7 month survival benefit in lung cancer
It’s not every day you hear news of a survival benefit as big as 7 months in lung cancer. But when you do, it’s not the usual suspects. Yes, a study has shown that an intervention improved the median survival of patients with lung cancer by 7 months, but the intervention is not a checkpoint inhibitor or a targeted therapy. Better yet, the journal isn't a usual suspect either: not NEJM, Lancet, JAMA or JCO. This is a nice RCT that was presented at ASCO 2016 but remained largely ignored. It is now published as a full paper in JNCI. In this phase 3 RCT, after the initial treatment of lung cancer patients with advanced disease, patients were randomized to usual care with routine follow up and CT scans or to an experimental arm. In the experimental arm, patients recorded their self-scored 12 pre-determined symptoms weekly with the data being transmitted to the treating oncologist via a web-mediated algorithm-based e-follow up application. This allowed an individualized schedule for imaging based on patient symptoms. The result? The patients in the experimental arm not only had a significantly longer OS with an extremely attractive hazard ratio (19 v 12 m, HR 0.32, p = 0.002) but the performance status of patients at detection of relapse was also better and more patients received subsequent optimal treatments. If by precision medicine we mean such individualized care instead of genomics-driven unproven care, I am all for it!
It should need no repeating but patients are their own best doctors. They know their body and their symptoms better than anybody else. So, it’s no surprise that their self-recording of symptom scores allowed detection of relapse early and institution of better and early care leading to prolonged survival. Now, the surprising thing is that we have seen a 7-month survival benefit in lung cancer with such a cheap intervention, but this is not making much news. Why aren’t patient groups demanding that this personalized care be instituted immediately across the world? Why is everyone lobbying to get nivolumab and pembrolizumab approved for lung cancer but turning a blind eye to this simple intervention? As I discussed in my last blog, a Center for Sense in Oncology is very much needed.
Notwithstanding the availability of this application, my take-home from this study is: Ask your patients to immediately contact the hospital if they feel something is wrong. Tell your patients that they don’t need to wait for their scheduled follow-up or scans if they sense that something is not normal. Listening to your patient carefully is more important than dosing him/her with an immunotherapy.
The good and the bad from Japan
Prophylactic cranial irradiation (PCI) after response with chemotherapy is the textbook standard of care for small-cell lung cancer (SCLC). While it is undebatable for limited-stage SCLC, a Japanese trial presented at ASCO 2014 cast doubt on its utility in extensive-disease SCLC (ED-SCLC), showing in fact shorter survival with PCI compared to observation alone. Unfortunately though, it took 3 years for the presentation to be published as a journal article. PCI was taken as SOC in ED-SCLC based on an earlier EORTC trial which showed survival benefits with PCI versus observation; however, that trial enrolled patients without an MRI scan at randomization. Hence, many patients might have had brain metastasis already for which cranial irradiation is effective and hence, might have had benefit. This Japanese trial mandated MRI screening at randomization and included only those patients for whom cranial irradiation was truly prophylactic, i.e. patients without brain metastases. For these patients, PCI doesn’t seem to work and instead could be detrimental (11.6 v 13.7 m, HR 1.27, p = 0.094). Although some people argue that these results are for Japanese patients only, I don’t see why that would be different for other countries provided similar MRI scans both at baseline and periodically can be done. This is important because cranial irradiation carries its own set of various short and long term toxicities. Although the guidelines in the US and Europe still recommend PCI, are we sure that the evidence of benefit justifies the risks involved?
This trial represents a good piece of Japanese research. A very pragmatic trial that was well-designed to answer an important question. The funding for this important research came from the Japanese government. Furthermore, the authors are very humble in their conclusion: “we conclude that prophylactic cranial irradiation is not essential for patients with extensive-disease small-cell lung cancer with any response to initial chemotherapy and a confirmed absence of brain metastases by MRI when patients are periodically assessed by MRI examination during follow-up and asymptomatic metastases are treated. However, physicians in countries other than Japan should be cautious when extrapolating the results of this study to general practice in view of different ethnicities and medical situations.”
Now the bad.
ABSOLUTE is another trial that convinced me the absolute necessity for the Center for Sense in Oncology. This study tested nab-paclitaxel (abraxane) versus paclitaxel for gastric cancer randomizing patients to 3 arms: 3 weekly abraxane, weekly abraxane and weekly paclitaxel. Surprisingly, this was a non-inferiority trial rather than a superiority trial and showed that weekly, but not 3 weekly abraxane, was non-inferior to weekly paclitaxel. There are a few problems with this: first, why non-inferior design? Abraxane is not a novel drug. Basically, abraxane is expensive paclitaxel. And weekly abraxane would cost 12 times more than paclitaxel for a month of treatment. Some argue that abraxane has a better safety profile, especially with regards to hypersensitivity reactions, but that alone doesn’t justify 12 times the price. Furthermore, in this trial, the hypersensitivity reactions were 5% in the paclitaxel group versus 1% each in each abraxane group. The trial used 2 abraxane arms and was designed as a non-inferior study against paclitaxel. The upper margin of hazard ratio that was considered acceptable to prove non-inferiority was 1.25. That means the investigators were ready to accept up to 25% higher rates of death in the abraxane arm! Well, why stop at 25%? Let’s consider 50% as acceptable non-inferiority margin and approve every drug.
It is also important to state here that except for hypersensitivity reactions, toxicities were higher and treatment interruptions more frequent in the abraxane cohort compared to paclitaxel cohort. This study was presented as a plenary in the 2016 Annual meeting of the Japanese Society of Medical Oncology and many oncologists appropriately questioned the rationale for this trial. It’s sad to think that more than 700 patients were randomized and millions spent to undertake this huge trial designed from outset to answer a very unimportant question. In case you are wondering, the Japanese government didn’t fund this trial.
Were you joking Mr. Feynman?
A study has shown that for immunotherapies, treatment effect sizes seen for PFS were not usually as high for OS. This study seems to have used robust methodology to arrive at neat results; however, there are a number of other conceptual problems. Although the study title suggests “immunotherapy”, the drugs used in the study include all monoclonal antibodies (mabs) such as bevacizumab and cetuximab. While technically all mabs can be labeled as immunotherapy, almost no oncologist would think of bevacizumab or cetuximab when someone mentions immunotherapy today. Almost invariably, everyone would think of immune checkpoint inhibitors (ICI) such as ipilimumab or nivolumab. Hence, the results such as “only 38% of the variability in the OS effect could be explained by the variability in the PFS effect” or the conclusions that “treatment effect sizes in trials of immunotherapy drugs were greater for PFS than for OS” are misleading. In fact, ICIs are known to cause the opposite effect: prolonging OS without prolonging PFS. Another problem is that of mixing apples and oranges. The study also includes ipilimumab (but not nivolumab or pembrolizumab) among other mabs. Pooling PFS and OS for ipilimumab and drugs like bevacizumab, trastuzumab, aflibercept makes no clinical sense. It is statistically correct, of course, but just makes no clinical sense. In his autobiography, the renowned theoretical physicist Richard Feynman recalls a problem from a mathematics book: ""John and his father go out to look at the stars. John sees two blue stars and a red star. His father sees a green star, a violet star, and two yellow stars. What is the total temperature of the stars seen by John and his father?”—and I would explode in horror... there’s no purpose whatsoever in adding the temperature of two stars." I now understand how you must have felt Feynman. Surely you weren’t joking, Mr. Feynman!
Reviewing eligibility criteria in oncology trials
It is well known that clinical trials don’t represent the real world scenario. Also, most of the real-world uses of drugs are just extensions or extrapolations of trial evidence because the trials use strict inclusion and exclusion criteria. Hence, we see that some trial benefits diminish in the real world or some unknown toxicities become apparent many years after the trial. Furthermore, strict eligibility criteria make many patients ineligible for the trial and they are thus deprived of access to good care. Hence this timely perspective from the FDA encourages broader patient participation in oncology trials such as including patients with brain metastasis or elderly or paediatric population. Once the drug is approved, it’s going to be used in many patients that don’t meet the trial criteria anyway. So, it’s important to have trial data among these patients. Industry might be afraid of losing the efficacy by doing so but appropriate suggestions, such as using relaxed criteria for safety assessment only, are provided.
How many bevacizumab trials do we need in ovarian cancer?
Too many, it seems. If you test the same drug in the same cancer many times, you might achieve statistical significance sometimes, as in this study. Among ovarian cancer patients who had relapsed on first line chemotherapy after 6 months or more (treatment free interval; TFI), this trial found that the addition of bevacizumab to the re-treatment chemotherapy improved survival but the p-value was not significant at 0.056. Later, they found that 45 patients were incorrectly stratified based on TFI. Upon correction of this data, the HR changed from 0.829 to 0.0823 and voila, the p was 0.0447! Magic! The trial is now positive.
I am not blaming the correction. Error and correction is in fact, a laudable exercise. There are two things I am concerned about: First, would they check for errors had the p been less than 0.05 in the first place? Would corrections have been announced if upon correction, the p or HR worsened? Shouldn’t all trials be rechecked for such inadvertent errors? Secondly, and most importantly, doesn’t this make a case for open data or data sharing? I would surely want to see this data validated by a third conflict free party. Wouldn't you?
Cancer Drugs Fund - a lesson from the UK to the world
The Cancer Drugs Fund (CDF) was established in the UK to improve access to cancer drugs not approved by NICE. This analysis found that the CDF spent more than £1.3 billion - the equivalent of 1 year’s total spend on all cancer drugs in the NHS - to cover 47 indications of which only 38% reported providing OS benefit. The authors conclude that “the CDF has not delivered meaningful value to patients or society. There is no empirical evidence to support a ‘drug only’ ring fenced cancer fund relative to concomitant investments in other cancer domains such as surgery and radiotherapy, or other non-cancer medicines.” This is an important message to other countries - cancer drugs represent the most expensive and least value-for-money investment in cancer care. Because financial resources aren’t infinite, every government should prioritize their investment in health resources. Spending billions on drugs that don’t improve survival or do so only by a median of a couple of months at the expense of investment on other cancer control strategies is not sensible.
Let me take a selfie
"Drugs that lack single-agent activity: are they worth pursuing in combination?" This is a question we asked in our recent commentary in Nature Reviews Clinical Oncology. It’s a short 2 page commentary which I invite my readers to take a look. We conclude: “Clinical trials are costly, arduous, and patients remain a scarce resource. The pursuit of agents lacking single-agent activity opens the door to numerous combinations, many potential clinical trials, and concerns regarding prioritization and streamlining of R&D efforts. Our analysis highlights that such agents often provide marginal benefits when combined with other drugs, have inconsistent effects on measures of activity, often have additive toxicity, and are used on the basis of inadequate end points. Thus, we believe the lack of single-agent activity is a reasonable basis to abandon anticancer drugs early in clinical development.”
Last, but not the least, thank you again.
Bishal Gyawali, MD is a postgraduate trainee in medical oncology at the Graduate School of Medicine, Nagoya University, Japan, where he is also a PhD candidate under the Japanese government scholarship. His areas of clinical and research interests include evidence-based oncology practice, global oncology, cancer policy, cancer management in resource-limited settings, financial toxicities of cancer treatment, clinical trial methods and supportive treatment of cancer. He has no conflicts of interest to declare. Dr Gyawali tweets at @oncology_bg. Read his previous blog posts here.