We need information on clinical features to determine clinical targets, and we need molecular markers, and I think there are two aspects of information on RAS status, for RAS mutant, which is a very strong negative predictor of efficacy, and patients should not receive anti-EGFR therapy;
And for the RAS wild type, perhaps, this is a weak positive predictor of efficacy, meaning that there may be better outcomes, and therefore, these patients may be considered for priority anti-EGFR therapy, but of course, it’s not mandatory that all these patients should receive such therapy.
However, we also need to understand that RAS is by no means the whole range of predictive markers for mCRC efficacy, tumors are very heterogeneous and RAS status can help us to define one particular group of them, but it is not enough, in the future for 80405 study and FIRE-3 study, many more refined studies are needed to reveal the important value of molecular markers, for example, different molecular subtypes of tumors, the For example, different molecular subtypes of tumors, efficacy prediction markers of anti-VEGF therapy, etc.
Finally, as far as I am concerned, the impact on clinical treatment decisions is the same. Previously, we had the ESMO consensus guidelines for colorectal cancer management, in which we elaborated from the perspective of clinical features and from the perspective of marker features. Recently, the ESMO consensus guidelines were further updated to ESMO clinical practice guidelines for the management of metastatic colorectal cancer, published in the August 2014 issue of Annals of Oncology. The guidelines begin by mentioning that extended testing for RAS must be performed prior to anti-EGFR therapy, and I don’t think anyone would object to that.
The guidelines also recommend that for patients with wild-type RAS, two chemotherapy regimens (FOLFOX/FOLFIRI) in combination with either of the two classes of antibodies should be considered as effective treatment options, and there are complex factors to consider when making specific regimen selection decisions. toxicity, and subjective intent. All of these factors need to be balanced. These recommendations of the clinical guidelines represent my own summary.
Prof. F. Ciardielo, Italy (session chair)
Metastatic colorectal cancer is a heterogeneous disease in which a subgroup is highly dependent on the EGFR signaling pathway and these patients are likely to benefit from anti-EGFR monotherapy, and the detection of KRAS and NRA is the first step to identify this group.
Therefore, all patients with metastatic colorectal cancer should undergo expanded testing for RAS prior to the initiation of first-line therapy to ensure that these patients can be offered access to all effective therapies. The available data suggest that for patients with all-RAS wild-type, two excellent treatment strategies available should be used in alternating order in first- and second-line therapy: FOLFOX/FOLFIRI in combination with either anti-EGFR monoclonal antibody or Bev.
For me, I would prioritize FOLFOX/FOLFIRI in combination with anti-EGFR therapy if tumor regression is an obvious element to consider in the treatment goals, i.e., if the tumor load is large, if the tumor causes significant symptoms, if it can potentially be converted to liver metastases or if the primary tumor is surgically resected together. The tolerability of treatment, toxicities and detailed informed communication with patients are all important in first-line treatment decisions.
Of course, more importantly in the field of clinical and translational research, we are still faced with many very important questions that have not yet been addressed, namely, how to optimize the treatment of patients with RAS wild-type mCRC. First, we still need to clarify the value of maintenance therapy and the most appropriate maintenance therapy for these patients, the phenomenon of resistance to chemotherapy and the value of reintroduction. Second, is there an appropriate sequence of treatment for these patients? Should anti-EGFR therapy or anti-VEGF therapy be administered first?
As Professor Cervantes said, if 100% of the patients are treated with subsequent therapy, maybe the order doesn’t matter, but if a lot of patients are shed when you go from first-line to second-line therapy, for example, a lot of patients end up not receiving anti-EGFR therapy at all, then maybe it matters for the final outcome. So I think the most important thing is for each clinician to choose an optimal treatment regimen based on their best understanding in order to try to overcome or prevent the occurrence of acquired resistance to anti-EGFR antibodies or anti-VEGF antibodies in therapy.