Position of a panel of international lung cancer experts on the approval decision for use of durvalumab in stage III non-small-cell lung cancer (NSCLC) by the Committee for Medicinal Products for Human Use (CHMP)

The two recent publications of durvalumab 1.Antonia S.J. Villegas A. Daniel D. et al.Durvalumab after chemoradiotherapy in stage III non–small-cell lung cancer.N Engl J Med. 2017; 377: 1919-1929Crossref PubMed Scopus (2469) Google Scholar, 2.Antonia S.J. Villegas A. Daniel D. et al.Overall survival with durvalumab after chemoradiotherapy in stage III NSCLC.N Engl J Med. 2018; 379: 2342-2350Crossref PubMed Scopus (1607) Google Scholar showing that consolidation immunotherapy with durvalumab following chemoradiotherapy (chemoRT) could significantly improve the outcome of patients with inoperable stage III NSCLC have given rise to high hopes among patients, clinicians, and lung cancer experts. No significant advances have been made in the treatment of stage III NSCLC patients for at least two decades. Stage III NSCLC represents 20–25% of NSCLC and worldwide approximately 500 000 patients are diagnosed every year. This stage includes either locally advanced primary tumors with infiltration of vital mediastinal organs or involvement of locoregional mediastinal lymph nodes. Over the past decade, concomitant chemotherapy and radiotherapy has remained the standard of care for unresectable stage III NSCLC, irrespective of histological subtype or molecular characteristics. Currently, only 15–30% of patients are alive five years after chemoRT, and this figure remains largely unchanged despite multiple phase III randomized trials integrating surgery, higher radiation dose or induction/consolidation with chemotherapy, biologics or vaccine treatment 3.Bradley J. Paulus R. Komaki R. et al.Standard-dose versus high-dose conformal radiotherapy with concurrent and consolidation carboplatin plus paclitaxel with or without cetuximab for patients with stage IIIA or IIIB nsclc (RTOG 0617): a randomised, two-by-two factorial phase 3 study.Lancet Oncol. 2015; 16: 187-199Abstract Full Text Full Text PDF PubMed Scopus (1396) Google Scholar. In recent years, immune-checkpoints blockades (ICB) have revolutionized the care of metastatic non-small-cell lung cancer (NSCLC), becoming the standard second-line strategy as monotherapy, as well as front line in combination with chemotherapy, in all NSCLC patients irrespective of the tumor programmed death-ligand 1 (PD-L1) status. Radiation causes cell death and consequently local release of tumor antigens that can generate an immune response. This process is described as ‘in-situ vaccination’, theoretically leading to subsequent cancer cells death in all existing tumor sites. Both chemotherapy and radiation therapy can promote immunogenic cell death of tumor cells by activating dendritic cells and enhancing antigen presentation, resulting in induction of adaptive immune responses. PACIFIC (NCT03693300) is the first phase III trial to evaluate the role of ICB using durvalumab, a selective anti-PD-L1 antibody. The trial was conducted in patients with locally advanced unresectable stage III NSCLC who were not selected based on histology or tumor PD-L1 score. PACIFIC was published in the NEJM in 2017 demonstrating that one year of durvalumab consolidation after concurrent chemoradiotherapy (chemoRT) offers an unprecedented benefit in progression-free survival (PFS), almost halving the risk of progression PFS hazard ratio (HR) = 0.52; 95% confidence interval (CI) 0.42–0.65; P < 0.001 as compared with placebo in the intention-to-treat (ITT) population. This benefit was observed in both non-squamous and squamous histology as well as in both stages IIIA and IIIB NSCLC, irrespective of PD-L1 status 1.Antonia S.J. Villegas A. Daniel D. et al.Durvalumab after chemoradiotherapy in stage III non–small-cell lung cancer.N Engl J Med. 2017; 377: 1919-1929Crossref PubMed Scopus (2469) Google Scholar. In 2018, at the first overall survival (OS) interim analysis, a statistically significant and clinically meaningful relative increase in OS (HR = 0.68; 95% CI 0.53–0.87; P = 0.0025) along with updated PFS (PFS HR = 0.51; 95% CI 0.41–0.63), confirmed the lasting benefit of durvalumab in the ITT population 2.Antonia S.J. Villegas A. Daniel D. et al.Overall survival with durvalumab after chemoradiotherapy in stage III NSCLC.N Engl J Med. 2018; 379: 2342-2350Crossref PubMed Scopus (1607) Google Scholar. The PACIFIC study was powered to detect meaningful relative benefit for the co-primary end points PFS and OS in the ITT population. Indeed, a statistically significant benefit was shown for both end points at the interim analysis stage. Consolidation durvalumab is the first treatment within the past two decades to show a significant improvement in survival in this curative-intent setting. Following these impressive results, the United States Food and Drug Administration, as well as regulatory agencies in Canada, Japan, Australia, Switzerland, Malaysia, Singapore, India and the United Arab Emirates have approved durvalumab as consolidation therapy after chemoRT in unselected unresectable stage III NSCLC. Unexpectedly, the European Medicines Agency (EMA) made the decision to limit the use of durvalumab for patients with PD-L1 expressing tumors. This was based on an unplanned post hoc analysis, in a relatively small subset, that could not unequivocally establish an OS benefit in PD-L1 negative tumors, though the ITT population demonstrated a robust survival benefit 4.European Agency Medicines: Imfizi: EPAR – Product information. 2018; Google Scholar, 5.European Medicines Agency: Imfizi: EPAR – Public Assessment Report. 2018; EMA/CHMP/548232/2018Google Scholar. Importantly, when PACIFIC was conceived (original protocol in 2014), no immunotherapy biomarker had been formally evaluated, and subsequently the predictive ability of PD-L1 for ICB was far from being suggested for routine tumor molecular characterization. For these reasons, tissue collection at diagnosis, although highly recommended, was not mandatory for entry into the study, acknowledging that such a tissue collection is anatomically difficult to obtain in many patients with stage III NSCLC. An exploratory PD-L1 analysis with an empirical cut-off of 25% was planned, while no stratification was carried out based on PD-L1 expression. Nevertheless, the EMA asked for an unplanned post hoc analysis based on PD-L1 expression levels on tumor cells at initial biopsy. In this unplanned biomarker-based analysis, while the durvalumab beneficial effect on PFS was consistent across PD-L1 expression, a benefit in OS was not established in the specific PD-L1-negative subgroup. EMA used the uncertainties identified in this subgroup as a basis for its recommendation excluding patients with PD-L1 < 1% or unknown from access to durvalumab 4.European Agency Medicines: Imfizi: EPAR – Product information. 2018; Google Scholar, 5.European Medicines Agency: Imfizi: EPAR – Public Assessment Report. 2018; EMA/CHMP/548232/2018Google Scholar. Of note, the ‘PD-L1 unknown’ subgroup demonstrated a strong survival benefit, consistent with that of the ITT population. The European Society for Medical Oncology (ESMO) recently developed the Magnitude of Clinical Benefit Scale (ESMO-MCBS) with the aim ‘to derive a relative ranking of the magnitude of clinically meaningful benefit that can be expected from a new anticancer treatment’ 6.Cherny N. Sullivan R. Dafni U. et al.A standardised, generic, validated approach to stratify the magnitude of clinical benefit that can be anticipated from anti-cancer therapies: the European Society for Medical Oncology Magnitude of Clinical Benefit Scale (ESMO-MCBS).Ann Oncol. 2015; 26: 1547-1573Abstract Full Text Full Text PDF PubMed Scopus (532) Google Scholar, 7.Cherny N. Dafni U. Bogaerts J. et al.ESMO-Magnitude of Clinical Benefit Scale Version 1.1.Ann Oncol. 2017; 28: 2340-2366Abstract Full Text Full Text PDF PubMed Scopus (325) Google Scholar. Drugs or treatment interventions that obtain the highest scores on the scale (A for curative; 4 or 5 for non-curative) should be accepted as standard of care in the ESMO guidelines, with the hope that they will become rapidly accessible across the European Union. Based on the ESMO-MCBS methodology, and taking into account the potential for cure in the locally advanced lung cancer scenario, maintenance durvalumab gets the maximal score of A (while based on the non-curative setting also gets the high score of 4, scoring PFS). Of note, the guidelines for MCBS, respecting statistical principles, evaluate the primary ITT analysis and restrict additional evaluation to only a limited number of pre-planned subgroup analyses, NOT allowing exploratory, post hoc comparisons. It is surprising that the EMA, a regulatory agency known to follow well-established statistical methodological principles, has concluded that patients with PD-L1 < 1% should not receive treatment with durvalumab. We disagree with the decision for the following reasons: •The conclusion is NOT based on the primary analysis population (ITT), as is customary and appropriate (CONSORT 8.Altman D.G. Schulz K.F. Moher D. et al.The revised consort statement for reporting randomized trials: explanation and elaboration.Ann Intern Med. 2001; 134: 663-694Crossref PubMed Scopus (3054) Google Scholar).•The conclusion is instead based on an unplanned, post hoc exploratory analysis requested by the regulator with an additional cutoff, different from the one pre-specified in the protocol (1% instead of planned 25%), contrary to basic statistical principles 9.Lagakos S. The challenge of subgroup analyses—reporting without distorting.N Engl J Med. 2006; 354: 1667-1669Crossref PubMed Scopus (407) Google Scholar.•PD-L1 expression can be dynamic and heterogeneous. This is influenced by tumor biology as well as technical and analytical parameters including the quality and quantity of biopsy samples, the type of antibody utilized, the various scoring systems and different cut points currently used to assess PD-L1 status in different tumor types 10.Takada K. Toyokawa G. Shoji F. et al.The significance of the PD-L1 expression in non-small-cell lung cancer: trenchant double swords as predictive and prognostic markers.Clin Lung Cancer. 2018; 19: 120-129Abstract Full Text Full Text PDF PubMed Scopus (51) Google Scholar, 11.Herrera F. Bourhis J. Coukos G. Radiotherapy combination opportunities leveraging immunity for the next oncology practice.CA Cancer J Clin. 2017; 67: 65-85Crossref PubMed Scopus (266) Google Scholar, 12.Patel K. Martinez A. Stahl J. et al.Increase in PD-L1 expression after pre-operative radiotherapy for soft tissue sarcoma.OncoImmunology. 2018; 7: e1442168Crossref PubMed Scopus (49) Google Scholar, 13.Song X. Shao Y. Jiang T. et al.Radiotherapy upregulates programmed death ligand-1 through the pathways downstream of epidermal growth factor receptor in glioma.EBioMedicine. 2018; 28: 105-113Abstract Full Text Full Text PDF PubMed Scopus (23) Google Scholar, 14.Sheng J. Fang W. Yu J. et al.Expression of programmed death ligand-1 on tumor cells varies pre and post chemotherapy in non-small cell lung cancer.Sci Rep. 2016; 6: 20090Crossref PubMed Scopus (137) Google Scholar, 15.Takamori S. Toyokawa G. Takada K. et al.Combination therapy of radiotherapy and anti-PD-1/PD-L1 treatment in nonesmall-cell lung cancer: a mini-review.Clin Lung Cancer. 2018; 19: 12-16Abstract Full Text Full Text PDF PubMed Scopus (56) Google Scholar, 16.Hecht M. Herold M. Erlenbach-Wünsch K. et al.PD-L1 is upregulated by radiochemotherapy in rectal adenocarcinoma patients and associated with a favourable prognosis.Eur J Cancer. 2016; 65: 52-60Abstract Full Text Full Text PDF PubMed Scopus (93) Google Scholar, 17.Fujimoto D. Uehara K. Sato Y. et al.Alteration of PD-L1 expression and its prognostic impact after concurrent chemoradiation therapy in non-small cell lung cancer patients.Sci Rep. 2017; 7: 11373Crossref PubMed Scopus (52) Google Scholar.•PD-L1 expression was analyzed in pre-treatment samples; the hypothesis for the combination strategy is that chemoRT induces changes in the tumor micro-environment, thus rendering it more suited for response to ICB. Therefore, in the absence of PD-L1 evaluation in the post-chemoRT samples, the predictive capacities of baseline PD-L1 expression cannot be accurately assessed.•Tissue collection at diagnosis was not mandatory, and thus tissue samples were available for only 451 (63%) of patients with only 148 patients having PD-L1 expression below 1%. There is no guarantee that the tumor samples are missing completely at random. Bias cannot be excluded.•Balance of characteristics within the PD-L1 subgroups is not guaranteed since no stratification by PD-L1 was carried out. Possible confounding effects from other prognostic factors cannot be fully taken into account.•Over-performance of the control arm in the PD-L1-negative subgroup as compared with the ITT placebo arm is observed. This is possibly a consequence of imbalances in prognostic/predictive factors for NSCLC outcome, the unplanned retrospective analysis, and the small subgroup sample size.•A high probability of false negative results exists. The design of the trial targeted an OS benefit of 27% (HR=0.73) in the ITT population (power 85%, two-sided α = 0.025). The power for detecting an OS benefit in the PD-L1-negative subgroup, with only 60 observed events is inadequate. In fact, with this number of events, using the study design α of 0.025, a power of 80% is reached only for a targeted benefit of at least 57% (HR = 0.43), which is an unreasonably high target. Not surprisingly based on the above caveats, the OS result is inconclusive in the PD-L1 negative subgroup, with the additional observation that it does not indicate harm in these patients. Of note, although the overall OS benefit was found significant, the regulator recognized that OS data was not fully mature even for the ITT population and recommended to continue collecting OS data post-approval 4.European Agency Medicines: Imfizi: EPAR – Product information. 2018; Google Scholar, 5.European Medicines Agency: Imfizi: EPAR – Public Assessment Report. 2018; EMA/CHMP/548232/2018Google Scholar. According to the surprising EMA recommendation driven by an exploratory analysis, European patients belonging to a group of high unmet need will not have access to durvalumab. This is a treatment which was proven in the ITT population through a well-designed randomized clinical trial, to provide significant benefit in both PFS and OS. Furthermore, the drug was well tolerated with a manageable safety profile. No other treatment after chemoRT is approved for these patients. Despite the MCBS high score in both evaluation settings, durvalumab is not approved in patients with PD-L1 unknown or negative status. This restriction is only applicable to European patients, contrary to the citizens of other developed countries, according to the UN definition. The EMA decision is a major cause for concern among the thoracic oncology community. It is hard for us to think of an alternate situation where approval is granted by a regulatory agency for an unplanned subset of patients that derived benefit, although results for the ITT population are negative. While we understand that the EMA decision is taken to protect patients from unnecessary treatments and possible toxicities, durvalumab treatment might be life-saving in a substantial patient subset. We are fully aware of the need for consideration of the economic benefit to patients with the addition of new treatment strategies. We are also fully supportive of the need to identify predictive biomarkers to individualize novel therapies to patients. From this standpoint, members of this panel have been at the forefront of developing biomarkers for immunotherapy. At a time when the therapeutic options are improving for lung cancer, a lethal disease for the vast majority of the patients, we worry that the present EMA decision could result in a lost therapeutic opportunity in the curative setting for a sizeable subset of patients. In the routine clinical setting, clinical decisions for the treatments of stage III NSCLC are often made on small biopsies, smears and cytological specimens, which may limit the ability to conduct PD-L1 testing for all patients. The EMA decision based on a post hoc subgroup analysis is against the integrity of the clinical trial rules on which the principles of evidence-based medicine are based. It could create an unprecedented situation unique to Europe with regards to management of stage III NSCLC compared with other countries, and set worrying precedents with respect to analysis of future clinical trial data. Subset analyses are typically used to generate new hypotheses. As it was done in the molecular characterization for the definition of breast cancer adjuvant treatment, we should wait for a prospective validation in the PD-L1-negative subgroup before denying these patients access to durvalumab. We support additional studies in patients with PD-L1 negative disease to verify the role of durvalumab. A conditional approval for the PD-L1 negative population, perhaps accompanied by a warning on the medication information leaflet about the lack of robust efficacy data in this specific context, would have been a more balanced decision. This would enable the agency to oblige the manufacturer to continue collecting OS data in the PD-L1-negative subset. Finally, we are also concerned that this decision will have a negative social impact creating unacceptable disparities across countries, with differential access to treatments; some patients will pay for the treatment, some patients with private insurance shall get access and some might decide to get treatment abroad, where the drug is available. None declared.