- Email: [email protected]
Disease kinetics for decision-making in advanced melanoma: a call for scenario-driven strategy trials
Personal View
Disease kinetics for decision-making in advanced melanoma: a call for scenario-driven strategy trials Jean Jacques Grob, Georgina V Long, Dirk Schadendorf, Keith Flaherty
In the past 5 years, the treatment of metastatic melanoma has changed from almost no effective treatment to the use of targeted and immune therapies with proven improvements in survival. The time has now come to define the optimal drug combinations, sequence of treatment, and drug regimens (intermittent vs continuous dosing) in the treatment of patients with metastatic melanoma. In view of the prevalence of advanced melanoma, finite resources, and the heterogeneity of disease characteristics, not all possibilities can be tested in therapeutic trials starting from an unselected population of patients with metastatic melanoma. In practice, clinicians rely on a few clinically derived signals, especially dynamic signals, to categorise patients into scenarios, from fast disease kinetics to slow disease kinetics, which drive clinicians’ therapeutic decision making. The realistic goals of therapy are different in each scenario. We recommend that these scenarios are incorporated into clinical trials as either patient inclusion criteria or stratification factors. This approach is not only feasible but is also the only way to generate evidence for more effective and individualised treatment strategies for patients with metastatic melanoma.
Introduction In the past 5 years, several drugs have shown clinical significance in the treatment of patients with metastatic melanoma—namely, two molecularly targeted drugs (a BRAF inhibitor1 and a MEK inhibitor2), two immune checkpoint inhibitors (CTLA-4 antibody3 and PD-1 antibody4,5), and two combinational therapies (a BRAF inhibitor plus a MEK inhibitor,6,7 and an anti-CTLA-4 antibody plus an anti-PD-1 antibody8). Yet, most patients will still succumb to their disease. Since the number of patients is finite, it is not possible to design trials that compare all existing treatment options, including optimal combinations, sequencing, and intermittent versus continuous dosing. Additionally, owing to the introduction of these drugs into widespread clinical use over a very short period of time, there has been little time to generate effective treatment strategies that are adapted to the many specific clinical situations with which clinicians are faced. Clinicians are expected to make strategic decisions regarding treatment options by the use of predictive markers and evidence from rigorous clinical trials. However, predictive markers in advanced melanoma are not robust and are neither sensitive nor specific.9–11 At present, evidence-based practice is impeded by clinical trials that oversimplify patient disease characteristics and the landscape of treatment options, only testing one drug against another, and only in stereotypical cases. In particular, phase 3 trials for melanoma are not sufficiently large, nor do their inclusion criteria cover the entire range of metastatic melanoma, to provide insights into the effectiveness of therapy in real-world subgroups. In this Personal View, we define clinically derived signals that could be used to categorise patients into three scenarios mainly on the basis of disease kinetics. We recommend the incorporation of these scenarios into the design of clinical trials to generate a base of evidence for more effective and individualised treatment strategies for patients with advanced melanoma. www.thelancet.com/oncology Vol 16 October 2015
Use of clinically derived signals to define different disease scenarios Clinicians who treat patients with advanced melanoma are consciously or unconsciously using clinically derived signals from the observation of their patients and of the behaviour of their metastatic disease. These signals have not yet been adequately defined to use as a basis for the inclusion, exclusion, or stratification of patients in therapeutic trials, but they are the major driver of clinicians’ therapeutic decisions when selecting a trial to be proposed to a patient or when selecting approved therapies. These signals can be separated into two categories: instantaneous signals, and dynamic signals that require time to characterise correctly. Instantaneous signals can be thought of as a photograph of the disease at a given moment in time. They include variables such as the presence of brain metastases, the presence of a tumour metastasis in a potentially life-threatening location, the radiological assessment of tumour burden, serum lactate dehydrogenase (LDH) levels,9 and the performance status of the patient.12,13 These instantaneous signals are often used as inclusion, exclusion, or stratification factors in the enrolment of patients for clinical trials. By contrast, dynamic signals are analogous to a motion picture and need a time period to record the clinical course of the disease. These signals give more information regarding the patient’s progress and possible outcome than instantaneous signals do, just as the next scene of a film can be predicted from the previous scene but not from a photograph. An assessment of the dynamic signals needs at least two consecutive measures of the same signal. They are regarded as important decisionmaking variables by most clinicians and are known as disease kinetics, disease progression, disease tempo, or disease aggressiveness, as mentioned in algorithms and guidelines.14–16 These dynamic signals can be assessed before treatment (ie, pretreatment disease kinetics) or during treatment (ie, kinetics under treatment).
Lancet Oncol 2015; 16: e522–26 Aix-Marseille University and APHM Hospital CHU Timone, Marseille, France (Prof J J Grob MD); Melanoma Institute Australia, University of Sydney, and Mater Hospital, Sydney, NSW, Australia (G V Long MBBS); University Hospital Essen and German Cancer Consortium, Essen, Germany (Prof D Schadendorf MD); and Massachusetts General Hospital Cancer Center, Boston, MA, USA (K Flaherty MD) Correspondence to: Prof Jean Jacques Grob, Aix-Marseille University and APHM Hospital CHU Timone, Marseille 13885, France [email protected]
e522
Personal View
Pretreatment disease kinetics Two successive measures of total tumour burden while the patient is not undergoing any treatment (ideally, two assessments within 4–12 weeks before treatment) provide an objective measure of disease kinetics17 and, in our view, is one of the most useful assessments to assist clinical decision making. However, the worsening of a patient’s performance status or an increase in concentration of serological markers (ie, LDH) over a short period of time might also be useful.10 Dynamic signals are sometimes inferred from instantaneous signals when no previous documentation is available. For example, a high tumour burden is usually regarded as an indicator of fast disease tempo and thus disease aggressiveness. However, a high tumour burden alone does not necessarily equate to fast disease tempo, or a low tumour load to slow disease tempo. A high tumour burden might occur in a patient with slowly progressing melanoma who has not been restaged for a prolonged period of time; by contrast, tumour load can be low in a patient with rapidly progressing melanoma when metastatic disease first manifests itself immediately after resection of high-risk regional disease or after a long period of indolence.
Kinetics under treatment The early changes of disease characteristics and kinetics after treatment has begun are important dynamic signals to assess disease progression, to which clinicians intuitively assign a high importance. If a patient’s disease is clearly progressing within the first few months of treatment with either BRAF inhibitor alone or with a combination of BRAF and MEK inhibitors, then he or she is unlikely to benefit from the same treatment at any time thereafter. However, the situation with immunotherapy is more complex than the situation with BRAF and MEK inhibitors. A small minority of patients whose disease has progressed within the first 12 weeks of anti-PD-1 therapy will benefit from the same treatment if it is continued.4,5,18 However, with anti-CTLA-4 therapy, absence of an obvious treatment response in the short term (within 12–16 weeks) can still result in durable disease control or treatment response, although the percentage of patients who benefit after initial progression is low.3
The different scenarios in advanced melanoma In practice, clinicians use clinically derived signals with a particular emphasis on dynamic signals. These signals are interpreted by the clinician as a scenario that drives a pragmatic decision regarding the ideal treatment strategy.
Initial scenarios Most newly diagnosed patients with distant metastatic melanoma can be simplified into three scenarios. In fast disease kinetics (ie, immediate danger), the first objective is to preserve life and relieve symptoms in the e523
short term. The ideal strategy is to use a fast-acting treatment with the highest possible response rate with acceptance of high toxicity. Long-term survival is only a secondary, although less realistic, objective. In intermediate disease kinetics (ie, non-immediate danger), the first objective is to prolong survival. The ideal strategy is to start with the treatment that will give the highest chance of a prolonged survival (3–5 years), and the strategy is altered depending on the result of treatment and toxicities. In slow disease kinetics, the first objective is to prolong survival with the best possible quality of life. The current strategy is to start with treatments such as surgery, radiosurgery, or any straightforward treatments that have low toxicity, low morbidity, and some potential to preserve long-term survival. This conservative approach might change soon and if new treatments can really improve 5-year survival, or even lead to a cure when they are given very early on, in these patients with low-aggressiveness disease. In this case, toxicity would no longer be a concern.
Adjustment of scenarios with treatment In everyday clinical practice, decisions regarding treatment and assessments of the benefit of a treatment often use signals other than classical outcomes used in clinical trials, such as objective response (by Response Evaluation Criteria in Solid Tumors) or progression-free survival. For clinicians, the assessment of treatment effectiveness can be simplified into two dominant scenarios. In the first scenario, where the patient benefits from treatment, the benefit can be interpreted as obvious or just relative to what was expected from the natural kinetics of the disease. The perception of benefit is a multifactorial integration of subjective factors, and the patient might be interpreted as deriving benefit even if some evidence of disease progression exists. In this scenario, the same treatment will be continued unless a much better treatment is available (ie, a treatment that can improve response, quality of life, toxicity, and survival). In the second scenario, the patient is not benefiting from treatment or is likely to lose benefit quickly. The definition of loss of benefit is a clear worsening of dynamic signals during treatment, such as a rapid increase in most pre-existing metastases, including the development of new metastases with or without lifethreatening implications. Such a worsening of dynamic signals might be accompanied by a loss in performance status or increased disease-related symptoms. Of note, the development of a single new metastasis or slow progression of a subset of pre-existing metastases alone is not an absolute indicator of loss of treatment benefit. However, if a loss of benefit is observed, the strategy is changed as soon as possible if another treatment strategy exists. www.thelancet.com/oncology Vol 16 October 2015
Personal View
Clinical decisions based on these dynamic signals in patients receiving treatment might directly affect the outcome of similarly designed trials, as seen in two very similar trials that used two different BRAF inhibitors, vemurafenib and dabrafenib. In the BREAK-3 trial,1 investigators had the option to continue treatment beyond disease progression (defined by Response Evaluation Criteria in Solid Tumors) if they perceived ongoing clinical benefit for the patient, whereas in the BRIM-3 trial2 the treatment had to be discontinued at disease progression, as defined by trial protocol. Overall survival was longer for patients in the BREAK-3 trial than for patients in the BRIM-3 trial. Although the two BRAF inhibitors might have different efficacy, the highly similar progression-free survival and disease responses for both agents support our interpretation that clinical decision making based on the interpretation of the dynamic signals account for the notable difference in overall survival reported in the two trials.
Need for scenario-driven clinical trials We cannot, and arguably should not, test all therapeutic combinations or sequences of the growing armamentarium of drugs in an undifferentiated, aggregate population of patients with advanced melanoma. However, with few definitive phase 3 trials establishing a clinical benefit of new targeted and immunotherapeutic agents in comparison with standard therapy, evidence-based practitioners tend to place undue significance on statistically underpowered subsets of patients from the trials, which they consider the closest to the scenarios of their patients in real life. We need to design dedicated therapeutic strategy trials for each of the three initial scenarios (fast, intermediate, and slow) to recapitulate the real decision-making process in daily clinical practice. Nowadays, we are convinced of the benefit of targeted therapy for appropriately selected patients based on molecular characteristics of the tumour. However, not a single trial in advanced melanoma has been based on the scenarios that clinicians use in daily decision making. Therefore, clinicians tend to distort or extrapolate the actual trial results to apply them to clinical practice. For example, the rapid onset of response in the use of BRAF and MEK inhibitors is interpreted as an argument in favour of these drugs in the fast disease kinetics scenario, although not a single trial has shown the superiority of BRAF and MEK inhibitors over other drugs in this setting. Data showed an apparently more rapid effect of anti-PD-1 monotherapy and of the combination of antiCTLA-4 plus anti-PD-1 antibodies than anti-CTLA-4 monotherapy in cross-trial comparison, which might lead to the same speculations about their potential effectiveness in fast disease kinetics scenarios. Furthermore, the benefit of BRAF and MEK inhibitors may even be better in a slow disease kinetics scenario than a fast disease kinetics scenario. For example, the www.thelancet.com/oncology Vol 16 October 2015
combination of a BRAF inhibitor plus a MEK inhibitor has a proportionately greater improvement in overall and progression-free survival than standard therapy in patients with normal serum LDH levels.19 Similarly, the delay of action of ipilimumab, an anti-PD-1 antibody, is interpreted in favour of the use of immunotherapy in the scenarios of intermediate and slow disease kinetics, although not a single trial has proved its superiority over BRAF inhibitors in these scenarios. The most important strategic questions in one scenario are usually irrelevant in another scenario. For example, testing an aggressive regimen combining anti-CTLA-4 plus anti-PD-1 antibodies versus the combination of BRAF plus MEK inhibitors might be more appropriate in a fast disease kinetics scenario than in a slow disease kinetics scenario. Only if these aggressive combinations show evidence of a much more prolonged long-term survival than with low-toxicity treatments would the exploration of these combination regimen in the slow kinetics setting be appropriate. Similarly, the testing of deintensified strategies, such as short-term or intermittent treatment with an anti-PD-1 antibody, or with a BRAF or MEK inhibitor, seems to be more warranted in a slow disease kinetics scenario, where the risk of a lost opportunity for another treatment later on in the patient’s treatment programme is lower than in a fast disease kinetics scenario. The design of the right trials in the right scenario, or at least the stratification of trials based on these scenarios, is important to answer the real questions clinicians face in the treatment of patients with advanced melanoma.
Design of scenario-driven clinical trials The present challenge is to agree on the definitions of each scenario so that they could be used in a reliable and reproducible manner in a clinical trial. An objective measure of initial disease kinetics is possible if the patient is monitored during a short period of time without treatment for 4–8 weeks with two successive CT scans before a treatment decision is made. This measure allows delineation of the three initial disease kineticsbased scenarios.17 Ethical considerations could be raised as to whether the loss of 1–2 months without treatment to ascertain the disease scenario is acceptable. In all cases, agreement by the patient and an explanation of the rationale for this approach are required. First, the long-term benefit of a scenario-driven strategy is likely to surpass the lost opportunity to initiate the therapy a few weeks earlier. Second, the usual loss of time in the daily management of patients (eg, validation of mutational status, completion of radiological staging studies, surgery, and pretreatment assessment) is often longer than a month, and this inevitable delay could be used to assess disease kinetics. Third, loss of time would be a problem mainly in fast-growing disease, but a very short period of 4 weeks is sufficient to ascertain the scenario of fast disease e524
Personal View
kinetics. From a practical point of view, with sufficient planning and resources, disease kinetics could be appropriately assessed without extra delay. Hopefully, easily measurable and reliable prognostic and predictive biomarkers will be discovered in the future and assist in the definition of the appropriate scenario for a specific patient. However, we believe that scenario-driven trials could be done before we have these validated biomarkers. The ultimate objective is not to perfectly delineate the three scenarios but to make trials more relevant to clinical practice. For example, irrespective of the criteria used to define patients with fast disease kinetics, choosing these criteria for inclusion in a trial will increase the proportion of patients with aggressive disease in the trial population, and the results will thus be more relevant to the fast scenario of metastatic disease in real practice, than any results of a post-hoc analysis of subgroups in a trial enrolling unselected patients. Thus, any trial with any definition of the three scenarios based on clinical factors such as radiology, PET scan, serum LDH levels, calculation of kinetics, and so on will be more useful for clinical practice than trials of metastatic patients without classification into the clinical scenarios. Even a trial with a purely subjective classification of patients by the clinician into the three scenarios as they are enrolled would be informative and useful, provided that participation is restricted to centres with melanoma expertise. However, a trial using this subjective classification into three scenarios as an enrolment criterion would only work if an adequate portfolio of at least one trial by scenario exists to control for the potential bias of clinicians misclassifying patients to give them access to a clinical trial versus standard therapy.
(eg, threatening brain metastases, serum LDH concentration >3 × upper limit of normal, or a rapid decline in performance status or organ function), or if a comparison with a previous CT scan done within the past 2 months gives a direct proof of fast disease kinetics, the patient will be classified in the scenario of fast disease kinetics upfront and immediately be offered a treatment strategy or a trial adapted to this scenario. In other situations, the wait of a few weeks (≥4 weeks) without treatment to assess disease kinetics is unlikely to induce substantial additional risks. Moreover, these few weeks are the typical time needed to organise treatments, arrange additional tests, and await tumour mutation testing in present clinical practice, and will not usually correspond to any extra delay for treatment. A second CT scan and test for serum LDH levels can be done at the time treatment begins, allowing an assessment of disease kinetics. An increase of more than 1 cm³ per day in the total tumour load between these two CT scans,17 the emergence of an immediate clinical threat within a month, or a major increase in serum LDH levels (double or more the previous value) will suggest a fast disease kinetics scenario. Such a rapid evolution demands a fastonset strategy and might save resources, which would probably be wasted if a slow-onset therapy such as ipilimumab had been started because of the lack of information about disease kinetics. In most other cases, a scenario of intermediate disease kinetics will be considered. However, if comparison with a previous disease assessment shows almost no changes over 4 months or more, the patient will be classified into a slow disease kinetics scenario. Strategies or trials adapted to scenarios of intermediate or slow disease kinetics will be proposed accordingly. The details of these classification criteria can be debated but can be accepted, in the large part, by any melanoma expert; these criteria can be used in clinical practice and for trial designs, without ethical problems, provided that the patient is properly informed. The time has come to test strategies, not only molecules, in the treatment of patients with advanced melanoma. Until we have relevant biological markers to classify different scenarios, the most relevant approach is to include kinetics-based scenarios as inclusion criteria or stratification factors in the design of clinical trials.
Proposal of a practical algorithm to characterise disease scenario
Contributors All authors designed and wrote this Personal View, and approved the final version before submission.
A CT scan should be done at the first consultation with an oncologist for a patient with stage IV or inoperable stage III melanoma. The following algorithm is practical, scientific, and ethical, since no extra delays will be introduced in the most dangerous situations, yet crucial information will be obtained to ensure the best interests of the patient in the long term. If the first assessment suggests an immediate emergency from the point of view of the clinician, irrespective of the reason
Declaration of interests JJG has received research grants to his institution from Roche, personal fees for lectures from Roche and GlaxoSmithKline, and personal fees for advisory boards from Roche, GlaxoSmithKline, Bristol-Myers Squibb, Merck, Novartis, Amgen, and Electra. GVL has acted as a consultant advisor for Amgen, Bristol-Myers Squibb, Merck, Novartis, Provectus Biopharmaceuticals, and Roche. DS has received honoraria for consultancy, speaker’s bureau, and advisory boards from Bristol-Myers Squibb, Merck, Novartis, Roche, Boehringer Ingelheim, Pfizer, Merck Serono, and GlaxoSmithKline. KF has acted as a consultant advisor for GSK, Novartis, Roche, and Merck.
Search strategy and selection criteria We searched MEDLINE for reports in English from Jan 1, 2004, to Dec 31, 2014. To ensure that no previous report has described what we proposed in our Personal View, we searched for “melanoma” with the following keywords: “strategy”, “kinetics”, “aggressiveness”, “aggressive”, “tempo”, “fast disease”, “slow disease”, “slow”, “fast”, “scenario”, “threat”, “guidelines”, and “LDH”.
e525
www.thelancet.com/oncology Vol 16 October 2015
Personal View
References 1 Hauschild A, Grob JJ, Demidov LV, et al. Dabrafenib in BRAFmutated metastatic melanoma: a multicentre, open-label, phase 3 randomised controlled trial. Lancet 2012; 380: 358–65. 2 Chapman PB, Hauschild A, Robert C, et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med 2011; 364: 2507–16. 3 Schadendorf D, Hodi FS, Robert C, et al. Pooled analysis of longterm survival data from phase II and phase III trials of ipilimumab in unresectable or metastatic melanoma. J Clin Oncol 2015; 33: 1889–94. 4 Topalian SL, Sznol M, McDermott DF, et al. Survival, durable tumor remission, and long-term safety in patients with advanced melanoma receiving nivolumab. J Clin Oncol 2014; 32: 1020–30. 5 Robert C, Ribas A, Wolchok JD, et al. Anti-programmed-deathreceptor-1 treatment with pembrolizumab in ipilimumab-refractory advanced melanoma: a randomized dose-comparison cohort of a phase 1 trial. Lancet 2014; 384: 1109–17. 6 Larkin J, Ascierto PA, Dréno B, et al. Combined vemurafenib and cobimetinib in BRAF-mutated melanoma. N Engl J Med 2014; 371: 1867–76. 7 Long GV, Stroyakovskiy D, Gogas H, et al. Combined BRAF and MEK inhibition versus BRAF inhibition alone in melanoma. N Engl J Med 2014; 371: 1877–88. 8 Postow MA, Chesney J, Pavlick AC, et al. Nivolumab and ipilimumab versus ipilimumab in untreated melanoma. N Engl J Med 2015; 372: 2006–17. 9 Balch CM, Soong SJ, Atkins MB, et al. An evidence-based staging system for cutaneous melanoma. CA Cancer J Clin 2004; 54: 131–49. 10 Weide B, Richter S, Büttner P, et al. Serum S100B, lactate dehydrogenase and brain metastasis are prognostic factors in patients with distant melanoma metastasis and systemic therapy. PLoS One 2013; 8: e81624.
www.thelancet.com/oncology Vol 16 October 2015
11
12
13
14
15
16
17
18
19
Ascierto PA, Kalos M, Schaer DA, Callahan MK, Wolchok JD. Biomarkers for immunostimulatory monoclonal antibodies in combination strategies for melanoma and other tumor types. Clin Cancer Res 2013; 19: 1009–20. Oken MM, Creech R, Tormey DC, et al. Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol 1982; 5: 649–55. Mor V, Laliberte L, Morris JN, Wiemann M. The Karnofsky Performance Status Scale. An examination of its reliability and validity in a research setting. Cancer 1984; 53: 2002–07. Leccia MT, Planchamp F, Sassolas B, et al. Management of patients with metastatic cutaneous melanoma: French national guidelines. French National Cancer Institute. Ann Dermatol Venereol 2014; 141: 111–21 (in French). Pflugfelder A, Kochs C, Blum A, et al. Malignant melanoma S3-guideline «diagnosis, therapy and follow-up of melanoma». J Dtsch Dermatol Ges 2013; 11 (Suppl 6): 1–116. Ascierto PA, Simeone E, Grimaldi AM, et al. Do BRAF inhibitors select for populations with different disease progression kinetics? J Transl Med 2013; 11: 61. Gaudy-Marqueste C, Archier E, Grob A, et al. Initial metastatic kinetics is the best prognostic indicator in stage IV metastatic melanoma. Eur J Cancer 2014; 50: 1120–24. Lyle MK, Lee JHJ, Menzies AM, et al. Lesion-specific patterns of response and progression with anti-PD-1 treatment in metastatic melanoma. 2014 ASCO Annual Meeting; Chicago, IL, USA; May 30–June 3, 2014. Abstract 9077. Robert C, Karaszewska B, Schachter J, et al. Improved overall survival in melanoma with combined dabrafenib and trametinib. N Engl J Med 2015; 372: 30–39.
e526
Disease kinetics for decision-making in advanced melanoma: a call for scenario-driven strategy trials Jean Jacques Grob, Georgina V Long, Dirk Schadendorf, Keith Flaherty
In the past 5 years, the treatment of metastatic melanoma has changed from almost no effective treatment to the use of targeted and immune therapies with proven improvements in survival. The time has now come to define the optimal drug combinations, sequence of treatment, and drug regimens (intermittent vs continuous dosing) in the treatment of patients with metastatic melanoma. In view of the prevalence of advanced melanoma, finite resources, and the heterogeneity of disease characteristics, not all possibilities can be tested in therapeutic trials starting from an unselected population of patients with metastatic melanoma. In practice, clinicians rely on a few clinically derived signals, especially dynamic signals, to categorise patients into scenarios, from fast disease kinetics to slow disease kinetics, which drive clinicians’ therapeutic decision making. The realistic goals of therapy are different in each scenario. We recommend that these scenarios are incorporated into clinical trials as either patient inclusion criteria or stratification factors. This approach is not only feasible but is also the only way to generate evidence for more effective and individualised treatment strategies for patients with metastatic melanoma.
Introduction In the past 5 years, several drugs have shown clinical significance in the treatment of patients with metastatic melanoma—namely, two molecularly targeted drugs (a BRAF inhibitor1 and a MEK inhibitor2), two immune checkpoint inhibitors (CTLA-4 antibody3 and PD-1 antibody4,5), and two combinational therapies (a BRAF inhibitor plus a MEK inhibitor,6,7 and an anti-CTLA-4 antibody plus an anti-PD-1 antibody8). Yet, most patients will still succumb to their disease. Since the number of patients is finite, it is not possible to design trials that compare all existing treatment options, including optimal combinations, sequencing, and intermittent versus continuous dosing. Additionally, owing to the introduction of these drugs into widespread clinical use over a very short period of time, there has been little time to generate effective treatment strategies that are adapted to the many specific clinical situations with which clinicians are faced. Clinicians are expected to make strategic decisions regarding treatment options by the use of predictive markers and evidence from rigorous clinical trials. However, predictive markers in advanced melanoma are not robust and are neither sensitive nor specific.9–11 At present, evidence-based practice is impeded by clinical trials that oversimplify patient disease characteristics and the landscape of treatment options, only testing one drug against another, and only in stereotypical cases. In particular, phase 3 trials for melanoma are not sufficiently large, nor do their inclusion criteria cover the entire range of metastatic melanoma, to provide insights into the effectiveness of therapy in real-world subgroups. In this Personal View, we define clinically derived signals that could be used to categorise patients into three scenarios mainly on the basis of disease kinetics. We recommend the incorporation of these scenarios into the design of clinical trials to generate a base of evidence for more effective and individualised treatment strategies for patients with advanced melanoma. www.thelancet.com/oncology Vol 16 October 2015
Use of clinically derived signals to define different disease scenarios Clinicians who treat patients with advanced melanoma are consciously or unconsciously using clinically derived signals from the observation of their patients and of the behaviour of their metastatic disease. These signals have not yet been adequately defined to use as a basis for the inclusion, exclusion, or stratification of patients in therapeutic trials, but they are the major driver of clinicians’ therapeutic decisions when selecting a trial to be proposed to a patient or when selecting approved therapies. These signals can be separated into two categories: instantaneous signals, and dynamic signals that require time to characterise correctly. Instantaneous signals can be thought of as a photograph of the disease at a given moment in time. They include variables such as the presence of brain metastases, the presence of a tumour metastasis in a potentially life-threatening location, the radiological assessment of tumour burden, serum lactate dehydrogenase (LDH) levels,9 and the performance status of the patient.12,13 These instantaneous signals are often used as inclusion, exclusion, or stratification factors in the enrolment of patients for clinical trials. By contrast, dynamic signals are analogous to a motion picture and need a time period to record the clinical course of the disease. These signals give more information regarding the patient’s progress and possible outcome than instantaneous signals do, just as the next scene of a film can be predicted from the previous scene but not from a photograph. An assessment of the dynamic signals needs at least two consecutive measures of the same signal. They are regarded as important decisionmaking variables by most clinicians and are known as disease kinetics, disease progression, disease tempo, or disease aggressiveness, as mentioned in algorithms and guidelines.14–16 These dynamic signals can be assessed before treatment (ie, pretreatment disease kinetics) or during treatment (ie, kinetics under treatment).
Lancet Oncol 2015; 16: e522–26 Aix-Marseille University and APHM Hospital CHU Timone, Marseille, France (Prof J J Grob MD); Melanoma Institute Australia, University of Sydney, and Mater Hospital, Sydney, NSW, Australia (G V Long MBBS); University Hospital Essen and German Cancer Consortium, Essen, Germany (Prof D Schadendorf MD); and Massachusetts General Hospital Cancer Center, Boston, MA, USA (K Flaherty MD) Correspondence to: Prof Jean Jacques Grob, Aix-Marseille University and APHM Hospital CHU Timone, Marseille 13885, France [email protected]
e522
Personal View
Pretreatment disease kinetics Two successive measures of total tumour burden while the patient is not undergoing any treatment (ideally, two assessments within 4–12 weeks before treatment) provide an objective measure of disease kinetics17 and, in our view, is one of the most useful assessments to assist clinical decision making. However, the worsening of a patient’s performance status or an increase in concentration of serological markers (ie, LDH) over a short period of time might also be useful.10 Dynamic signals are sometimes inferred from instantaneous signals when no previous documentation is available. For example, a high tumour burden is usually regarded as an indicator of fast disease tempo and thus disease aggressiveness. However, a high tumour burden alone does not necessarily equate to fast disease tempo, or a low tumour load to slow disease tempo. A high tumour burden might occur in a patient with slowly progressing melanoma who has not been restaged for a prolonged period of time; by contrast, tumour load can be low in a patient with rapidly progressing melanoma when metastatic disease first manifests itself immediately after resection of high-risk regional disease or after a long period of indolence.
Kinetics under treatment The early changes of disease characteristics and kinetics after treatment has begun are important dynamic signals to assess disease progression, to which clinicians intuitively assign a high importance. If a patient’s disease is clearly progressing within the first few months of treatment with either BRAF inhibitor alone or with a combination of BRAF and MEK inhibitors, then he or she is unlikely to benefit from the same treatment at any time thereafter. However, the situation with immunotherapy is more complex than the situation with BRAF and MEK inhibitors. A small minority of patients whose disease has progressed within the first 12 weeks of anti-PD-1 therapy will benefit from the same treatment if it is continued.4,5,18 However, with anti-CTLA-4 therapy, absence of an obvious treatment response in the short term (within 12–16 weeks) can still result in durable disease control or treatment response, although the percentage of patients who benefit after initial progression is low.3
The different scenarios in advanced melanoma In practice, clinicians use clinically derived signals with a particular emphasis on dynamic signals. These signals are interpreted by the clinician as a scenario that drives a pragmatic decision regarding the ideal treatment strategy.
Initial scenarios Most newly diagnosed patients with distant metastatic melanoma can be simplified into three scenarios. In fast disease kinetics (ie, immediate danger), the first objective is to preserve life and relieve symptoms in the e523
short term. The ideal strategy is to use a fast-acting treatment with the highest possible response rate with acceptance of high toxicity. Long-term survival is only a secondary, although less realistic, objective. In intermediate disease kinetics (ie, non-immediate danger), the first objective is to prolong survival. The ideal strategy is to start with the treatment that will give the highest chance of a prolonged survival (3–5 years), and the strategy is altered depending on the result of treatment and toxicities. In slow disease kinetics, the first objective is to prolong survival with the best possible quality of life. The current strategy is to start with treatments such as surgery, radiosurgery, or any straightforward treatments that have low toxicity, low morbidity, and some potential to preserve long-term survival. This conservative approach might change soon and if new treatments can really improve 5-year survival, or even lead to a cure when they are given very early on, in these patients with low-aggressiveness disease. In this case, toxicity would no longer be a concern.
Adjustment of scenarios with treatment In everyday clinical practice, decisions regarding treatment and assessments of the benefit of a treatment often use signals other than classical outcomes used in clinical trials, such as objective response (by Response Evaluation Criteria in Solid Tumors) or progression-free survival. For clinicians, the assessment of treatment effectiveness can be simplified into two dominant scenarios. In the first scenario, where the patient benefits from treatment, the benefit can be interpreted as obvious or just relative to what was expected from the natural kinetics of the disease. The perception of benefit is a multifactorial integration of subjective factors, and the patient might be interpreted as deriving benefit even if some evidence of disease progression exists. In this scenario, the same treatment will be continued unless a much better treatment is available (ie, a treatment that can improve response, quality of life, toxicity, and survival). In the second scenario, the patient is not benefiting from treatment or is likely to lose benefit quickly. The definition of loss of benefit is a clear worsening of dynamic signals during treatment, such as a rapid increase in most pre-existing metastases, including the development of new metastases with or without lifethreatening implications. Such a worsening of dynamic signals might be accompanied by a loss in performance status or increased disease-related symptoms. Of note, the development of a single new metastasis or slow progression of a subset of pre-existing metastases alone is not an absolute indicator of loss of treatment benefit. However, if a loss of benefit is observed, the strategy is changed as soon as possible if another treatment strategy exists. www.thelancet.com/oncology Vol 16 October 2015
Personal View
Clinical decisions based on these dynamic signals in patients receiving treatment might directly affect the outcome of similarly designed trials, as seen in two very similar trials that used two different BRAF inhibitors, vemurafenib and dabrafenib. In the BREAK-3 trial,1 investigators had the option to continue treatment beyond disease progression (defined by Response Evaluation Criteria in Solid Tumors) if they perceived ongoing clinical benefit for the patient, whereas in the BRIM-3 trial2 the treatment had to be discontinued at disease progression, as defined by trial protocol. Overall survival was longer for patients in the BREAK-3 trial than for patients in the BRIM-3 trial. Although the two BRAF inhibitors might have different efficacy, the highly similar progression-free survival and disease responses for both agents support our interpretation that clinical decision making based on the interpretation of the dynamic signals account for the notable difference in overall survival reported in the two trials.
Need for scenario-driven clinical trials We cannot, and arguably should not, test all therapeutic combinations or sequences of the growing armamentarium of drugs in an undifferentiated, aggregate population of patients with advanced melanoma. However, with few definitive phase 3 trials establishing a clinical benefit of new targeted and immunotherapeutic agents in comparison with standard therapy, evidence-based practitioners tend to place undue significance on statistically underpowered subsets of patients from the trials, which they consider the closest to the scenarios of their patients in real life. We need to design dedicated therapeutic strategy trials for each of the three initial scenarios (fast, intermediate, and slow) to recapitulate the real decision-making process in daily clinical practice. Nowadays, we are convinced of the benefit of targeted therapy for appropriately selected patients based on molecular characteristics of the tumour. However, not a single trial in advanced melanoma has been based on the scenarios that clinicians use in daily decision making. Therefore, clinicians tend to distort or extrapolate the actual trial results to apply them to clinical practice. For example, the rapid onset of response in the use of BRAF and MEK inhibitors is interpreted as an argument in favour of these drugs in the fast disease kinetics scenario, although not a single trial has shown the superiority of BRAF and MEK inhibitors over other drugs in this setting. Data showed an apparently more rapid effect of anti-PD-1 monotherapy and of the combination of antiCTLA-4 plus anti-PD-1 antibodies than anti-CTLA-4 monotherapy in cross-trial comparison, which might lead to the same speculations about their potential effectiveness in fast disease kinetics scenarios. Furthermore, the benefit of BRAF and MEK inhibitors may even be better in a slow disease kinetics scenario than a fast disease kinetics scenario. For example, the www.thelancet.com/oncology Vol 16 October 2015
combination of a BRAF inhibitor plus a MEK inhibitor has a proportionately greater improvement in overall and progression-free survival than standard therapy in patients with normal serum LDH levels.19 Similarly, the delay of action of ipilimumab, an anti-PD-1 antibody, is interpreted in favour of the use of immunotherapy in the scenarios of intermediate and slow disease kinetics, although not a single trial has proved its superiority over BRAF inhibitors in these scenarios. The most important strategic questions in one scenario are usually irrelevant in another scenario. For example, testing an aggressive regimen combining anti-CTLA-4 plus anti-PD-1 antibodies versus the combination of BRAF plus MEK inhibitors might be more appropriate in a fast disease kinetics scenario than in a slow disease kinetics scenario. Only if these aggressive combinations show evidence of a much more prolonged long-term survival than with low-toxicity treatments would the exploration of these combination regimen in the slow kinetics setting be appropriate. Similarly, the testing of deintensified strategies, such as short-term or intermittent treatment with an anti-PD-1 antibody, or with a BRAF or MEK inhibitor, seems to be more warranted in a slow disease kinetics scenario, where the risk of a lost opportunity for another treatment later on in the patient’s treatment programme is lower than in a fast disease kinetics scenario. The design of the right trials in the right scenario, or at least the stratification of trials based on these scenarios, is important to answer the real questions clinicians face in the treatment of patients with advanced melanoma.
Design of scenario-driven clinical trials The present challenge is to agree on the definitions of each scenario so that they could be used in a reliable and reproducible manner in a clinical trial. An objective measure of initial disease kinetics is possible if the patient is monitored during a short period of time without treatment for 4–8 weeks with two successive CT scans before a treatment decision is made. This measure allows delineation of the three initial disease kineticsbased scenarios.17 Ethical considerations could be raised as to whether the loss of 1–2 months without treatment to ascertain the disease scenario is acceptable. In all cases, agreement by the patient and an explanation of the rationale for this approach are required. First, the long-term benefit of a scenario-driven strategy is likely to surpass the lost opportunity to initiate the therapy a few weeks earlier. Second, the usual loss of time in the daily management of patients (eg, validation of mutational status, completion of radiological staging studies, surgery, and pretreatment assessment) is often longer than a month, and this inevitable delay could be used to assess disease kinetics. Third, loss of time would be a problem mainly in fast-growing disease, but a very short period of 4 weeks is sufficient to ascertain the scenario of fast disease e524
Personal View
kinetics. From a practical point of view, with sufficient planning and resources, disease kinetics could be appropriately assessed without extra delay. Hopefully, easily measurable and reliable prognostic and predictive biomarkers will be discovered in the future and assist in the definition of the appropriate scenario for a specific patient. However, we believe that scenario-driven trials could be done before we have these validated biomarkers. The ultimate objective is not to perfectly delineate the three scenarios but to make trials more relevant to clinical practice. For example, irrespective of the criteria used to define patients with fast disease kinetics, choosing these criteria for inclusion in a trial will increase the proportion of patients with aggressive disease in the trial population, and the results will thus be more relevant to the fast scenario of metastatic disease in real practice, than any results of a post-hoc analysis of subgroups in a trial enrolling unselected patients. Thus, any trial with any definition of the three scenarios based on clinical factors such as radiology, PET scan, serum LDH levels, calculation of kinetics, and so on will be more useful for clinical practice than trials of metastatic patients without classification into the clinical scenarios. Even a trial with a purely subjective classification of patients by the clinician into the three scenarios as they are enrolled would be informative and useful, provided that participation is restricted to centres with melanoma expertise. However, a trial using this subjective classification into three scenarios as an enrolment criterion would only work if an adequate portfolio of at least one trial by scenario exists to control for the potential bias of clinicians misclassifying patients to give them access to a clinical trial versus standard therapy.
(eg, threatening brain metastases, serum LDH concentration >3 × upper limit of normal, or a rapid decline in performance status or organ function), or if a comparison with a previous CT scan done within the past 2 months gives a direct proof of fast disease kinetics, the patient will be classified in the scenario of fast disease kinetics upfront and immediately be offered a treatment strategy or a trial adapted to this scenario. In other situations, the wait of a few weeks (≥4 weeks) without treatment to assess disease kinetics is unlikely to induce substantial additional risks. Moreover, these few weeks are the typical time needed to organise treatments, arrange additional tests, and await tumour mutation testing in present clinical practice, and will not usually correspond to any extra delay for treatment. A second CT scan and test for serum LDH levels can be done at the time treatment begins, allowing an assessment of disease kinetics. An increase of more than 1 cm³ per day in the total tumour load between these two CT scans,17 the emergence of an immediate clinical threat within a month, or a major increase in serum LDH levels (double or more the previous value) will suggest a fast disease kinetics scenario. Such a rapid evolution demands a fastonset strategy and might save resources, which would probably be wasted if a slow-onset therapy such as ipilimumab had been started because of the lack of information about disease kinetics. In most other cases, a scenario of intermediate disease kinetics will be considered. However, if comparison with a previous disease assessment shows almost no changes over 4 months or more, the patient will be classified into a slow disease kinetics scenario. Strategies or trials adapted to scenarios of intermediate or slow disease kinetics will be proposed accordingly. The details of these classification criteria can be debated but can be accepted, in the large part, by any melanoma expert; these criteria can be used in clinical practice and for trial designs, without ethical problems, provided that the patient is properly informed. The time has come to test strategies, not only molecules, in the treatment of patients with advanced melanoma. Until we have relevant biological markers to classify different scenarios, the most relevant approach is to include kinetics-based scenarios as inclusion criteria or stratification factors in the design of clinical trials.
Proposal of a practical algorithm to characterise disease scenario
Contributors All authors designed and wrote this Personal View, and approved the final version before submission.
A CT scan should be done at the first consultation with an oncologist for a patient with stage IV or inoperable stage III melanoma. The following algorithm is practical, scientific, and ethical, since no extra delays will be introduced in the most dangerous situations, yet crucial information will be obtained to ensure the best interests of the patient in the long term. If the first assessment suggests an immediate emergency from the point of view of the clinician, irrespective of the reason
Declaration of interests JJG has received research grants to his institution from Roche, personal fees for lectures from Roche and GlaxoSmithKline, and personal fees for advisory boards from Roche, GlaxoSmithKline, Bristol-Myers Squibb, Merck, Novartis, Amgen, and Electra. GVL has acted as a consultant advisor for Amgen, Bristol-Myers Squibb, Merck, Novartis, Provectus Biopharmaceuticals, and Roche. DS has received honoraria for consultancy, speaker’s bureau, and advisory boards from Bristol-Myers Squibb, Merck, Novartis, Roche, Boehringer Ingelheim, Pfizer, Merck Serono, and GlaxoSmithKline. KF has acted as a consultant advisor for GSK, Novartis, Roche, and Merck.
Search strategy and selection criteria We searched MEDLINE for reports in English from Jan 1, 2004, to Dec 31, 2014. To ensure that no previous report has described what we proposed in our Personal View, we searched for “melanoma” with the following keywords: “strategy”, “kinetics”, “aggressiveness”, “aggressive”, “tempo”, “fast disease”, “slow disease”, “slow”, “fast”, “scenario”, “threat”, “guidelines”, and “LDH”.
e525
www.thelancet.com/oncology Vol 16 October 2015
Personal View
References 1 Hauschild A, Grob JJ, Demidov LV, et al. Dabrafenib in BRAFmutated metastatic melanoma: a multicentre, open-label, phase 3 randomised controlled trial. Lancet 2012; 380: 358–65. 2 Chapman PB, Hauschild A, Robert C, et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med 2011; 364: 2507–16. 3 Schadendorf D, Hodi FS, Robert C, et al. Pooled analysis of longterm survival data from phase II and phase III trials of ipilimumab in unresectable or metastatic melanoma. J Clin Oncol 2015; 33: 1889–94. 4 Topalian SL, Sznol M, McDermott DF, et al. Survival, durable tumor remission, and long-term safety in patients with advanced melanoma receiving nivolumab. J Clin Oncol 2014; 32: 1020–30. 5 Robert C, Ribas A, Wolchok JD, et al. Anti-programmed-deathreceptor-1 treatment with pembrolizumab in ipilimumab-refractory advanced melanoma: a randomized dose-comparison cohort of a phase 1 trial. Lancet 2014; 384: 1109–17. 6 Larkin J, Ascierto PA, Dréno B, et al. Combined vemurafenib and cobimetinib in BRAF-mutated melanoma. N Engl J Med 2014; 371: 1867–76. 7 Long GV, Stroyakovskiy D, Gogas H, et al. Combined BRAF and MEK inhibition versus BRAF inhibition alone in melanoma. N Engl J Med 2014; 371: 1877–88. 8 Postow MA, Chesney J, Pavlick AC, et al. Nivolumab and ipilimumab versus ipilimumab in untreated melanoma. N Engl J Med 2015; 372: 2006–17. 9 Balch CM, Soong SJ, Atkins MB, et al. An evidence-based staging system for cutaneous melanoma. CA Cancer J Clin 2004; 54: 131–49. 10 Weide B, Richter S, Büttner P, et al. Serum S100B, lactate dehydrogenase and brain metastasis are prognostic factors in patients with distant melanoma metastasis and systemic therapy. PLoS One 2013; 8: e81624.
www.thelancet.com/oncology Vol 16 October 2015
11
12
13
14
15
16
17
18
19
Ascierto PA, Kalos M, Schaer DA, Callahan MK, Wolchok JD. Biomarkers for immunostimulatory monoclonal antibodies in combination strategies for melanoma and other tumor types. Clin Cancer Res 2013; 19: 1009–20. Oken MM, Creech R, Tormey DC, et al. Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol 1982; 5: 649–55. Mor V, Laliberte L, Morris JN, Wiemann M. The Karnofsky Performance Status Scale. An examination of its reliability and validity in a research setting. Cancer 1984; 53: 2002–07. Leccia MT, Planchamp F, Sassolas B, et al. Management of patients with metastatic cutaneous melanoma: French national guidelines. French National Cancer Institute. Ann Dermatol Venereol 2014; 141: 111–21 (in French). Pflugfelder A, Kochs C, Blum A, et al. Malignant melanoma S3-guideline «diagnosis, therapy and follow-up of melanoma». J Dtsch Dermatol Ges 2013; 11 (Suppl 6): 1–116. Ascierto PA, Simeone E, Grimaldi AM, et al. Do BRAF inhibitors select for populations with different disease progression kinetics? J Transl Med 2013; 11: 61. Gaudy-Marqueste C, Archier E, Grob A, et al. Initial metastatic kinetics is the best prognostic indicator in stage IV metastatic melanoma. Eur J Cancer 2014; 50: 1120–24. Lyle MK, Lee JHJ, Menzies AM, et al. Lesion-specific patterns of response and progression with anti-PD-1 treatment in metastatic melanoma. 2014 ASCO Annual Meeting; Chicago, IL, USA; May 30–June 3, 2014. Abstract 9077. Robert C, Karaszewska B, Schachter J, et al. Improved overall survival in melanoma with combined dabrafenib and trametinib. N Engl J Med 2015; 372: 30–39.
e526